Proceedings Volume 8009

Third International Conference on Digital Image Processing (ICDIP 2011)

Ting Zhang
cover
Proceedings Volume 8009

Third International Conference on Digital Image Processing (ICDIP 2011)

Ting Zhang
View the digital version of this volume at SPIE Digital Libarary.

Volume Details

Date Published: 28 June 2011
Contents: 2 Sessions, 128 Papers, 0 Presentations
Conference: 3rd International Conference on Digital Image Processing 2011
Volume Number: 8009

Table of Contents

icon_mobile_dropdown

Table of Contents

All links to SPIE Proceedings will open in the SPIE Digital Library. external link icon
View Session icon_mobile_dropdown
  • Front Matter: Volume 8009
  • Third International Conference on Digital Image Processing
Front Matter: Volume 8009
icon_mobile_dropdown
Front Matter: Volume 8009
This PDF file contains the front matter associated with SPIE Proceedings Volume 8009, including the Title Page, Copyright information, Table of Contents, Introduction, and Conference Committee listing.
Third International Conference on Digital Image Processing
icon_mobile_dropdown
A new watermarking approach based on combination of reversible watermarking and CDMA in spatial and DWT domain
S. Bekkouche, A. Chouarfia
Image watermarking can be defined as a technique that allows insertion of imperceptible and indelible digital data into an image. In addition to its initial application which is the copyright, watermarking can be used in other fields, particularly in the medical field in order to contribute to secure images shared on the network for telemedicine applications. In this report we study some watermarking methods and the comparison result of their combination, the first one is based on the CDMA (Code Division Multiple Access) in DWT and spatial domain and its aim is to verify the image authenticity whereas the second one is the reversible watermarking (the least significant bits LSB and cryptography tools) and the reversible contrast mapping RCM its objective is to check the integrity of the image and to keep the Confidentiality of the patient data. A new scheme of watermarking is the combination of the reversible watermarking method based on LSB and cryptography tools and the method of CDMA in spatial and DWT domain to verify the three security properties Integrity, Authenticity and confidentiality of medical data and patient information .In the end ,we made a comparison between these methods within the parameters of quality of medical images. Initially, an in-depth study on the characteristics of medical images would contribute to improve these methods to mitigate their limits and to optimize the results. Tests were done on IRM kind of medical images and the quality measurements have been done on the watermarked image to verify that this technique does not lead to a wrong diagnostic. The robustness of the watermarked images against attacks has been verified on the parameters of PSNR, SNR, MSE and MAE which the experimental result demonstrated that the proposed algorithm is good and robust in DWT than in spatial domain.
Efficient algorithm for salt and pepper noise removal
Gouse Mohiuddin Kosgiker
In this paper an efficient nonlinear decision based filter is proposed to remove salt and pepper impulse noise. Propose filter is a two stage filter that incorporates a powerful impulse noise detection method, called the modified boundary discriminative noise detection (MBDND) [1] to determine whether the current pixel is corrupted or not. In the second stage a Euclidean distance algorithm [2] is used to restore the corrupted pixels. Extensive experimental results demonstrate that proposed filters performs significantly better than many existing, well accepted and recently proposed median and decision based filters for both gray scale and color images corrupted up to 70% of salt and pepper noise.
Multifocus image fusion scheme based on nonsubsampled contourlet transform
Xinxing Zhou, Dianhong Wang, Zhijuan Duan, et al.
This paper proposes a novel multifocus image fusion scheme based on nonsubsampled contourlet transform (NSCT). The selection principles for different subband coefficients in NSCT domain are discussed in detail. In order to be consistent with the characteristics of the human visual system and improve the robustness of the fusion algorithm to the noise, the NSCT-DCT energy is first developed. Based on it, the clarity measure and bandpass energy contrast are defined and employed to motivate the pulse coupled neural networks (PCNN) for the fusion of lowpass and bandpass subbands, respectively. The performance of the proposed fusion scheme is assessed by experiments and the results demonstrate that the algorithm proposed in the paper compares favorably to wavelet-based, contourlet-based and NSCTbased fusion algorithms in terms of visual appearances and objective criterion.
A high performance hardware implementation image encryption with AES algorithm
Ali Farmani, Mohamad Jafari, Seyed Sohrab Miremadi
This paper describes implementation of a high-speed encryption algorithm with high throughput for encrypting the image. Therefore, we select a highly secured symmetric key encryption algorithm AES(Advanced Encryption Standard), in order to increase the speed and throughput using pipeline technique in four stages, control unit based on logic gates, optimal design of multiplier blocks in mixcolumn phase and simultaneous production keys and rounds. Such procedure makes AES suitable for fast image encryption. Implementation of a 128-bit AES on FPGA of Altra company has been done and the results are as follow: throughput, 6 Gbps in 471MHz. The time of encrypting in tested image with 32*32 size is 1.15ms.
Automatic honeycombing detection based on watershed transform
Yanjie Zhu, Jianguo Zhang, Wenjie Dong
Honeycombing is a common diffuse lung symptom in High-Resolution computed Tomography (HRCT), indicating the fibrosis of the lung. The purpose of this study was to develop an automatic scheme to detect honeycombing pattern accurately. The scans of 30 patients with diffuse lung disease were enrolled in the study. The lung region identified by threshold and morphological operations was pre-segmented by watershed transform to be divided into proper regions of interest (ROIs). Then texture features selected by recursive feature elimination algorithm were calculated within each ROI. Support vector machine (SVM) is used to generate rules with the training examples provided by experienced radiologists and knowledge-guided strategy was applied to reduce false positive regions. The proposed system achieved an accuracy of 92.8%, a sensitivity of 87.6% and a specification of 93.9%. The strategy is sufficiently accurate for objective and quantitative analysis of honeycombing in lung CT images.
Proposed optimization for AdaBoost-based face detection
Jiu Xu, Satoshi Goto
In this paper, a novel approach is proposed for face detection in still image based on the AdaBoost algorithm. First, face candidates are detected by AdaBoost Algorithm. Since a lot of influence might exist, such as size of the image, illumination and noise, some non-faces windows might also be detected as face candidates, or some faces might be missed. In order to solve these problems and get better performances, we take use of skin color information in the YCbCr color space together with the edge information of the color image. In this way, we are able to remove some non-faces that have been wrongly detected as faces and add some possible missed faces as well. Experimental results show that the hit rate could be improved and false alarm could also be reduced by this method.
Neural networks type MLP in the process of identification chosen varieties of maize
P. Boniecki, K. Nowakowski, R. Tomczak
During the adaptation process of the weights vector that occurs in the iterative presentation of the teaching vector, the the MLP type artificial neural network (MultiLayer Perceptron) attempts to learn the structure of the data. Such a network can learn to recognise aggregates of input data occurring in the input data set regardless of the assumed criteria of similarity and the quantity of the data explored. The MLP type neural network can be also used to detect regularities occurring in the obtained graphic empirical data. The neuronal image analysis is then a new field of digital processing of signals. It is possible to use it to identify chosen objects given in the form of bitmap. If at the network input, a new unknown case appears which the network is unable to recognise, it means that it is different from all the classes known previously. The MLP type artificial neural network taught in this way can serve as a detector signalling the appearance of a widely understood novelty. Such a network can also look for similarities between the known data and the noisy data. In this way, it is able to identify fragments of images presented in photographs of e.g. maze's grain. The purpose of the research was to use the MLP neural networks in the process of identification of chosen varieties of maize with the use of image analysis method. The neuronal classification shapes of grains was performed with the use of the Johan Gielis super formula.
Cloud base height survey based on stereo image
Guosheng Li, Zongjian Lin, Shuqing Ma, et al.
Cloud observation is an important factor for weather application and change of cloud base height plays a vital role in development of future weather system. Aim at requirement of cloud observation, a method on survey of cloud base height (CBH) is proposed based on stereo image. The main contents include image match and calculation of CBH which makes use of forward intersection of photogrammetry. It overcomes discontinuity, strong subjectivity and qualitative analysis of traditional eye observation. By application of National Day's weather safeguard, it tests that method of stereo imaging surveying is a way of direct measurement with better precision and is possible in technology also.
The face hallucinating two-step framework using hallucinated high-resolution residual
H. M. M. Naleer, Yao Lu, Yaozu An
In video surveillance, the attention human faces are frequently of small size. Image hallucination is an imperative factor disturbing the face classification by human and computer. In this paper, we propose a two-step face hallucination framework by means of training data sets which have a small quantity of low and high resolution images. In the first step, the global face is constructed based on optimal weights of training images. In the second step, a local residual compensation method bases on position patch via residual training face image data sets. Moreover, the hallucinated highresolution residual image which is obtained by the identical process can be subsequent for the local face. Finally, the hallucinated high-resolution residual image is appended with the input low-resolution face image which is interpolated to the high-resolution image dimension by an upsampling factor. Experiments fully demonstrate that our framework is very flexible and accomplishs good performance via small training data sets.
Astronomical image restoration through atmosphere turbulence by lucky imaging
Shixue Zhang, Yuanhao Wu, Jinyu Zhao, et al.
In this paper, we develop a lucky imaging system to restore astronomical images through atmosphere turbulence. Our system takes very short exposures, on the order of the atmospheric coherence time. The rapidly changing turbulence leads to a very variable point spread function (PSF), and the variability of the PSF leads to some frames having better quality than the rest. Only the best frames are selected, aligned and co-added to give a final image with much improved angular resolution. Our system mainly consists of five parts: preprocessing, frame selection, image registration, image reconstruction, and image enhancement. Our lucky imaging system has been successfully applied to restore the astronomical images taken by a 1.23m telescope. We have got clear images of moon surface and Jupiter, and our system can be demonstrated to greatly improve the imaging resolution through atmospheric turbulence.
Identification process of corn and barley kernel damages using neural image analysis
K. Nowakowski, P. Boniecki, R. J. Tomczak, et al.
The subject of the study was to develop a neural model for the identification of mechanical damage in grain caryopses based on digital photographs. The authors has selected a set of universal features that distinguish between damaged and healthy caryopses. The study has produced an artificial neural network of a multilayer perceptron type whose identification capacity approximates that of a human.
Fingerprint recognition using image processing
Surekha Dholay, Akassh A. Mishra
Finger Print Recognition is concerned with the difficult task of matching the images of finger print of a person with the finger print present in the database efficiently. Finger print Recognition is used in forensic science which helps in finding the criminals and also used in authentication of a particular person. Since, Finger print is the only thing which is unique among the people and changes from person to person. The present paper describes finger print recognition methods using various edge detection techniques and also how to detect correct finger print using a camera images. The present paper describes the method that does not require a special device but a simple camera can be used for its processes. Hence, the describe technique can also be using in a simple camera mobile phone. The various factors affecting the process will be poor illumination, noise disturbance, viewpoint-dependence, Climate factors, and Imaging conditions. The described factor has to be considered so we have to perform various image enhancement techniques so as to increase the quality and remove noise disturbance of image. The present paper describe the technique of using contour tracking on the finger print image then using edge detection on the contour and after that matching the edges inside the contour.
Image recognition technology based on projection entropy
Dejun Tang, Weishi Zhang, Chen Wang, et al.
This paper is a study on the shortages of algorithm based on grayscale with low matching speed and feature-based algorithm with low matching rates, meanwhile, the fast algorithm based on grayscale and feature-based algorithm for high matching rate are discussed, and try to combine them, expect to achieve fast matching with high matching rate. Introducing the concept of image entropy and projection characteristics to the image matching completely, to define the image local projection entropy, and an image matching method based on local projection entropy is proposed. Image matching method based on local projection entropy has good ability against geometric distortion, and further reduces the computation; the results indicate that this is a simple and effective method of image matching.
Automated rice leaf disease detection using color image analysis
Reinald Adrian D. L. Pugoy, Vladimir Y. Mariano
In rice-related institutions such as the International Rice Research Institute, assessing the health condition of a rice plant through its leaves, which is usually done as a manual eyeball exercise, is important to come up with good nutrient and disease management strategies. In this paper, an automated system that can detect diseases present in a rice leaf using color image analysis is presented. In the system, the outlier region is first obtained from a rice leaf image to be tested using histogram intersection between the test and healthy rice leaf images. Upon obtaining the outlier, it is then subjected to a threshold-based K-means clustering algorithm to group related regions into clusters. Then, these clusters are subjected to further analysis to finally determine the suspected diseases of the rice leaf.
Background subtraction based on nonparametric Bayesian estimation
Yan He, Donghui Wang, Miaoliang Zhu
Background subtraction, the task of separating foreground pixels from background pixels in a video, is an important step in video processing. Comparing with the parametric background modeling methods, nonparametric methods use a model selection criterion to choose the right number of components for each pixel online. We model the background subtraction problem with the Dirichlet process mixture, which constantly adapts both the parameters and the number of components of the mixture to the scene.
Development of a patient-specific model for calculation of pulmonary function
Hualiang Zhong, Mingyue Ding, Benjamin Movsas, et al.
The purpose of this paper is to develop a patient-specific finite element model (FEM) to calculate the pulmonary function of lung cancer patients for evaluation of radiation treatment. The lung model was created with an in-house developed FEM software with region-specific parameters derived from a four-dimensional CT (4DCT) image. The model was used first to calculate changes in air volume and elastic stress in the lung, and then to calculate regional compliance defined as the change in air volume corrected by its associated stress. The results have shown that the resultant compliance images can reveal the regional elastic property of lung tissue, and could be useful for radiation treatment planning and assessment.
Image decomposition based on modified bidimensional empirical mode decomposition
Faten Ben Arfia, Mohamed Ben Messaoud, Mohamed Abid
In this paper we develop an adaptive algorithm for decomposition of greyscales images. This method is highly adaptive decomposition image called Bidimentional Empirical Mode Decomposition (BEMD). It is based on the characterization of the image through its decomposition in Intrinsic Mode Function (IMF) where it can be decomposed into basis functions called IMF and a residue. This method offered a good result in visual quality, unfortunately this method consume an important execution time. To overcome this problem we proposed a new approach using Block based BEMD method where the input image is subdivided into blocks. Then the BEMD is applied on each of the four blocks separately. This method offered a good solution to reduce the execution time.
The level set method for medical image segmentation with a new regularization
Wei Zheng, Chunye Sun, Fang Wang, et al.
The local area information level set method is applied to segment the medical images with intensity inhomogeneous and a new kind of Gaussian kernel regularization method is used to simplify operation, this regularization can not only ensure the smoothness of the level set function, but also eliminate the requirement of re-initialization. This method which we call Improved Local Binary Fitting method (ILBF) has a shorter time consuming compared with the LBF method, so it can be widely used in medical image segmentation with its high efficiency and accuracy.
3D object recognition using kernel construction of phase wrapped images
Hong Zhang, Hongjun Su
Kernel methods are effective machine learning techniques for many image based pattern recognition problems. Incorporating 3D information is useful in such applications. The optical profilometries and interforometric techniques provide 3D information in an implicit form. Typically phase unwrapping process, which is often hindered by the presence of noises, spots of low intensity modulation, and instability of the solutions, is applied to retrieve the proper depth information. In certain applications such as pattern recognition problems, the goal is to classify the 3D objects in the image, rather than to simply display or reconstruct them. In this paper we present a technique for constructing kernels on the measured data directly without explicit phase unwrapping. Such a kernel will naturally incorporate the 3D depth information and can be used to improve the systems involving 3D object analysis and classification.
A feature extraction method brought to visual inspection system for micro-platform
Xiaodong Song, Donghong Si, Haili Xu, et al.
In this paper, a feature extraction method has been studied for visual inspection system of micro-platform. In this paper, an improved algorithm for Harris to improve the accuracy and detection speed based on introducing the principle Harris algorithm for the importance of the feature extraction for image matching, camera calibration, three-dimensional reconstruction and so on. Finally, a very good corner has been got by the algorithm research in MATLAB and experimental. And we proved the feasibility of this method.
Research on the method of kinematical error measurement of the micro-platform
Hang Li, Haili Xu, Xiaodong Song, et al.
In order to study the micro-table three-dimensional motion measurement errors, this article proposed that realizes the goal of three dimensional position and orientation examination based on the monocular vision. The micro work table abstracts to rectangular platform with a regular shape, the rectangle platform selects vertices and edges as the feature elements. Based on the measurement principle of monocular vision, with the coordinate transformation, it obtains method for solving three-dimensional position and orientation based on the feature element.
Analysis of watershed landscape pattern change based on TM images
Chongwei Li, Sa Wang, Yongli Zhao
Based on analysis of remote sensing images, statistic data of Yuqiao watershed in Tianjin from 1999 to 2009, and with the technology of Remote Sensing (abbr. RS) and Geographic Information System (abbr. GIS), At the same time, according to Maximum Likelihood Classifier (MLC) , Support Vector Machines (SVM) classifier ,statistical analysis and correlation analysis, we can quantitatively analyze landscape pattern change of Yuqiao watershed by calculating its Landscape diversity index, landscape shape index, Patch density, Edge density and contagion, etc. As a result, significant land-use changes have taken place in the Yuqiao watershed over during the ten years due to urbanization. Farmland has a notable increase. Meanwhile; there is a remarkable decrease in grass and shrub land. The farmland landscape's LSI from 145.72 increased 207.89 from 1999 to 2009, according to the complex and anomalous of farmland, which indicates that the way people cultivate the farmland, not only causes farmland quantity increase, but also makes the shape becomes complex. In the end, some advice will be given that human beings should adjust land-use structure in lake districts. The study of the integration of TM images and GIS technique is an effective approach to analyze the landscape changes in the watershed.
Fire detection combining HSI model and improved codebook model
Xiang-Yang Yu, Ming-Yu Gao, Zhi-Wei He
The paper presents a fire detection method using HSI model and an improved Codebook model. Based on the statistical feature of fire, we can extract fire-like pixels with HSI model. Owing to the fact that fire is extending and flickering, we propose an improved Codebook model for background subtraction. As a result, moving objects (for example, fire, moving cars) can be detected. At last, utilizing a pixel fusion method, we add the dynamic and statistical fire-like pixels with different weighted values. Experimental results show that our fire detection algorithm is effective and robust.
Destriping IRS image based on image processing methods
Qiang Chen, Yong Hu, Cailan Gong, et al.
China Environment and disaster monitoring and forecasting satellite was successfully launched in 2008, the Infrared Multispectral Camera (IRS) on the satellite compared to other similar instruments has outstanding features and widely application. However, because it scans by multiple sensors and differential sensitivities of sensors to incoming radiation, the images display periodic strips, it impacts follow-up quantitative research. In this paper, image processing methods, including two-point algorithm based on image, histogram matching algorithm and moment matching algorithm are put into practice in destriping IRS thermal infrared channel images, also called non-uniformity correction. By evaluating the average line of standard deviation, the average standard deviation and generalized noise law, strips caused by nonuniformity has been significantly removed. In addition, characteristic of image strips in the frequency domain is also obvious.
Successive difference detection based adaptive iterative median filter for image restoration
Divya Balachandran, Saranya Rangaraj, Nithya Krishnamoorthy
The paper proposes a universal noise removing filter capable of removing a mix of salt and pepper, random and Gaussian noise. In addition, the filter can also remove degradations like scratches, streaks, grids etc. that may corrupt images in real time. The algorithm exploits the fact that noise is a violation of spatial coherence of image intensities. In the detection phase, the corrupted and uncorrupted pixels are identified by computing the successive difference of the sorted pixels in the detection window. The correction phase uses the adaptive median filter that is iterated through the image until all the noise pixels are restored. The iteration with the window dimension not exceeding 5 x 5 ensures better preservation of image details. For an image that is corrupted with Salt and Pepper noise of density 60%, Random noise of density 20% and Gaussian noise of Standard Deviation 20, the image restored by this filter has a PSNR as high as 22 dB. The best feature of the proposed Successive Difference Detection Based Adaptive Iterative Median Filter (SDD-AIMF) is the graceful degradation in performance as the noise density increases, which is not the case with popular algorithms. The quantitative and qualitative results clearly prove that the proposed algorithm has better image restoration capabilities than many other popular techniques in literature.
A case study of learning writing in service-learning through CMC
Yunxiang Li, LiLi Ren, Xiaomian Liu, et al.
Computer-mediated communication ( CMC ) through online has developed successfully with its adoption by educators. Service Learning is a teaching and learning strategy that integrates community service with academic instruction and reflection to enrich students further understanding of course content, meet genuine community needs, develop career-related skills, and become responsible citizens. This study focuses on an EFL writing learning via CMC in an online virtual environment of service places by taking the case study of service Learning to probe into the scoring algorithm in CMC. The study combines the quantitative and qualitative research to probe into the practical feasibility and effectiveness of EFL writing learning via CMC in service learning in China.
Digital halftoning using a modified pulse-coupled neural network
Huawei Duan, Guangxue Chen
We report the application of modified pulse-coupled neural network (PCNN) models as an image processing tool to improve the quality of digital halftoning. Four factors including weight matrice, internal activity computation, type of error diffusion and linking coefficient were researched and optimized in terms of the PSNR metric and visual inspection on halftoning simulations. Experimental results show that the optimized PCNN model is able to yield satisfying halftoning outputs, which has better quality than that obtained by using the traditional order dither algorithm. Moreover, because of the utilization of random function in the modified PCNN model, simulated images generated from that PCNN model eliminate the periodic visual defect that the order dither innately has and therefore can potentially get rid of moiré pattern if used for printing color image. This research, on the one hand, provides a new way to do digital halftoning, on the other hand, expands the application field of the PCNN method.
An efficient approach to detected edge of image
Mingqin Liu, Xiaoguang Zhang
An efficient approach to detected edge of image by extract local fractal feature of image is proposed in this paper. The method that how to extract local fractal feature of image is proposed and then given the results of experiments. By comparison with other methods,it has been shown that our method is both efficient and accurate. Practical results on artificial and natural textured images are presented. It has been shown that our method is efficient. Finally, the method was applied to detected edge of image, we can obtain a clearly and accurately edge.
Transformer fault prediction based on particle swarm optimization and SVM
Yan Zhang, Bide Zhang
Forecasting of dissolved gases content in power transformer oil is very significant to detect incipient failures of transformer early and ensure normal operation of entire power system. Forecasting of dissolved gases content in power transformer oil is a complicated problem due to its nonlinearity and the small quantity of training data. Support vector machine (SVM) has been successfully employed to solve regression problem of nonlinearity and small sample. However, it is different to choice the best parameters of the SVM ,In this study, support vector machine is proposed to forecast dissolved gases content in power transformer oil, among which Particle Swarm Optimization (PSO) are used to determine free parameters of support vector machine. The experimental data from the electric power company in Sichuan are used to illustrate the performance of proposed PSO-SVM model. The experimental results indicate that the proposed PSO-SVM model can achieve greater forecasting accuracy than grey model (GM) under the circumstances of small sample. Consequently, the PSO-SVM model is a proper alternative for forecasting dissolved gases content in power transformer oil.
Infrared and visible image fusion using NSCT and GGD
Xiuqiong Zhang, Cuiyin Liu, Tao Men, et al.
In order to fuse the visible and infrared images captured in low visibility conditions, a method based on nonsubsampled contourlet transform (NSCT) and generalized Gaussian distribution (GGD) is proposed in this paper. The statistical character of the directional coefficients decomposed by NSCT meet the GGD. So, the coefficients are estimated using absolute moment estimation in local neighbor in directional coefficients. The estimated scale parameter is used to measure the saliency and compute the weight. The fused coefficients are obtained by the weighted average and are reconstructed the final fused image. Compared to the DWT and SIDWT, the proposed method has b superior fusion performance.
Change detection of lung cancer using image registration and thin-plate spline warping
Dawood M. S. Almasslawi, Ehsanollah Kabir
Lung cancer has the lowest survival rate comparing to other types of cancer and determination of the patient's cancer stage is the most vital issue regarding the cancer treatment process. In most cases accurate estimation of the cancer stage is not easy to achieve. The changes in the size of the primary tumor can be detected using image registration techniques. The registration method proposed in this paper uses Normalized Mutual Information metric and Thin-Plate Spline transformation function for the accurate determination of the correspondence between series of the lung cancer Computed Tomography images. The Normalized Mutual Information is used as a metric for the rigid registration of the images to better estimate the global motion of the tissues and the Thin Plate Spline is used to deform the image in a locally supported manner. The Control Points needed for the transformation are extracted semiautomatically. This new approach in change detection of the lung cancer is implemented using the Insight Toolkit. The results from implementing this method on the CT images of 8 patients provided a satisfactory quality for change detection of the lung cancer.
An efficient and robust 3D mesh compression based on 3D watermarking and wavelet transform
Ezzeddine Zagrouba, Saoussen Ben Jabra, Yosra Didi
The compression and watermarking of 3D meshes are very important in many areas of activity including digital cinematography, virtual reality as well as CAD design. However, most studies on 3D watermarking and 3D compression are done independently. To verify a good trade-off between protection and a fast transfer of 3D meshes, this paper proposes a new approach which combines 3D mesh compression with mesh watermarking. This combination is based on a wavelet transformation. In fact, the used compression method is decomposed to two stages: geometric encoding and topologic encoding. The proposed approach consists to insert a signature between these two stages. First, the wavelet transformation is applied to the original mesh to obtain two components: wavelets coefficients and a coarse mesh. Then, the geometric encoding is done on these two components. The obtained coarse mesh will be marked using a robust mesh watermarking scheme. This insertion into coarse mesh allows obtaining high robustness to several attacks. Finally, the topologic encoding is applied to the marked coarse mesh to obtain the compressed mesh. The combination of compression and watermarking permits to detect the presence of signature after a compression of the marked mesh. In plus, it allows transferring protected 3D meshes with the minimum size. The experiments and evaluations show that the proposed approach presents efficient results in terms of compression gain, invisibility and robustness of the signature against of many attacks.
Negative obstacle detection from image sequences
Tingbo Hu, Yiming Nie, Tao Wu, et al.
Negative obstacle detection has been a challenging topic. In the previous researches, the distance that negative obstacles can be detected is so near that vehicles have to travel at a very low speed. In this paper, a negative obstacle detection algorithm from image sequences is proposed. When negative obstacles are far from the vehicle, color appearance models are used as the cues of detecting negative obstacles, while negative obstacles get closer, geometrical cues are extracted from stereo vision. Furthermore, different cues are combined in a Bayesian framework to detect obstacles in image sequences. Massive experiments show that the proposed negative obstacle detection algorithm is quite effective. The alarming distance for 0.8 m width negative obstacle is 18m, and the confirming distance is 10 m. This supplies more space for vehicles to slow down and avoid obstacles. Then, the security of the UGV running in the field can be improved remarkably.
Binary image authentication based on watermarking algorithm
Behrang Masoodifar, S. Mojtaba Hashemi, Omid Zarei
A digital image watermark embedding and extracting algorithm is presented based on the Finite Ridgelet Transform (FRT) which can efficiently represent image with linear singularities. In general RT also has directional sensitivity so that among the transformed coefficients the most significant one represents the most energetic direction of straight edges in an image. In this paper effect of RT is compared with wavelet transform in watermarking application. Different noises with different PSNR are added into the watermarked image in the experiments and the results are of robustness and transparency.
Iterated denoising and fusion to improve the image quality of wavelet-based coding
Beibei Song
An iterated denoising and fusion method is presented to improve the image quality of wavelet-based coding. Firstly, iterated image denoising is used to reduce ringing and staircase noise along curving edges and improve edge regularity. Then, we adopt wavelet fusion method to enhance image edges, protect non-edge regions and decrease blurring artifacts during the process of denoising. Experimental results have shown that the proposed scheme is capable of improving both the subjective and the objective performance of wavelet decoders, such as JPEG2000 and SPIHT.
Remote sensing image fusion based on M-band dual-tree wavelet transform
Beibei Song
The paper proposes a new fusion method of visible and near infrared remote sensing images based on the multidirectional and multiscale M-band dual-tree wavelet transform(DTWT). The M-band DTWT decomposition is performed on each source image, then the high-frequency subband fusion coefficients are calculated by synthesizing modulus maxima selection and weighted average based on local coefficient matching, and average rule is adopted for low-frequency coefficients fusion. Experiment results show the proposed fusion method outperforms other typical ones, no matter in visual observation and objective evaluation criterion.
Analysis and improvement of SNR using time slicing
Srikrishna Karanam, Amarjot Singh, Devinder Kumar, et al.
Noise is a very important factor which in most cases, plays an antagonistic role in the vast field of image processing. Thus noise needs to be studied in great depth in order to improve the quality of images. The quantity of signal in an image, corrupted by noise is generally described by the term Signal-to-Noise ratio. Capturing multiple photos at different focus settings is a powerful approach for improving SNR. The paper analyses a frame work for optimally balancing the tradeoff's between defocus and sensor noise by experimenting on synthetic as well as real video sequences. The method is first applied to synthetic image where the improvement in SNR is studied by the ability of Hough transform to extract the number of lines with respect to the variation in SNR. The paper further experiments on real time video sequences while the improvement in SNR is analyzed using different edge operators like Sobel, Canny, Prewitt, Roberts and Laplacian. The result obtained is further analyzed using different edge operators. The main aim is to detect the edges at different values of SNR which will be a prominent measure of the signal strength as well as clarity of an image. The paper also explains in depth the modeling of noise leading to better understanding of SNR. The results obtain from both synthetic image and real time video sequences elaborate the increase in SNR with the increment in the total number of time slices in a fixed budget leading to clear pictures. This technique can be very effectively applied to capture high quality images from long distances.
Geometrical regularization of nonrigid registration using local anisotropic structure and joint saliency map
Jiawei Zhou, Binjie Qin
Nonrigid image registration is a crucial task to study local structural/volumetric change in many applications. The presence and resection of brain tumor in pre- and intra-operative brain images will greatly distort local anatomical structure and introduce non-corresponding outlier features. This can cause serious conflicts in achieving a smoothly varying deformation field in nonrigid registration. In this paper, a novel regularizing scheme, which is based on local anisotropic structure and Joint Saliency Map weighted regularization, is introduced in registration to aim at handling local complex deformation and outliers. The sparse displacement is regularized to adapt its smoothness as well as orientation according to the local anisotropic structure. Moreover, the Joint Saliency Map guides the assignment of data certainty so that the reliable corresponding structural voxels are emphasized in regularization. The results show that our method is sufficiently accurate and effective to both local large deformation and outliers while maintaining an overall smooth deformation field.
Zoning types of mountain forest restoration in the upper Min River supported by image auto-recognition system
Q. Wang, F. C. Li
Image auto- recognition system is a processing platform for extracting the grid data of characteristic element from original map and converting grid data to vector data. Selecting five maps about precipitation, evaporation, vegetation, soil erosion, and human activities, this paper applied images auto- recognition system and GIS technique to define and distinguish the boundary between natural restoration, human reconstruction and arid zone in the Upper Reaches of Min River. The results show that: images auto-recognition system efficiently extracted the grid data of characteristic element from original landscape patterns, and converted the grid data to vector data accurately.
Face recognition based on color model
Ying Zhang, Daoli Huang, Hong Guo, et al.
Over the past ten years face segmentation has developed rapidly and various algorithms have been proposed. In this paper we will demonstrate a face detection system based on skin color and the spaces RGB, normalized RGB, HSV and YCbCr are concentrated here. Through combing them the more accurate face region will be detected.
3D object recognition using line structure
Zhaojin Lu, Yonghong Li, Jian Wang, et al.
This paper presents an intelligent approach to recognize 3D objects using line structure correspondences. The proposed approach simultaneous recognizes an object and estimates the pose of the object. In order to achieve this goal, three challenges should be solved. First of all, line structures that human usually used to describe an object is used to represent the object. A set of such feature representation that shares the same properties with corresponding model line structures are first generated from images. Secondly, the structure correspondences are evaluated and ranked by additional features in the image. Only the most meaningful correspondences are selected. Each correspondence contributes a pose hypothesis with a transformation matrix. Finally, the approximate model pose hypotheses are estimated and refined based on the selected correspondences.
Study on methods of filtering the random noise from GPS CV observed data
Bian Li, Lili Qu, Yuping Gao
Measuring noise becomes one of the major factors to influence comparison result precision after correcting each time delay error in GPS Common View (GPS CV) comparison data. So it is the key to filter and smooth measuring noises of GPS CV observed data, which affect the precision for GPS CV comparison results directly. In order to eliminate GPS CV measuring noises, the de-noise mean on wavelet transform was studied. The method of threshold wavelet de-noise is to use wavelet transform technique to decompose and reconstruct signal with noise, and confine the thresholds of wavelet coefficients so as to eliminate the noise from signal. This paper presents an application of the wavelet threshold de-noise to filtering random noise from GPS observed data. The results show that wavelet filtering noise methods are effective and satisfied in filtering GPS CV measuring noises.
Parallel of low-level computer vision algorithms on a multi-DSP system
Huaida Liu, Pingui Jia, Lijian Li, et al.
Parallel hardware becomes a commonly used approach to satisfy the intensive computation demands of computer vision systems. A multiprocessor architecture based on hypercube interconnecting digital signal processors (DSPs) is described to exploit the temporal and spatial parallelism. This paper presents a parallel implementation of low level vision algorithms designed on multi-DSP system. The convolution operation has been parallelized by using redundant boundary partitioning. Performance of the parallel convolution operation is investigated by varying the image size, mask size and the number of processors. Experimental results show that the speedup is close to the ideal value. However, it can be found that the loading imbalance of processor can significantly affect the computation time and speedup of the multi- DSP system.
Moving object detection and tracking based on improved Surendra background updating algorithm
Cheng-liang Wang, Liang-liang Jia, Juan-juan Chen
Background subtraction method is usually used to detect the moving objects. But the establishment of the background model is vulnerable due to some external environmental influences, e.g. the noise, the brightness variations and so on. This paper presents a method based on three frame difference arithmetic of Surendra background updating algorithm to build the background model. By difference calculation of the current frame and background model, and deal with the binary images by applying the difference images, then make the morphology operation for removing the noises to get the smooth images, mark the different connected regions among binary images. According to different labels in connected regions, this paper defines different center of mass of moving objects, and analyzes the image features of moving object area, so as to achieve the detection and track of moving objects.
Wavelet transform for skew angle detection in printed Persian documents
Samira Nasrollahi Dizajyekan, Afshin Ebrahimi
When a document is fed to a scanner for digitization, it suffers from some degrees of skew. This skew has a detrimental effect on character recognition. In this paper we present two novel skew estimation methods for scanned Persian documents that estimates skew angles based on the wavelet decompositions and projection profile analysis. In the first algorithm, one row of the skewed document images is selected and is decomposed by the wavelet transform and in the second algorithm the whole document image is decomposed. Then, in both algorithms the matrix containing the absolute values of the horizontal sub-band coefficients, is then rotated through a range of angles. A projection profile is computed at each angle, and the angle that maximizes the maximum projection profile is regarded as the skew angle. The first algorithm is used for limited angle but second algorithm is used for wide angles.
Feature extraction from printed Persian sub-words using Haar wavelet transform
Samira Nasrollahi Dizajyekan, Afshin Ebrahimi
This article presents a novel set of shape descriptors which are especially well-suited for the recognition of printed Persian sub-words based on their holistic shapes. The descriptor set is derived from the wavelet transform of a sub-word's image. The proposed algorithm is used to extract features from 87804 sub-words of 4 fonts and 3 sizes. To evaluate the feature extraction results, this algorithm was used to obtain recognition rate for a set of sub-words in a printed Persian text document. Features of an unknown sub-word are extracted and compared with all sub-words features in the dictionary and the desired sub-word is identified. In this stage to increase the recognition rate, dot features of the unknown sub-word are used as the second feature and compared with dot codes of 10 last sub-words in before stage and the sub-word with maximum similarity is extracted as correct recognized sub-word.
Research on the distribution characteristics of secondary geological disasters induced by 5.12 earthquake in Wenchuan county based on RS and GIS
Hui Yu, Yong Luo, Zhi-Jun Zheng, et al.
Field observation, remote sensing combined with GIS technique were adopted to analysis the spatial distribution characteristics of secondary geological disasters induced by Wenchuan earthquake. The study results show that: (1) The earthquake-damaged slash mostly distributed in the two sides of main fault zones, earthquake-affected areas and disaster density decreased with the increase of distance from the main fault. (2) The earthquake-damaged slash mostly distributed in seismic intensity of IX and above. The region with greater seismic intensity had larger disaster density. Collapse and landslide distributed in high seismic intensity zone. (3) Hard rocks were easily caused secondary geological disasters by seismic waves. (4) The earthquake damaged slash mostly distributed within 0.5 km away from the river and decreased with the increase of distance from river. Collapse and landslide often appear at valley and the steep slope which more than 40 degree. The biggest hazard density appeared at altitude between 2000 m and 2500 m.
An improved structure propagation based image inpainting
Bianjing Bai, Zhenjiang Miao, Zhen Tang
Image inpainting technique has been widely used for completing missing parts caused by the removal of foreground or background elements from an image in a plausible way. A novel and efficient image inpainting method is proposed in this paper. First, structure propagation is used to synthesize regions in the unknown region along these salient structures specified by user. After structure completion, a finer algorithm is used to fill in the remaining unknown regions. It can prevent from erroneous matching blocks and reduce the breaking of salient structures which human eyes are sensitive to. Experiment results show the high efficiency and quality of the algorithm.
Several methods of smoothing motion capture data
Jingjing Qi, Zhenjiang Miao, Shujun Zhang
Human motion capture and editing technologies are widely used in computer animation production. We can acquire original motion data by human motion capture system, and then process it by motion editing system. However, noise embed in original motion data maybe introduced by extracting the target, three-dimensional reconstruction process, optimizing algorithm and devices itself in human motion capture system. The motion data must be modified before used to make videos, otherwise the animation figures will be jerky and their behavior is unnatural. Therefore, motion smoothing is essential. In this paper, we compare and summarize three methods of smoothing original motion capture data.
Age classification using facial feature extraction on female and male images
Fatemeh Mirzaei, Önsen Toygar
This paper presents age classification on facial images using subpattern-based Local Binary Patterns (LBP) method. Classification of age intervals are conducted separately on female and male facial images since the aging process for female and male is different for human beings in real life. The age classification performance of the holistic approaches is compared with the performance of subpattern-based LBP approach in order to demonstrate the performance differences between these two types of approaches. To be consistent with the research of others, our work has been tested on two publicly available databases namely FGNET and MORPH. The experiments are performed on these aging databases to demonstrate the age classification performance on female and male facial images of human beings using subpatternbased LBP method with several parameter settings. The results are then compared with the results of age classification of the holistic PCA and holistic subspace LDA methods.
A straight line detection method based on chain codes
Fan Li, Guozhen Li
Straight line detection is still a challenging task in image processing. Though many methods have been put forward before, most of them have some limitations. In this paper, we propose a novel method for straight line detection based on chain codes. Different from the former methods based on chain codes, this method doesn't rely too much on the Freeman's criteria. A more robust and reasonable criterion is adopted for detection, and an algorithm is presented to make this criterion works. Then we seek for a solution to the problem of corner points locating error that exists in all the chain codes tracking methods. We also suggest a criterion for correcting corner points, and present a strategy which can save computation time. The results show that our method has better performance than the traditional methods.
Using the Canny edge detector and mathematical morphology operators to detect vegetation patches
Qingsheng Liu, Jinfa Dong, Gaohuan Liu, et al.
Numbers and areas and locations of vegetation community patch are the important parameters for vegetation function and structure researches. In this paper, these parameters of vegetation patches are detected using the Canny edge detector and mathematical morphology operators. Firstly, part of a SPOT 5 fusion-ready color image is transformed into the gray image, then, stretched according to the histogram of the gray image in order to enhance the interesting vegetation patches. Secondly, using the Wiener filter to remove the noise and Canny edge detector to find the edges of the targets in the gray image. Finally, vegetation patches are detected based on the mathematical morphology criterion of circle and ellipse object and the centers of the patches are located. The experiments show that integration the Canny edge detector with the algorithms for extracting circle and ellipse object based on mathematical morphology are simple and effective for detecting vegetation patches.
Some results on preserving K-member simply separable relations in partial K-valued logic
Xin Yang, Fen Xu, Hongliang Zhu
According to the completeness theory in partial K-valued logic, the number of preserving K-member simply separable relation in partial K-valued logic is concluded; further more, we use a new method to present similar relation furthermore, we prove they are not similar.
The design of automatic close valve of isolating pollution in situ
Zi-ming Kou, Juan Wu
According to accident in Dalian New Port, The automatic close valve of isolating pollution in-situ is put forward. Form model on bond grand of valve to study the valve's open and close performance and dynamic response. Reveal the influence rule of pressure variation and load variation for the valve. Provide a theoretical basis for proper design and controlling of the valve.
A multilevel regression analysis based nonlocal means denoising algorithm
Jin Xu, Pengcheng Zheng, Rui Lv
This paper focuses on image denoising under the powerful framework-non local means. First, the introduction and development of NL-means is discussed. Second, a powerful scheme based on linear regression analysis for the classification of image meaningful parts is proposed. Third, an improved version of NL-means is carried out, which uses a novel patch similarity rule based on quadratic regression analysis. This multilevel regression analysis based algorithm can better describe and smooth the noisy image and finally, experimental results validate the algorithm in both effectiveness and efficiency.
Effective image annotation based on the diverse density algorithm and keywords correlation
Keping Wang, Zhigang Zhang
Automatic image annotation is significance for image understanding and retrieve of web image, so it becomes the new hot research topic in recent years. This paper proposes an effective annotation method based on the Diverse Density (DD) algorithm and keywords correction. The method includes two sub-processes, basic image annotation and annotation refinement. In the basic label process, we use the improved DD algorithm to find the visual feature vector for some semantic concept. The general DD algorithm uses all instances in the positive bags as start points of the optimization process, which will greatly increase the computing time. We cluster the same visual feature regions in the positive bags and use the clustering centers as start points instead of all instances. Moreover, the negative instances have been used to guild the selection of start point. Then, we integrate the improved DD algorithm into the Bayesian framework to realize the initial image annotation. In the annotation refinement, the correlations between keywords are added to refine those candidate annotations from the prior process. Finally, experimental results and comparisons on the Corel image set are given to illustrate the performance of the new algorithm.
An improved codebook model for detecting moving object under complex dynamic background
Xianyong Fang, Biao He, Bin Luo, et al.
Background subtraction is popular in moving object detection, but it is weak under the complex dynamic background where there are serious variations of illumination and color in the background pixels. This paper studies one of the background subtraction method, codebook, and propose an improved codebook model to resolve this problem. The novelty of the model lies in two aspects: 1) it applies HSV color space into the block-based codebook for robust codebook construction; 2) it also includes an efficient correction mechanism to update the codebook and eliminate false targets during detection. Experimental results demonstrate the proposed codebook model can detect the moving target effectively.
Peripheral nerve enhancement based on multi-scale Hessian matrix
Xiuli Ma, Hui Li
To improve the precision of nerve segmentation in CT images, a new comparability function is proposed in this paper to enhance the contrast between nerve structure and other surrounding tissues. It is based on nerve's characteristic, i.e. dark tubular structure, and a thorough analysis of the multi-scale Hessian matrix. By comparability function, the gray range of interested nerve structure can be automatically determined, which combines the multi-scale Hessian matrix eigenvalues with intensity information of original nerve CT images. The experimental results show that the improved algorithm can not only enhance the continuous nerve of tubular structure, but also clearly reflect its bifurcations and crossovers. It is very important and significant to the computer-aided disease diagnosis of peripheral nervous system.
A novel robust digital image watermarking method using SVD and GA
F. Golshan, K Mohammadi
A novel evolutionary-based watermarking algorithm for digital images is proposed. Robustness and imperceptibility are two important properties in digital image watermarking which compete with each other. In this paper a DCT and SVD based intelligent algorithm is applied to make a tradeoff between these two properties. First of all, a cover image is divided into 8×8 blocks and some of them which are special ones are transformed to DCT domain. The singular value decomposition is applied to DCT coefficients and singular values change according to a binary watermark image. The binary watermark image is obtained by Genetic Algorithm to solve the optimization problem between robustness and imperceptibility. So the novelty of this method is image adaptability. Robustness of the proposed method against several attacks such as filtering, noise contamination, JPEG compression and some geometrical attacks is good. In comparison with a recently similar existing work, experimental results show improvement in both imperceptibility and robustness.
Driver fatigue recognition based on supervised LPP and MKSVM
Wei Huang, Wei Zhang
Driver fatigue is a significant factor in many traffic accidents. In this paper, a novel approach is proposed to recognize driver fatigue. First of all, in order to extract effective feature of fatigue expression from face images, supervised locality preserving projections (SLPP) is adopted, which can solve the problem that LPP ignores the within-class local structure by adopting prior class label information. And then multiple kernels support vector machines (MKSVM) is employed to recognizing fatigue expression, Compared to SVM, which can improve the interpretability of decision function and performance of fatigue recognition. Experimental results are shown to demonstrate the effectiveness of the proposed method.
Estimating smoothness term for BP stereo matching with sparse disparities
Junxue He, Zhanming Li
The belief propagation stereo matching algorithm has high computational cost and varied parameters. To address these problems and perform stereo matching in practical applications, we implement BP algorithm on cuda and propose an approach to estimate parameters of smoothness term in the energy function automatically. We use Harris corner detector and Pyramidal KLT feature tracker algorithm to extract a sparse disparity map. The rate of increase in the discontinuity cost can be compute out approximately. Experiments demonstrate that our approach is feasible and the approximate parameters of discontinuity cost can be computed out quickly.
Localization of facial region in digital images
Raj Kumar Gupta, Aditya Chowdhury, Rahul Roy
We have developed and implemented an algorithm for the localization of facial region in a digital image consisting of multiple faces. The algorithm utilizes the basic colour-segmentation methods where the skin and hair regions are identified using the standard colour models. However, the implementation of merely the skin and hair models yields both the facial and non-facial regions. In order to filter out the non-facial region, we have introduced a quantization and a filtering module. The filter module essentially evaluates the proximity of the connected components associated with that of skin and hair regions. We have tested the algorithm on various images under various conditions. We found that the algorithm is capable of localizing the facial region even in a harsh condition.
Based on linear spectral mixture model (LSMM) unmixing remote sensing image
Jiaodi Liu, Weibin Cao
There are mixed pixels in remote sensing images ordinarily, this is a difficulty of the pixel classification (ie, unmixing) in remote sensing image processing.Linear spectral separation, estimating the value end of Genpo degree, for spatial modeling, through the non-constrained mixed pixel decomposition,with cotton, corn, tomatoes and soil four endmembers to decompose mixed pixels, Got four endmember abundance images and the RMS error image, the planting area of cotton and cotton-growing area of the measurement in the decomposition of mixed pixel block, and obtained unmixing accuracy. Experimental results show that: a simple linear mixed model modeling, and computation is greatly reduced, high precision, strong adaptability.
Complex diffusion and shock filter processes for image enhancement and denoising
Zhu Yan, Mingju Chen, Pingxian Yang
The diffusion process can simultaneously enhance, sharpen and denoise image. The diffusion coefficient is locally adjust according to image gradient, so it has many formats diffusion process according to the set of criteria, suck as P-M diffusion,complex diffusion and forward and backward diffusion. In the complex diffusion, the imaginary part of image serve as approximate second derivative of image. So using the imaginary to control the diffusion coefficient can combine the forward and backward complex diffusion.The forward and backward complex diffusion choice forward or backward diffusion according to the imaginary part. The forward diffusion denoise and smooth the image,and the backward diffusion magnify the noise and sharpen the edges, so the the forward and backward complex diffusion can't denoise detail part of image effectively.The shock filer sharpen the edges can take for inverse diffusion, in the paper,we use the imaginary part to control the shock filter as the backward part and the nonlinear complex diffusion as the forward part to combine a novel forward and backward complex diffusion. This novel isn't selectivity diffusion but forward and backward diffusion both diffuse simultaneously.The results of experiments demonstrate the novel denoise the image and retain the details more effective than the forward and backward complex diffusion.
A novel data processing technique for image reconstruction of penumbral imaging
Hongwei Xie, Hongyun Li, Zeping Xu, et al.
CT image reconstruction technique was applied to the data processing of the penumbral imaging. Compared with other traditional processing techniques for penumbral coded pinhole image such as Wiener, Lucy-Richardson and blind technique, this approach is brand new. In this method, the coded aperture processing method was used for the first time independent to the point spread function of the image diagnostic system. In this way, the technical obstacles was overcome in the traditional coded pinhole image processing caused by the uncertainty of point spread function of the image diagnostic system. Then based on the theoretical study, the simulation of penumbral imaging and image reconstruction was carried out to provide fairly good results. While in the visible light experiment, the point source of light was used to irradiate a 5mm×5mm object after diffuse scattering and volume scattering. The penumbral imaging was made with aperture size of ~20mm. Finally, the CT image reconstruction technique was used for image reconstruction to provide a fairly good reconstruction result.
Discharge light and carbonization distribution characteristics at XLPE-silicon rubber interface with silicon-grease in tracking failure test
Liang Gu, W. Liu, S. L. Lei, et al.
Imaging processing method was adopted to investigate the effect of silicon grease on tracking failure of the XLPEsilicon rubber interface by analyzing the distribution characteristics of discharge light and carbonization at the interface. Three interfaces were set up by pressing together a slice of XLPE and a slice of transparent silicon rubber. One filled silicon grease and the other partly filled the grease. As comparison, the third one filled on grease. High voltage (AC 50 Hz) was applied on a pair of flat-round electrodes sandwiched at the interface with their insulation distance of 5 mm. When the test voltage was raised to a certain value, discharge occurred and discharge light appeared and carbonization accumulated at the interface. The discharge light from discharge to the failure and the carbonization after the failure was recorded with a digital video recorder and then the images were analyzed with image processing method. Obtained results show that silicon grease at the interface weakens the transportation of charge and enhances the interfacial breakdown strength. However, interfacial discharge and tracking failure easily occur once discharge appears. Image processing method is helpful to understand the tracking failure process and mechanism of XLPE cable joint.
A new framework for the fusion of object and scene based on IHS transform
You-dong Ding, Xiao-cheng Wei, Hai-bo Pang, et al.
By introducing the concept of IHS transform and intensity modulation, this paper proposes a new framework for the fusion of object and scene. A comparative experiment within the framework using the standard IHS transform fusion technique and wavelet technique as intensity fusion tools respectively has proved its practicality well. Furthermore, regarding the particularity of the fusion of object and scene, we adopt a more appropriate assessment scheme which combines local quality evaluation method with global quality evaluation method to assess the fusion quality objectively besides subjective evaluation.
Multispectral color space representation based on BP neutral network
Qiang Liu, Xiaoxia Wan, Chong Teng
A new multispectral color space representation workflow was proposed, together with a color group classification method which based on the BP neutral network. This workflow was applied to multispectral color representation in color datasets grouping experiment and achieved excellent spectral representation accuracy.
Human multimedia display interface based on human activity recognition
Yiting Shang, Eung-Joo Lee
In this paper, we will propose a Human Multimedia Display Interface. The interface uses the tracking of human hand movements to control the IP-TV. This paper presents an improved CAMSHIFT algorithm to control an IP-TV system. The CAMSHIFT algorithm (Continuously Adaptive MeanShift) is a method of using color information[1]. It can do tracking with a specific color of the target. In some typical environmental constraints, it can obtain good tracking performance. However, as the question of noise, large area similar to the color interference and so on, only by CAM-SHIFT algorithm it is not competent. Against these issues we propose an improved CAMSHIFT algorithm[2].
A robust and fast super-resolution for license plate in traffic videos
Junxi Sun, Jun Wu, Guangqiu Chen, et al.
A robust and fast super-resolution method to enhance the license plate of moving vehicles is proposed. The method contains three steps: sub-pixel motion estimation, irregular data interpolation and edge preserved image restoration. Experiment results show the proposed algorithm has better performance than others.
An information hiding method based on LSB and tent chaotic map
Jianhua Song, Qun Ding
In order to protect information security more effectively, a novel information hiding method based on LSB and Tent chaotic map was proposed, first the secret message is Tent chaotic encrypted, and then LSB steganography is executed for the encrypted message in the cover-image. Compared to the traditional image information hiding method, the simulation results indicate that the method greatly improved in imperceptibility and security, and acquired good results.
Face recognition using DWT compression and PSO-based DCT feature selection
Yigui Sun, Dexiang Zhou
In this paper, a robust face recognition algorithm based on Discrete Wavelet Transform (DWT), Discrete Cosine Transform (DCT) and Particle Swarm Optimization (PSO) is presented. Initially, 2D-DWT is used to compress the data at various levels, which also removes the high frequency noise from the input image. Then DCT is applied to the resulting image to extract coefficients. Finally, the proposed PSO-based feature selection algorithm is utilized to search the feature space for the optimal feature subset where features are carefully selected according to a well defined discrimination criterion. Experimental results compared to the recently proposed algorithms on the ORL face database show that the proposed approach is promising; it is able to select small subsets and still improve the classification accuracy.
Chinese wine classification system based on micrograph using combination of shape and structure features
Chinese wines can be classification or graded by the micrographs. Micrographs of Chinese wines show floccules, stick and granule of variant shape and size. Different wines have variant microstructure and micrographs, we study the classification of Chinese wines based on the micrographs. Shape and structure of wines' particles in microstructure is the most important feature for recognition and classification of wines. So we introduce a feature extraction method which can describe the structure and region shape of micrograph efficiently. First, the micrographs are enhanced using total variation denoising, and segmented using a modified Otsu's method based on the Rayleigh Distribution. Then features are extracted using proposed method in the paper based on area, perimeter and traditional shape feature. Eight kinds total 26 features are selected. Finally, Chinese wine classification system based on micrograph using combination of shape and structure features and BP neural network have been presented. We compare the recognition results for different choices of features (traditional shape features or proposed features). The experimental results show that the better classification rate have been achieved using the combinational features proposed in this paper.
Transmission line galloping image monitoring system based on digital signal processor
Hai Peng Ren, Zhan Feng Ma
An embedded image monitoring system based on TMS320DM642 Digital Signal Processor (DSP) is proposed for the transmission line monitoring. The system can detect galloping, ice or snow covering, and other abnormal status of the transmission line in a real time mode. The image detection algorithms are compared using the controlled experiment under the complex weather environment, thereby, a set of image processing algorithms is proposed for transmission lines image monitoring. The DSP/BOIS multi-threaded programming techniques are used to realize the algorithm in the DSPs' embedded software. A wireless communication based on General Packet Radio Service (GPRS) module is designed to transmit the detection results and the changed information of the image to the monitoring center, so that the operators can get the real time status of the transmission line. The application of the system will play an important role in the condition-based maintenance of power transmission lines and improve the reliability of power delivery system.
Research of image segmentation based on genetic algorithm
Meining Zhu, Zhili Hu, Xiumin Chen
Through the in-depth study on the existing genetic algorithms and image segmentation technologies, based on the practical application of image segmentation, this paper improves the basic genetic algorithm and applies it in image segmentation. Practices show that this can simplify the calculation steps, consequently improve the image segmentation efficiency.
Application to monitoring of tailings dam based on 3D laser scanning technology
Fang Ren, Aiwu Zhang
This paper presented a new method of monitoring of tailing dam based on 3D laser scanning technology and gave the method flow of acquiring and processing the tailing dam data. Taking the measured data for example, the author analyzed the dam deformation by generating the TIN, DEM and the curvature graph, and proved that it's feasible to global monitor the tailing dam using 3D laser scanning technology from the theory and method.
Aerial lidar data classification using weighted support vector machines
Ning Guo, Gang Xu
This paper presents our research on classifying scattered 3D aerial Lidar height data into ground, vegetable (trees) and man-made object (buildings) using Support Vector Machine algorithm. To this end, the most basic theory of SVM is first outlined and with concern to the fact that features are differed in their contribution to classification, Weighted Support Vector Machine (W-SVM) technique is proposed. Second, four features consist of height, height variation, plane fitting error and Lidar return intensity are identified for classification purposes. In this step, features are normalized respectively and their weight that indicates feature's contribution to certain class or multi-class as a whole are calculated and specified. Third, Based on W-SVM technique, one 1AAA1 solution to multi-class classification is proposed by integration "one against one" and "one against all" solution together. Finally, the classification results of LIDAR data with presented technique clearly demonstrate higher classification accuracy and valuable conclusions are given as well.
Robust image matching based on wavelet transform and SIFT
Jian-fang Dou, Jian-xun Li
Template matching is the process of determining the presence and the location of a reference image or an object inside a scene image under analysis by a spatial cross-correlation process. Conventional cross-correlation type algorithms are computationally expensive. In this paper, an algorithm for a robust template matching method based on the combination of the wavelet transform method and SIFT is proposed. Discrete wavelet transform is done firstly on a reference image and a template image, and low frequency parts of them is extracted, then we use harris corner detection to detect the interesting point in low frequency parts of them to determined the matching candidate region of template image in reference image, extracting SIFT features on the matching candidate region and template image, The extracted SIFT features are matched by k-d tree and bidirectional matching strategy. Experiment show that, the algorithm can improve the accuracy of matching and at the same time to reduce the computation load.
Selection of regularization parameter based on synchronous noise in total variation image restoration
Peng Liu, Dingsheng Liu, Zhiwen Liu
In this article, we apply the total variation method to image restoration. We propose a method to calculate the regularization parameter in which we establish the relationship between the noise and the regularization parameter. To correctly estimate the variance of the noise remaining in image, we synchronously iterate a synthesized noise with the observed image in deconvolution. We take the variance of the synthesized noise to be the estimate of the variance of the noise remaining in the estimated image, and we propose a new regularization term that ensures that the synthetic noise and the real noise change in a synchronous manner. The similarity in the statistical properties of the real noise and the synthetic noise can be maintained in iteration. We then establish the relationship between the variance of synthetic noise and the regularization parameter. In every iteration, the regularization parameter is calculated by using the formula proposed for the relationship. The experiments confirm that, by using this method, the performance of the total variation image restoration is improved.
A quick algorithm of counting flow accumulation matrix for deriving drainage networks from a DEM
Yanping Wang, Yonghe Liu, Hongbo Xie, et al.
Computerized auto-extraction of drainage networks from Digital Elevation Model (DEM) has been widely used in hydrological modeling and relevant studies. Several essential procedures need to be implemented in eight-directional(D8) watershed delineation method, among which a problem need to be resolved is the lack of a high efficiency algorithm for quick and accurate computation of flow accumulation matrix involved in river network delineations. For the problem of depression filling, the algorithm presented by Oliver Planchon has resolved it. This study was aimed to develop a simple and quick algorithm for flow accumulation matrix computations. For this purpose, a simple and high efficiency algorithm of the time complexity of O(n) compared to the commonly used code of the time complexity of O(n2) orO(nlogn) , has been developed. Performance tests on this newly developed algorithm were conducted for different size of DEMs, and the results suggested that the algorithm has a linear time complexity with increasing sizes of DEM. The computation efficiency of this newly developed algorithm is many times higher than the commonly used code, and for a DEM of size 1000*1000, flow accumulation matrix computation can be completed within only several seconds compared with about few minutes needed by common used algorithms.
Effect of water on neural-network-based soil image recognizer and classifier
Nagaraj V Dharwadkar, D. G. Savakar, S. S. Panchal, et al.
From the last few years, more attention has been directed towards the usage of information technology in agriculture. This new way of farming offers the promise of improving farm profitability. Using Internet the farmers can collect data like geographical- referred yield, weather, soil and other important data related to farming. The aim is to use these data to produce area-specific crop production decisions. For increasing the production quality of crop soil plays very important role. To help the farmer in deciding how to increase the crop quality based on soil. We have proposed, soil image recognizer and classifier, which classifies soil image samples based on their color and morphological features. Different types of soil image samples considered like red soil, black soil, black cotton soil. Using color and morphological features a Neural Network Based Classifier is designed. The effect of water on the soil image classifier is analyzed by adding the water into different portions of soil samples. The accuracy of the soil image classifier is improved by considering wet soil samples.
A strategy of car detection via sparse dictionary
Guo-Qing Jin, Ying-Hui Dong
In recent years there is a growing interest in the study of sparse representation for object detection. These approaches heavily depend on local salient image patches, thus weakening the global contribution to the object identification of other less informative signals.Our generic approach not only employs the informative representation by linear transform, but also keeps all the spatial dependence implied among the objects. As an example,car images can be represented using parts from a vocabulary, along with spatial relations observed among them.Our approach is conducted with the quantitative measurement in developing the car detector at every stage. The theory underneath the optimal solution is the maximal mutual information carried out by the system. Our goal is to keep the maximal mutual information transmitted from stage to stage so that only the least uncertainty about the class identification remains based on the observation of classifier's output.
Warship radar cross section determination and reduction, and hindrances in optimizing radar cross section reduction on warships
Jawad Khan, WenYang Duan
From the begining of military warfare, it has always been extremely important to know the enemy position and hide oneself to capitalize on elements of surprise and initiative, and same is true for naval warfare. Radar is the primary instrument used for detecting enemy platforms today.Radar detects a target by clocking time taken by a known pulse of electromagnetic energy to get to the target and return. Radar cross section (RCS) is the measure of reflective strength of a target. Reducing the RCS of a platform implies its late detection, used to capitalize on surprise and initiative. RCS is also important for survivability evaluation since most modern weapons use installed radars during final engagement phase. As a result, RCS of a warship has transformed into a very important design factor for stealth to achieve surprise, initiative and survivability. Thus accurate RCS determination and RCS reduction are matters of extreme importance. The purpose of this study is to provide an understanding RCS reduction and RCS determination methods used on warships today. In doing so, this study will discuss importance of RCS, radar fundamentals and RCS basics, RCS reduction and RCS determination methods. It will also present hindrances in optimizing RCSR on warships, impact of these hindrances on navies around the world, and comment on possible remedies to these hindrances.
Supervised sparsity preserving projections for face recognition
Yanfeng Sun, Jiangang Zhao, Yongli Hu
Sparsity preserving projection (SPP) is a recently proposed unsupervised linear dimensionality reduction method for face recognition, which is based on the recently-emerged sparse representation theory. It aims to find a low-dimensional subspace to best preserve the global sparse reconstructive relationship of the original data. In this paper, we propose a supervised variation on SPP called supervised sparsity preserving projection (SSPP). The SSPP method explicitly takes into account the within-class weight as well as between-class weight and assigns different weights to them, which attempts to strengthen the discriminating power and generalization ability of embedded data representation. The effectiveness of the proposed SSPP method is verified on two standard face databases (Yale, AR).
Lossless image compression utilizing reference points coding
Yi-Fei Tan, Wooi-Nee Tan, Kae-Yann Tan
This paper proposes a lossless image compression method utilizing the neighboring pixels to determine the reference point values. The proposed method scans every pixel row by row and assigns a 2-bit reference point value to each pixel by comparing its intensity value to the neighboring pixels' intensity values. The intensity value will be stored to a new file only when the comparison fails to find a neighborhood pixel with the same intensity value. The compression is achieved as only the information of 2-bit reference point values for all pixels and certain intensity values are required for storage. The suggested method is tested on various types of images and the results show that it performs well for most of the images.
Calibration of digital camera IOP using radial alignment constrain and vanish point geometry
Gang Xu, Zenglai Dong, Jiejun Wang
This paper presents our research on calibrating IOP of digital camera with mono-view. First, the Radial Alignment Constrain (RAC) from Tsai's two-stage calibration is outlined and extended to handle with camera principle point parameters. Second, we relate principle point to other camera parameters by means of two vanishing points from the perspective projection of group lines, parallel to World Coordinate System X, Y axis, respectively, and with the relation, the function-dependence among parameters within the equation derived from extended RAC are eliminated. Third, based on the extended RAC, a new nonlinear equation for IOP calibration purpose is formulated and optimally solved with Least Square technique as well as initial value for parameters are determined. Finally, a set of stimulated images generated with virtual plane grid lines and known camera parameters are used for calibration experiment. The comparison of proposed approach with Tsai's two-stage calibration is also given in this paper and valuable conclusions are conducted as well.
Progressive image transmission using pyramid structure and pattern matching coding
A pyramidal data structure suited for coding and progressive transmission of images is proposed in this work. A mean pyramid representation of an image is first built up by forming a sequence of reduced-size images. A pyramid of difference images is then generated by subtracting the previous coded image from the original image at each level of the pyramid. Progressive transmission is achieved by sending all the nodes in the difference pyramid starting from the top level and ending at the bottom level. To gain efficiency, a pattern matching- based coding algorithm is applied to the difference pyramid of the image on a level-by-level basis. The proposed coding method, compresses the difference images by using a set of parameters computed based the visual activity of individual image blocks. The coding efficiency of the proposed technique along with the low computational complexity and simple parallel implementation of the pyramid approach allows for a high compression ratio as well as a good image quality. Satisfactory coded images have been obtained at bit rates in the range of 0.30 - 0.33 bits per pixel (bpp).
Random ambience using high fidelity images
Nur Azman Abu, Shahrin Sahib
Most of the secure communication nowadays mandates true random keys as an input. These operations are mostly designed and taken care of by the developers of the cryptosystem. Due to the nature of confidential crypto development today, pseudorandom keys are typically designed and still preferred by the developers of the cryptosystem. However, these pseudorandom keys are predictable, periodic and repeatable, hence they carry minimal entropy. True random keys are believed to be generated only via hardware random number generators. Careful statistical analysis is still required to have any confidence the process and apparatus generates numbers that are sufficiently random to suit the cryptographic use. In this underlying research, each moment in life is considered unique in itself. The random key is unique for the given moment generated by the user whenever he or she needs the random keys in practical secure communication. An ambience of high fidelity digital image shall be tested for its randomness according to the NIST Statistical Test Suite. Recommendation on generating a simple 4 megabits per second random cryptographic keys live shall be reported.
Application of the digital watermarking technique in 2D barcode certificate anti-counterfeit systems
MuSheng Chen, ShunDa Lin
At present, two dimensional barcode has been used in many fields. The safety of information in barcode is important, so this article brings up an effective two dimensional barcode encryption technology to assure it. Either two-dimensional barcode or digital watermarking technique is one of the most important parts and research focuses in anti-counterfeit fields. This paper designs and realizes a whole set of certificate administration system based on QRcode. On this platform the digital watermarking technique based on the spatial domain is used to encrypt the two dimensional barcode. The combination of two dimensional barcode and digital watermarking can improve the security and secrecy of personal information, and realize real anti-counterfeit certificates.
Integrating TPS and level transformation for the mosaic of high-resolution image sequence from civil unmanned aerial vehicle
Jun Wu, Gang Xu, Guangjing He
Low-cost high-solution imagery is attractive to various applications. This paper presents our research on the automatic mosaic of high-resolution image sequence from civil Unmanned Aerial Vehicle (UAV). First, the image geometry distortions resulted from perspective projection are discussed and based on them, the key issue to the mosaic of images from UAV are described. Second, two common techniques for the registration of remote sensing image, homography alignment and Thin Plate Spline (TPS) transformation, are outlined, compared and based on their complement with each other, the integration of two techniques for the mosaic of long-ranged image with geometry distortion is proposed. Third, one leveled TPS technique is developed to integrate homography with TPS as well as its parameters are estimated with level transformation or Least Square Technique. Finally, the Mosaic of UAV high-resolution Image sequence is experimented with proposed approach and valuable conclusions are conducted as well.
Vision-based obstacle recognition system for automated lawn mower robot development
Zalhan Mohd Zin, Ratnawati Ibrahim
Digital image processing techniques (DIP) have been widely used in various types of application recently. Classification and recognition of a specific object using vision system require some challenging tasks in the field of image processing and artificial intelligence. The ability and efficiency of vision system to capture and process the images is very important for any intelligent system such as autonomous robot. This paper gives attention to the development of a vision system that could contribute to the development of an automated vision based lawn mower robot. The works involve on the implementation of DIP techniques to detect and recognize three different types of obstacles that usually exist on a football field. The focus was given on the study on different types and sizes of obstacles, the development of vision based obstacle recognition system and the evaluation of the system's performance. Image processing techniques such as image filtering, segmentation, enhancement and edge detection have been applied in the system. The results have shown that the developed system is able to detect and recognize various types of obstacles on a football field with recognition rate of more 80%.
Geometric moment based nonlocal-means filter for ultrasound image denoising
Yangchao Dou, Xuming Zhang, Mingyue Ding, et al.
It is inevitable that there is speckle noise in ultrasound image. Despeckling is the important process. The original nonlocal means (NLM) filter can remove speckle noise and protect the texture information effectively when the image corruption degree is relatively low. But when the noise in the image is strong, NLM will produce fictitious texture information, which has the disadvantageous influence on its denoising performance. In this paper, a novel nonlocal means (NLM) filter is proposed. We introduce geometric moments into the NLM filter. Though geometric moments are not orthogonal moments, it is popular by its concision, and its restoration ability is not yet proved. Results on synthetic data and real ultrasound image show that the proposed method can get better despeckling performance than other state-of-the-art method.
Fast image matching algorithm based on projection characteristics
Lijuan Zhou, Xiaobo Yue, Lijun Zhou
Based on analyzing the traditional template matching algorithm, this paper identified the key factors restricting the speed of matching and put forward a brand new fast matching algorithm based on projection. Projecting the grayscale image, this algorithm converts the two-dimensional information of the image into one-dimensional one, and then matches and identifies through one-dimensional correlation, meanwhile, because of normalization has been done, when the image brightness or signal amplitude increasing in proportion, it could also perform correct matching. Experimental results show that the projection characteristics based image registration method proposed in this article could greatly improve the matching speed, which ensuring the matching accuracy as well.
Automatic image-based detection of paper fiber ends
Juha Hirvonen, Pooya Saketi, Pasi Kallio
Understanding the properties of paper fibers and paper fiber bonds can be really significant in improving the quality of paper. The problem in gathering measurement data from individual paper fibers is the lack of reliable and efficient research instruments. Also, without automation the yield of experiments will be low and the results will depend on the skills of the operator. This paper presents an image-based method to automatically detect the endpoints of paper fibers. The method will be utilized in automatic control of a novel and tailor-made paper fiber manipulation and measurement platform. Performance of the method is extremely promising with adequate speed and very accurate results.
A vehicle license plate localization method using color barycenters hexagon model
Xiaolei Wan, Jingbo Liu, Jin Liu
Detecting and extracting the region of a license plate is the challenging first step in the vehicle license plate recognition system. In this paper, we propose a new approach for vehicle license plate localization using a Color Barycenters Hexagon (CBH) model. In CBH model, full color images are calculated the color barycenters and get the barycenters region, then automatic select the idea threshold curves to separate the Region of Interest (ROI) of barycenters aiming to localize the region of the vehicle license plate. Experimental results demonstrate that our method is effective, has strong practicability. Even if the practically images have many noise, our method can ideally localize vehicle license plate on various scene images when source image with suitably thresholding and operations.
Quantized image patches co-occurrence matrix: a new statistical approach for texture classification using image patch exemplars
Zhonghua Liu, Jingyan Wang, Yongping Li, et al.
The statistical distribution of image patch exemplars has been shown to be an effective approach to texture classification. In this paper, the joint distribution of pairs of patches for texture classification from single images is investigated. We developed a statistical method of examining texture that considers the spatial relationship of image patches, which is called the quantized patches co-occurrence matrix (QPCM). In our method, the images are first slipt into small image patches, and then the patches are quantized to the closest patch cluster centers (textons) which is learned form training images. By calculating how often pairs of patches with specific quantized values (texton labels) and in a specified spatial relationship occur in an image, we create the QPCM for images representation. Moreover, we developed a fusion framework for texture classification by fusing 4 QPCM functions with specified neighboring spatial relationship and 3 other statistical representations of image patches, which is called QPCM-SVM classifier. The effectiveness of the proposed texture classification methodology is demonstrated via an extensive consistent evaluation in standard benchmarks that clearly shows better performance against state-of-the-art statistical approach using image patch exemplars.
Image super-resolution using the specialized interior-point method
Guang-mei Ren, Xiao-feng Li, Ning Zhou
Using sparse representation for image super-resolution (SR) has been a hot topic in image processing recently. Many of them use the standard interior-point method to solve the l1-minimization problems which is essential in sparse representation. In this paper, we present a specialized interior-point method to do the image SR by using Preconditioned Conjugate Gradient (PCG) to compute the search step, and get the state-of-the-art results both qualitatively and quantitatively. The results showed that our method can achieve higher PSNR than conventional ones.
An automatic image mosaic algorithm based on SURF detector
Meide Shao, Ping Shi, Xuyuan Gu, et al.
In this paper, an image mosaic method based on SURF feature matching is proposed, which consists of the following steps: First, detecting SURF feature descriptor and finding matching pairs by mutual matching; Then, modified PROSAC (Progressive Sample Consensus) is proposed to remove the mismatch pairs and estimate homography matrix; Finally, adjusting the image to the same coordinate by transformation matrix and blending image by weight averaged linear blending. The experiment results show that the proposed method can detect descriptor rapidly and blend images accurately.
An improved video matting technique based on GrabCut
Lei Yang, Chunxi Ma, Xiaoyu Wu, et al.
Now GrabCut algorithm is a comparatively ideal interactive image segmentation method, and it has the advantages of high precision segmentation and low interaction. But so far GrabCut operator is an interactive segmentation method only be used in static images. To solve this problem, this paper proposes a method of video matting, which is based on the GrabCut operator. This method is going to achieve video auto-mating by transmitting the matting result of a frame to the next frame. And this method improves the processing speed. By experiments, this method demonstrates a wonderful segmentation result.
False match elimination for face recognition based on SIFT algorithm
Xuyuan Gu, Ping Shi, Meide Shao
The SIFT (Scale Invariant Feature Transform) is a well known algorithm used to detect and describe local features in images. It is invariant to image scale, rotation and robust to the noise and illumination. In this paper, a novel method used for face recognition based on SIFT is proposed, which combines the optimization of SIFT, mutual matching and Progressive Sample Consensus (PROSAC) together and can eliminate the false matches of face recognition effectively. Experiments on ORL face database show that many false matches can be eliminated and better recognition rate is achieved.
Characteristic research on Hong Kong "I learned" series computer textbooks
Jinyan Hu, Zhongxia Liu, Yuanyuan Li, et al.
Currently, the construction of information technology textbooks in the primary and middle schools is an important content of the information technology curriculum reform. The article expect to have any inspire and reference on inland China school information technology teaching material construction and development through the analyzing and refining the characteristics of the Hong Kong quality textbook series - "I learn · elementary school computer cognitive curriculum".
P-M diffusion for image denoise with the fidelity term
Pingxian Yang, Mingju Chen, Jing Wang
Diffusion has received a lot of attention and has experienced significant developments, it can simultaneously enhance, sharpen and denoise image. The diffusion coefficient is locally adjust according to image features such as edges, textures, and moments, so it has many formats diffusion process according to the set of gradient. Suck as P-M diffusion and total variation. The aim of the present paper is to study the total variation then replace the smoothed intensity function with the P-M diffusion function and derive the fidelity term to get a novel nonlinear anisotropic P-M diffusion. And then deduce the new diffusion application in discrete two dimension space for image denoise. The results of experiments demonstrate the novel P-M diffusion denoise the image and retain the details more effective than traditional P-M diffusion.
Wavelet video encoder with rate-control algorithm
Zhijie Zhao, Li Ma, Xuesong Jin, et al.
In this paper, we investigate the rate-distortion (R-D) performance of the embedded wavelet coder and the frame dependency between the reference frame and the predictive frame. Then, according to the rate-distortion model and the frame dependency to solve the rate control problem to a bit allocation problem for each frame. Experimental results show that the superior performance of the embedded wavelet video coder with the proposed rate control scheme.
An algorithm for identifying digital image orientation based on c#
Da-Chun Jia, Xu-Dong Yao, Xin Jia
In research for machine vision, the discrimination of image orientation is one of practical problems. An algorithm for identifying image orientation was proposed, and to be realized by c# programming.
Natural scene text location oriented Chinese environment
Xiaopei Liu, Zhaoyang Lu, Jing Li
Natural scene text location is the precondition of the text recognition, a fast and effective technique is proposed for text location oriented Chinese environment. First, the color edge of the image is detected in consideration of the stroke directions of Chinese characters, then the collection of candidate text blocks is obtained by post-processing; Second, because text differs from non-text in texture, local binary pattern (LBP) operator is employed to extract the texture feature of candidate text block, then feed them into SVM classifier for verification, thus a great number of non-text blocks are excluded, and a higher accuracy rate is obtained. Finally, the algorithm is tested on our database, which includes 1000 images, and experimental results show that the proposed algorithm works well and very robust under the conditions of uneven lighting, blurring, and low contrast etc.
A component inspection algorithm based on low-dimensional image feature
Jianjie Wu, Yuhui Zhang
The images captured by image-array-based automatic optical inspection devices may be inconsistent in lightness, definition and uniformity, which greatly influence the accuracy of the inspection result. To solve such a problem, a component inspection algorithm based on low-dimensional image feature is proposed. It doesn't compare inspected image with standard image from pixel to pixel like traditional algorithms do. Instead it compares key image feature extracted from the image by designing feature functions and computing attributes. The essential of the algorithm is to transfer the original high-dimensional information contained within the image to one-dimensional feature data. Experiments show that the algorithm can ensure efficient and low-storage real-time component inspection.
Research on pavement crack recognition methods based on image processing
Yingchun Cai, Yamin Zhang
In order to overview and analysis briefly pavement crack recognition methods , then find the current existing problems in pavement crack image processing, the popular methods of crack image processing such as neural network method, morphology method, fuzzy logic method and traditional image processing .etc. are discussed, and some effective solutions to those problems are presented.
Tracking and identifying a magnetic spheroid target using unscented particle filter
Mingming Yang, Daming Liu, Liting Lian, et al.
In this paper we use the recursive Bayesian estimation method to solve the tracking and identification problem of a target modeled by an equivalent magnetic spheroid. Target positions, velocity, heading, magnetic moments and size are defined as the state vector, which is estimated from noisy magnetic field measurements by a sequential Monte Carlo based method known as particle filter. In order to improve the performance of the filter, the unscented Kalman filter is applied to generate the transition prior as the proposal distribution. A simulated experiment is given to test the performance of the unscented particle filter, and the results show that the filter is suitable for magnetic target's track and identification.
The analysis of image acquisition in LabVIEW
Wuni Xu, Lanxiang Zhong
In this paper, four methods of image acquisition in LabVIEW were described, and its realization principles and the procedures in combination with different hardware architectures were illustrated in the virtual instrument laboratory. Experiment results show that the methods of image acquisition in LabVIEW have many advantages such as easier configuration, lower complexity and stronger practicability than in VB and C++. Thus the methods are fitter to set the foundation for image processing, machine vision, pattern recognition research.
Moving object recognition from video sequence images based on wavelet transform and neural network
Kun Zhang, Cuirong Wang
In this paper, we present a new method of moving object recognition from video images based on wavelet transform and neural network. We first separate the target moving object images from the video Sequence Images, then use the wavelet coefficients statistics as the feature description which can form the training set for the Neural Network. The experimental result shows that this method is practicability, feasibility and good precision.
(2D)2PCA+(2D)2LDA: a new feature extraction for face recognition
Guohong Huang
In this paper, we combine the advantages of (2D)2PCA and (2D)2LDA, and propose a two-stage framework: "(2D)2PCA+(2D)2LDA". In the first stage, a two-directional 2D feature extraction technique, (2D)2PCA, is employed to condense the dimension of image matrix; in the second stage, the two-directional 2D linear discriminant analysis (2D)2LDA is performed in the (2D)2PCA subspace to find the optimal discriminant feature vectors. In addition, the proposed method can take full advantage of the descriptive information and discriminant information of the image. Experiments conducted on ORL and Yale face databases demonstrate the effectiveness and robustness of the proposed method.
A new method for extracting instantaneous frequency feature of emitter signal
Junfeng Tan, Xiaopeng Yan, Jiantao Wang, et al.
Currently, mostly used method for Instantaneous Frequency (IF) feature extraction of radar signal is two-dimensional graph, which is easy for people to distinguish, but not for machines. This article proposes a feature extraction algorithm based on instantaneous frequency of radiation source. Using the statistical change difference of different IF signals, we formulate the classification feature vectors describing the various types of signal pulse modulation laws, according which we do some classification experiments. The experiments prove that instantaneous frequency characteristics of radiation sources can characterize the change difference of different modulation signals and have some anti-noise performance. We are sure that this method can be used in signal sorting, and can be realized in practice.
The design and implementation of effective face detection and recognition system
Yigui Sun
In the paper, a face detection and recognition system (FDRS) based on video sequences and still image is proposed. It uses the AdaBoost algorithm to detect human face in the image or frame, adopts Discrete Cosine Transforms (DCT) for feature extraction and recognition in face image. The related technologies are firstly outlined. Then, the system requirements and UML use case diagram are described. In addition, the paper mainly introduces the design solution and key procedures. The FDRS's source-code is built in VC++, Standard Template Library (STL) and Intel Open Source Computer Vision Library (OpenCV).
Image segmentation based on kernel PCA and shape prior
Xiaoping Wan, Djamal Boukerroui, Jean-Pierre Cocquerez
The introduction of shape priori in the segmentation model ameliorates effectively the poor segmentation result due to the using of the image information alone to segment the image including noise, occlusion, or missing parts. But the presentation of shape via Principal Component Analysis (PCA) brings on the limitation of the similarity between the objet and the prior shape. In this paper, we proposed using Kernel PCA (KPCA) to capture the shape information - the variability. KPCA can present better shape prior knowledge. The model based on KPCA allows segmenting the object with nonlinear transformation or a quite difference with the priori shape. Moreover, since the shape model is incorporated into the deformable model, our segmentation model includes the image term and the shape term to balance the influence of the global image information and the shape prior knowledge in proceed of segmentation. Our model and the model based on PCA both are applied to synthetic images and CT medical images. The comparative results show that KPCA can more accurately identify the object with large deformation or from the noised seriously background.
Perimeter intrusion detection based on improved Surendra background update algorithm
Feng Ding, Hong Wang, Hongsheng Zhong, et al.
An improved Surendra background update algorithm for moving targets detection is proposed in this paper, which is beneficial to the background update and can automatically change threshold. Firstly, the initial background frame can be obtained through the arithmetic average to the first few frame images. Secondly, The OTSU method is used for adaptive threshold selection to background updating, so that the moving objects can be obtained by morphological processing. Experiments show that the improved Surendra background updating algorithm can detect the moving targets effectively and overcome the original method's shortcoming which needs to manually intervene the threshold. The impact of such background disturbances and illumination changes to moving targets detection has a good inhibition.
Approach of microscopic images mosaic revising based on adjacent image
Haishun Wang, Rong Wang, Limin Chen
Image mosaic splices several adjacent overlapped images into an integrated seamless picture which could be significant in medical image processing. However, because of image acquisition, a mismatch could occur as a result of adjacent image stitching data and cumulative errors. The current method which is effective in certain ways still has room for improvement regarding processing speed and effectiveness, particularly in accuracy. This paper proposed a new image mosaic revising algorithms based on the relativity of adjacent images location, expounding the principal of image mosaic unit, the equations on splicing parameters and the simplified rules as well as achieving automatic calculation through application of revised algorithm. By experiment, the 20 groups inaccurate pathological mosaic images were revised rapidly and accurately with error controlled within a pixel. It is proved that the approach is effective in revising the error matching in microscopic images mosaic.
Discussion about signals and systems teaching
Zhaodi Guo
The features of "Signals and Systems" were analyzed, and three methods were proposed to improve the effect of classroom teaching. These methods are combination of traditional teaching with multimedia teaching, guiding students to understand the physical meaning of the conclusion formula, and integrating theory with practice.
The design of infrared alarm system based on ZigBee technology
Qiang Yin, Haiyang Yu, Xingling Shao, et al.
In view of the existing shortcomings of the existing infrared system,such as complicated structure, high transmitting power, expensive cost and etc. A infrared alarm system based on ZigBee technology was presented in this paper. The system consists of owner-alarm, the monitoring room in the housing estate and network router. Through the alarm information detected by infrared sensors, it starts the way of local sound alarm or networking alarm. The results of test demonstrate that the system is convenient for use and reliable for performance.
Animation of shape-controlled smoke flow
Yongxia Zhou, Kangjian Wang, Lingmin He, et al.
We present a new method to produce smoke animations. By adding some forces, we efficiently controlled the smoke's target shape and kept smoke alive. Given an arbitrary object's shape represented by a geometric model, our method can generate an animation in which smoke flows out somewhere and quickly forms the shape. It looks like a cloud of moving smoke that fundamentally maintains the shape and fluid-like behavior. By adding two control forces to the physics-based free flow, the shape can be controlled well while smoke flowing. In our system, the free flow and the shape-controlled flow can be easily switched to each other. The additional computation to controlling the shape is negligible compared with the free flow.
Alarm system using pyroelectric infrared sensors for measuring distance and alarming
Haiyang Yu, Qiang Yin, Wei Yang, et al.
A novel and reliable alarm system consists of a pyroelectric sensor, an infrared len, amplified module, microcontroller and alarm module. According to different signal peak-to-peak values which person at different distances causes the system, the system can get the distance of the person and decide whether the system sends the information to computer. Before working the system needs to calibrate the person's characteristic values at different distances using the method described in this paper.
Design and implementation of a wireless video surveillance system based on ARM
Yucheng Li, Dantao Han, Juanli Yan
A wireless video surveillance system based on ARM was designed and implemented in this article. The newest ARM11 S3C6410 was used as the main monitoring terminal chip with the embedded Linux operating system. The video input was obtained by the analog CCD and transferred from analog to digital by the video chip TVP5150. The video was packed by RTP and transmitted by the wireless USB TL-WN322G+ after being compressed by H.264 encoders in S3C6410. Further more, the video images were preprocessed. It can detect the abnormities of the specified scene and the abnormal alarms. The video transmission definition is the standard definition 480P. The video stream can be real-time monitored. The system has been used in the real-time intelligent video surveillance of the specified scene.
Research of motion estimation algorithm based on H.264
Xiaoling Yao
Through the study on the existing motion estimation algorithms, for the problems existed in them, this paper improves the motion estimation algorithm. The improved algorithm reduces the complexity of the motion search algorithm through pre-judgment of motion types, dynamic selection of motion search patterns, early end of the search process and other technologies, which avoids the block matching calculation for some points with small probability, thus to achieve the purpose of reducing the complexity of the whole motion estimation algorithm, to improve the algorithm efficiency.
Medical image registration using MLS deformation
Lintao Zheng, Guiping Qian
We present a new method for medical image registration, which is based on the moving least squares (MLS) transformation. Under condition that some of structures should be minimally deformed during registration, our method can offer advantages over other non-rigid registration methods. Our method has been successfully used for medical image registration. The feasibility of the proposed method is demonstrated and compared with the TPS based image registration technique. The experimental results show that the propose method can get better results than the TPS method.
The MG/OPT algorithm for dense optical flow
El Mostafa Kalmoun, Luis Garrido, Vicent Caselles
We introduce the use of optimization-based multigrid techniques for dense optical flow computation. In particular, we evaluate the performance of a multigrid optimization (MG/OPT) algorithm based on a line search strategy for large-scale optimization like truncated Newton. Our experimental tests have shown that the algorithm outperforms the truncated Newton method even implemented with a coarse to fine strategy.
Algorithm comparison of regional economic resources niche categories
Firstly, regional economic classification was made by clustering algorithms, which could be referenced by Support Vector Machine classification. Then Support Vector Machine classification was made to classificatory regional economic resources niche. Furthermore, more detailed classification was made under different economic resources niche indicators, which solve the problem of inadequate samples, making Categories strong operational and practical significance. Thus, more reasonable classification could be made by support vector machine model through the establishment of relatively small samples.
A new adaptive object detection technique based on the wavelet co-occurrence features
Keivan Ashabi, Ali Mahmoodi, Elmira Vafabakhsh, et al.
This Object detection involves processing images for detecting, classifying, and tracking targets embedded in a background scene. This paper presents an adaptive algorithm for detecting a specified target objects embedded in visual images for tracking application. The developed algorithm employs a novel technique using the wavelet co-occurrence features for detecting object based on template matching. Several signatures as contrast, energy, entropy and maximum probability are computed from wavelet co-occurrence features for object window and compares with features of image windows. The results of the proposed algorithm are very adaptive in variant condition with clutters.
LMS filter for noise cancellation using Simulink
K. T. Talele, Ashish Shrivastav, Kunal Utekar, et al.
In this paper we have proposed the simplified implementation of adaptive noise cancellation using LMS filter. The LMS algorithm belongs to the family of stochastic gradient algorithms. It is one of the efficient algorithms in adaptive filtering.