An automatic segmentation method for building facades from vehicle-borne LiDAR point cloud data based on fundamental geographical data
Author(s):
Yongqiang Li;
Jie Mao;
Lailiang Cai;
Xitong Zhang;
Lixue Li
Show Abstract
In this paper, the author proposed a segmentation method based on the fundamental geographic data, the algorithm describes as following: Firstly, convert the coordinate system of fundamental geographic data to that of vehicle- borne LiDAR point cloud though some data preprocessing work, and realize the coordinate system between them; Secondly, simplify the feature of fundamental geographic data, extract effective contour information of the buildings, then set a suitable buffer threshold value for building contour, and segment out point cloud data of building facades automatically; Thirdly, take a reasonable quality assessment mechanism, check and evaluate of the segmentation results, control the quality of segmentation result. Experiment shows that the proposed method is simple and effective. The method also has reference value for the automatic segmentation for surface features of other types of point cloud.
Automatic registration of laser-scanned point clouds based on planar features
Author(s):
Minglei Li;
Xinyuan Gao;
Li Wang;
Guangyun Li
Show Abstract
Automatic multistation registration of laser-scanned point clouds is a research hotspot in laser-scanned point clouds registration. Some targets such as common buildings have plenty of planar features, and using these features as constraints properly can bring about high accuracy registration results. Starting from this, a new automatic multistation registration method using homologous planar features of two scan stations was proposed. In order to recognize planes from different scan stations and get plane equations in corresponding scan station coordinate systems, k-means dynamic clustering method was improved to be adaptive and robust. And to match the homologous planes of the two scan stations, two different procedures were proposed, respectively, one of which was based on the “common” relationship between planes and the other referenced RANSAC algorithm. And the transformation parameters of the two scan station coordinate systems were calculated after homologous plane matching. Finally, the transformation parameters based on the optimal match of planes was adopted as the final registration result. Comparing with ICP algorithm in experiment, the method is proved to be effective.
A point matching algorithm based on reference point pair
Author(s):
Huanxin Zou;
Youqing Zhu;
Shilin Zhou;
Lin Lei
Show Abstract
Outliers and occlusions are important degradation in the real application of point matching. In this paper, a novel point matching algorithm based on the reference point pairs is proposed. In each iteration, it firstly eliminates the dubious matches to obtain the relatively accurate matching points (reference point pairs), and then calculates the shape contexts of the removed points with reference to them. After re-matching the removed points, the reference point pairs are combined to achieve better correspondences. Experiments on synthetic data validate the advantages of our method in comparison with some classical methods.
Extraction of power lines from mobile laser scanning data
Author(s):
Qing Xiang;
Jonathan Li;
Chenglu Wen;
Pengdi Huang
Show Abstract
Modern urban life is becoming increasingly more dependent on reliable electric power supply. Since power outages cause substantial financial losses to producers, distributors and consumers of electric power, it is in the common interest to minimize failures of power lines. In order to detect defects as early as possible and to plan efficiently the maintenance activities, distribution networks are regularly inspected. Carrying out foot patrols or climbing the structures to visually inspect transmission lines and aerial surveys (e.g., digital imaging or most recent airborne laser scanning (ALS) are the two most commonly used methods of power line inspection. Although much faster in comparison to the foot patrol inspection, aerial inspection is more expensive and usually less accurate, in complex urban areas particularly. This paper presents a scientific work that is done in the use of mobile laser scanning (MLS) point clouds for automated extraction of power lines. In the proposed method, 2D power lines are extracted using Hough transform in the projected XOY plane and the 3D power line points are visualized after the point searching. Filtering based on an elevation threshold is applied, which is combined with the vehicle’s trajectory in the horizontal section.
Extraction of rod-like objects from vehicle-borne LiDAR data
Author(s):
Zhenzhen Wu;
Huiyun Liu;
Kuangyu Li;
Jie Mao;
Xitong Zhang
Show Abstract
A rod-like objects extraction method based on clustering is presented. Firstly, project the original point clouds onto the horizontal plane, divide them into grids, and take a single grid as data processing unit to remove the ground points; Secondly, make the grid based on processing data for point clouds detection and numbered, give the same attribute values and cluster the object points using eight neighborhood search method. Then, take the clustered single point clouds as processing units, take advantage of various object features, such as height, density projection, the projection area and shape to exclude the other object points progressively, and achieve the fine extraction of the rod-like objects. The experiment tests the validity of the method described in the text of the extraction of the rod-like objects in road environment.
Extraction of tree crowns from mobile laser scanning data using a marked point process model
Author(s):
Jonathan Li;
Yongtao Yu;
Haiyan Guan;
Zheng Gong
Show Abstract
For the purpose of realistic visualisation in 3D city models, we present a marked point process based method for extracting tree-crowns from mobile laser scanning (MLS) data. First, we apply a modified IDW interpolation to generate a geo-referenced feature image, by which a histogram analysis is applied to separate high objects(e.g. trees and lightpoles) from low objects(e.g. road, ground, low vegetation). Next, we calculate grey differences of each pixel with its neighbors to find the local maxima as potential tree-crown seeds, and then use a grouping-and-centralizing procedure to remove the redundants from the seeds. Finally, we employ a marked point process to the generated geo-referenced image via the seeds. Two experiments have been conducted to test the efficiency and feasibility of our tree-extraction algorithm using RIEGL VMX-450 MLS data.
Geometric validation of a mobile laser scanning system for urban applications
Author(s):
Haiyan Guan;
Jonathan Li;
Yongtao Yu;
Yan Liu
Show Abstract
Mobile laser scanning (MLS) technologies have been actively studied and implemented over the past decade, as their application fields are rapidly expanding and extending beyond conventional topographic mapping. Trimble’s MX-8, as one of the MLS systems in the current market, generates rich survey-grade laser and image data for urban surveying. The objective of this study is to evaluate whether Trimble MX-8 MLS data satisfies the accuracy requirements of urban surveying. According to the formula of geo-referencing, accuracies of navigation solution and laser scanner determines the accuracy of the collected LiDAR point clouds. Two test sites were selected to test the performance of Trimble MX-8. Those extensive tests confirm that Trimble MX-8 offers a very promising tool to survey complex urban areas.
Delineation of individual tree crowns for mobile laser scanning data
Author(s):
Rosen Wu;
Yiping Chen;
Chenglu Wen;
Cheng Wang;
Jonathan Li
Show Abstract
The information of individual trees plays an important role in urban surveying and mapping. With the development of Light Detection and Ranging (LiDAR) technology, 3-Dimenisonal (3D) structure of trees can be generated in point clouds with high spatial resolution and accuracy. Individual tree segmentations are used to derive tree structural attributes such as tree height, crown diameter, stem position etc. In this study, a framework is proposed to take advantage of the detailed structures of tree crowns which are represented in the mobile laser scanning (MLS) data. This framework consists of five steps: (1) Automatically detect and remove ground points using RANSAC; (2) Compress all the above ground points to image grid with 3D knowledge reserved; (3) Simplify and remove unqualified grids; (4) Find tree peaks using a heuristic searching method; (5) Delineate the individual tree crowns by applying a modified watershed method. In an experiment on the point clouds on Xiamen Island, China, individual tree crowns from MLS point cloud data are successfully extracted.
Road traffic sign detection and classification from mobile LiDAR point clouds
Author(s):
Shengxia Weng;
Jonathan Li;
Yiping Chen;
Cheng Wang
Show Abstract
Traffic signs are important roadway assets that provide valuable information of the road for drivers to make safer and easier driving behaviors. Due to the development of mobile mapping systems that can efficiently acquire dense point clouds along the road, automated detection and recognition of road assets has been an important research issue. This paper deals with the detection and classification of traffic signs in outdoor environments using mobile light detection and ranging (Li- DAR) and inertial navigation technologies. The proposed method contains two main steps. It starts with an initial detection of traffic signs based on the intensity attributes of point clouds, as the traffic signs are always painted with highly reflective materials. Then, the classification of traffic signs is achieved based on the geometric shape and the pairwise 3D shape context. Some results and performance analyses are provided to show the effectiveness and limits of the proposed method. The experimental results demonstrate the feasibility and effectiveness of the proposed method in detecting and classifying traffic signs from mobile LiDAR point clouds.
Feature selection for quality assessment of indoor mobile mapping point clouds
Author(s):
Fangfang Huang;
Chenglu Wen;
Cheng Wang;
Jonathan Li
Show Abstract
Owing to complexity of indoor environment, such as close range, multi-angle, occlusion, uneven lighting conditions and lack of absolute positioning information, quality assessment of indoor mobile mapping point clouds is a tough and challenging task. It is meaningful to evaluate the features extracted from indoor point clouds prior to further quality assessment. In this paper, we mainly focus on feature extraction depend upon indoor RGB-D camera for the quality assessment of point cloud data, which is proposed for selecting and screening local features, using random forest algorithm to find the optimum feature for the next step’s quality assessment. First, we collect indoor point clouds data and classify them into classes of complete or incomplete. Then, we extract high dimensional features from the input point clouds data. Afterwards, we select discriminative features through random forest. Experimental results on different classes demonstrate the effective and promising performance of the presented method for point clouds quality assessment.
Effects analysis of array geometry for resolving performance based on spatial average ambiguity function
Author(s):
Guofeng Zha;
Hongqiang Wang;
Yongqiang Cheng;
Yuliang Qin
Show Abstract
For analyzing the three dimension (3D) spatial resolving performance of Multi-Transmitter Single-Receiver (MTSR) array radar with stochastic signals, the spatial average ambiguity function (SAAF) was introduced. The analytic expression of SAAF of array radar with stochastic is derived. To analyze the effects of array geometry, comparisons are implemented for three typical array geometries including circular, decussate and planar configuration. Simulated results illustrate that the spatial resolving performance is better for the circular array than that of others. Furthermore, it is shown that the array aperture size and the target’s radial range are the main factors impacting the resolving performance.
Effect analysis and design on array geometry for coincidence imaging radar based on effective rank theory
Author(s):
Guofeng Zha;
Hongqiang Wang;
Zhaocheng Yang;
Yongqiang Cheng;
Yuliang Qin
Show Abstract
As a complementary imaging technology, coincidence imaging radar (CIR) achieves super-resolution in real aperture staring radar imagery via employing the temporal-spatial independent array detecting (TSIAD) signals. The characters of TSIAD signals are impacted by the array geometry and the imaging performance are influenced by the relative imaging position with respect to antennas array. In this paper, the effect of array geometry on CIR system is investigated in detail based on the judgment criteria of the effective rank theory. In course of analyzing of these influences, useful system design guidance about the array geometry is remarked for the CIR system. With the design guidance, the target images are reconstructed based on the Tikhonov regularization algorithm. Simulation results are presented to validate the whole analysis and the efficiency of the design guidance.
Design and implementation of H.264 based embedded video coding technology
Author(s):
Jian Mao;
Jinming Liu;
Jiemin Zhang
Show Abstract
In this paper, an embedded system for remote online video monitoring was designed and developed to capture and record the real-time circumstances in elevator. For the purpose of improving the efficiency of video acquisition and processing, the system selected Samsung S5PV210 chip as the core processor which Integrated graphics processing unit. And the video was encoded with H.264 format for storage and transmission efficiently. Based on S5PV210 chip, the hardware video coding technology was researched, which was more efficient than software coding. After running test, it had been proved that the hardware video coding technology could obviously reduce the cost of system and obtain the more smooth video display. It can be widely applied for the security supervision [1].
Semi-supervised feature learning for hyperspectral image classification
Author(s):
Pengfei Zhang;
Liujuan Cao;
Cheng Wang;
Jonathan Li
Show Abstract
Hyperspectral image has high-dimensional Spectral–spatial features, those features with some noisy and redundant information. Since redundant features can have significant adverse effect on learning performance. So efficient and robust feature selection methods are make the best of labeled and unlabeled points to extract meaningful features and eliminate noisy ones. On the other hand, obtaining sufficient accurate labeled data is either impossible or expensive. In order to take advantage of both precious labeled and unlabeled data points, in this paper, we propose a new semisupervised feature selection method, Firstly, we use labeled points are to enlarge the margin between data points from different classes; Secondly, we use unlabeled points to find the local structure of the data space; Finally, we compare our proposed algorithm with Fisher score, PCA and Laplacian score on HSI classification. Experimental results on benchmark hyperspectral data sets demonstrate the efficiency and effectiveness of our proposed algorithm.
An improved SIFT algorithm based on KFDA in image registration
Author(s):
Peng Chen;
Lijuan Yang;
Jinfeng Huo
Show Abstract
As a kind of stable feature matching algorithm, SIFT has been widely used in many fields. In order to further improve the robustness of the SIFT algorithm, an improved SIFT algorithm with Kernel Discriminant Analysis (KFDA-SIFT) is presented for image registration. The algorithm uses KFDA to SIFT descriptors for feature extraction matrix, and uses the new descriptors to conduct the feature matching, finally chooses RANSAC to deal with the matches for further purification. The experiments show that the presented algorithm is robust to image changes in scale, illumination, perspective, expression and tiny pose with higher matching accuracy.
Change detection based on integration of multi-scale mixed-resolution information
Author(s):
Li Wei;
Cheng Wang;
Chenglu Wen
Show Abstract
In this paper, a new method of unsupervised change detection is proposed by modeling multi-scale change detector based on local mixed information and we present a method of automated threshold. A theoretical analysis is presented to demonstrate that more comprehensive information is taken into account by the integration of multi-scale information. The ROC curves show that change detector based on multi-scale mixed information(MSM) is more effective than based on mixed information(MIX). Experiments on artificial and real-world datasets indicate that the multi-scale change detection of mixed information can eliminate the pseudo-change part of the area. Therefore, the proposed algorithm MSM is an effective method for the application of change detection.
Automatic extraction of lunar impact craters from Chang’E images based on Hough transform and RANSAC
Author(s):
Zhongfei Luo;
Zhizhong Kang
Show Abstract
This article proposed an algorithm combining Hough transform and RANSAC algorithm for automatic extraction of lunar craters. (1) In order to suppress noise, the images were filtered; (2) The edge of image were extracted, subsequently, eliminate false edge points by qualifying the gradient direction and the area of connected domain; (3) The edge images were segmented through Hough transform, gathering the same crater edge points together; (4) The edge images after segmentation were fitted using RANSAC algorithm, getting the high precision parameter. High precision of the algorithm was verified by the experiments of images acquired by the Chang’E-1 satellites.
Accumulating pyramid spatial-spectral collaborative coding divergence for hyperspectral anomaly detection
Author(s):
Hao Sun;
Huanxin Zou;
Shilin Zhou
Show Abstract
Detection of anomalous targets of various sizes in hyperspectral data has received a lot of attention in reconnaissance and surveillance applications. Many anomaly detectors have been proposed in literature. However, current methods are susceptible to anomalies in the processing window range and often make critical assumptions about the distribution of the background data. Motivated by the fact that anomaly pixels are often distinctive from their local background, in this letter, we proposed a novel hyperspectral anomaly detection framework for real-time remote sensing applications. The proposed framework consists of four major components, sparse feature learning, pyramid grid window selection, joint spatial-spectral collaborative coding and multi-level divergence fusion. It exploits the collaborative representation difference in the feature space to locate potential anomalies and is totally unsupervised without any prior assumptions. Experimental results on airborne recorded hyperspectral data demonstrate that the proposed methods adaptive to anomalies in a large range of sizes and is well suited for parallel processing.
Space resection model calculation based on Random Sample Consensus algorithm
Author(s):
Xinzhu Liu;
Zhizhong Kang
Show Abstract
Resection has been one of the most important content in photogrammetry. It aims at the position and attitude information of camera at the shooting point. However in some cases, the observed values for calculating are with gross errors. This paper presents a robust algorithm that using RANSAC method with DLT model can effectually avoiding the difficulties to determine initial values when using co-linear equation. The results also show that our strategies can exclude crude handicap and lead to an accurate and efficient way to gain elements of exterior orientation.
A robust endmember constrained non-negative matrix factorization method for hyperspectral unmixing
Author(s):
Jinjun Liu
Show Abstract
This paper presents a new method based non-negative matrix factorization (NMF) for hyperspectral unmixing, termed robust endmember constrained NMF (RECNMF). The objective function of RECNMF can not only reduce the effect of noise and outliers but also can reduce the size of convex formed by the endmembers and the correlation between the endmembers. The algorithm is solved by the projected gradient method. The effectiveness of RECNMF is illustrated by comparing its performance with the state-of-the-art algorithms in simulated data.
Vehicle detection from high-resolution aerial images based on superpixel and color name features
Author(s):
Ziyi Chen;
Liujuan Cao;
Zang Yu;
Yiping Chen;
Cheng Wang;
Jonathan Li
Show Abstract
Automatic vehicle detection from aerial images is emerging due to the strong demand of large-area traffic monitoring. In this paper, we present a novel framework for automatic vehicle detection from the aerial images. Through superpixel segmentation, we first segment the aerial images into homogeneous patches, which consist of the basic units during the detection to improve efficiency. By introducing the sparse representation into our method, powerful classification ability is achieved after the dictionary training. To effectively describe a patch, the Histogram of Oriented Gradient (HOG) is used. We further propose to integrate color information to enrich the feature representation by using the color name feature. The final feature consists of both HOG and color name based histogram, by which we get a strong descriptor of a patch. Experimental results demonstrate the effectiveness and robust performance of the proposed algorithm for vehicle detection from aerial images.
The live service of video geo-information
Author(s):
Wu Xue;
Yongsheng Zhang;
Ying Yu;
Ling Zhao
Show Abstract
In disaster rescue, emergency response and other occasions, traditional aerial photogrammetry is difficult to meet real-time monitoring and dynamic tracking demands. To achieve the live service of video geo-information, a system is designed and realized—an unmanned helicopter equipped with video sensor, POS, and high-band radio. This paper briefly introduced the concept and design of the system. The workflow of video geo-information live service is listed. Related experiments and some products are shown. In the end, the conclusion and outlook is given.
The infrared target enhancement method based on optimization at the whole directional polarization
Author(s):
Yan Zhang;
Ji-Cheng Li;
Sha-fei Wang;
Ting Gong
Show Abstract
An infrared target enhancement method based on optimization in the whole directional polarization is studied in this paper. By using the description relationship between the stokes vector of incident light and the intensity of emergent light, the analytical formula between the intensity of emergent light and the polarizing angle is deduced, and thus virtually derives the intensity of emergent light from 0°to 360° polarizing angle. Then according to the criterion of maximum contrast between target and background, the searching of optimal polarizing angle is iteratively realized, and finally gets the enhanced infrared target image. The feasibility and validity of the algorithm are validated by using real long wave infrared (LWIR) polarization images of target. Experimental results show that, the enhanced image using proposed algorithm possesses obvious suppression effect of background clutter, and the quantitative evaluation under two kinds of image quality evaluation indexes of average gradient and image entropy also validates the effectiveness of our algorithm in infrared target enhancement.
Automated extraction of urban trees from mobile LiDAR point clouds
Author(s):
Fan W.;
Chenglu W.;
Jonathan L.
Show Abstract
This paper presents an automatic algorithm to localize and extract urban trees from mobile LiDAR point clouds. First, in order to reduce the number of points to be processed, the ground points are filtered out from the raw point clouds, and the un-ground points are segmented into supervoxels. Then, a novel localization method is proposed to locate the urban trees accurately. Next, a segmentation method by localization is proposed to achieve objects. Finally, the features of objects are extracted, and the feature vectors are classified by random forests trained on manually labeled objects. The proposed method has been tested on a point cloud dataset. The results prove that our algorithm efficiently extracts the urban trees.
Ship detection in panchromatic images: a new method and its DSP implementation
Author(s):
Yuan Yao;
Zhiguo Jiang;
Haopeng Zhang;
Mengfei Wang;
Gang Meng
Show Abstract
In this paper, a new ship detection method is proposed after analyzing the characteristics of panchromatic remote sensing images and ship targets. Firstly, AdaBoost(Adaptive Boosting) classifiers trained by Haar features are utilized to make coarse detection of ship targets. Then LSD (Line Segment Detector) is adopted to extract the line features in target slices to make fine detection. Experimental results on a dataset of panchromatic remote sensing images with a spatial resolution of 2m show that the proposed algorithm can achieve high detection rate and low false alarm rate. Meanwhile, the algorithm can meet the needs of practical applications on DSP (Digital Signal Processor).
Automatic target extraction in complicated background for camera calibration
Author(s):
Xichao Guo;
Cheng Wang;
Chenglu Wen;
Ming Cheng
Show Abstract
In order to perform high precise calibration of camera in complex background, a novel design of planar composite target and the corresponding automatic extraction algorithm are presented. Unlike other commonly used target designs, the proposed target contains the information of feature point coordinate and feature point serial number simultaneously. Then based on the original target, templates are prepared by three geometric transformations and used as the input of template matching based on shape context. Finally, parity check and region growing methods are used to extract the target as final result. The experimental results show that the proposed method for automatic extraction and recognition of the proposed target is effective, accurate and reliable.
Spatial average ambiguity function for array radar with stochastic signals
Author(s):
Guofeng Zha;
Hongqiang Wang;
Yongqiang Cheng;
Yuliang Qin
Show Abstract
For analyzing the spatial resolving performance of multi-transmitter single-receiver (MTSR) array radar with stochastic signals, the spatial average ambiguity function (SAAF) is introduced based on the statistical average theory. The analytic expression of SAAF and the corresponding resolutions in vertical range and in horizontal range are derived. Since spatial resolving performance is impacted by many parameters including signal modulation schemes, signal bandwidth, array aperture’s size and target’s spatial position, comparisons are implemented to analyze these influences. Simulation results are presented to validate the whole analysis.
SAR target classification based on multiscale sparse representation
Author(s):
Huaiyu Ruan;
Rong Zhang;
Jingge Li;
Yibing Zhan
Show Abstract
We propose a novel multiscale sparse representation approach for SAR target classification. It firstly extracts the dense SIFT descriptors on multiple scales, then trains a global multiscale dictionary by sparse coding algorithm. After obtaining the sparse representation, the method applies spatial pyramid matching (SPM) and max pooling to summarize the features for each image. The proposed method can provide more information and descriptive ability than single-scale ones. Moreover, it costs less extra computation than existing multiscale methods which compute a dictionary for each scale. The MSTAR database and ship database collected from TerraSAR-X images are used in classification setup. Results show that the best overall classification rate of the proposed approach can achieve 98.83% on the MSTAR database and 92.67% on the TerraSAR-X ship database.
Benchmark on outdoor scenes
Author(s):
Hairong Zhang;
Cheng Wang;
Yiping Chen;
Fukai Jia;
Jonathan Li
Show Abstract
Depth super-resolution is becoming popular in computer vision, and most of test data is based on indoor data sets with ground-truth measurements such as Middlebury. However, indoor data sets mainly are acquired from structured light techniques under ideal conditions, which cannot represent the objective world with nature light. Unlike indoor scenes, the uncontrolled outdoor environment is much more complicated and is rich both in visual and depth texture. For that reason, we develop a more challenging and meaningful outdoor benchmark for depth super-resolution using the state-of-the-art active laser scanning system.
Epipolar geometry comparison of SAR and optical camera
Author(s):
Dong Li;
Yunhua Zhang
Show Abstract
In computer vision, optical camera is often used as the eyes of computer. If we replace camera with synthetic aperture radar (SAR), we will then enter a microwave vision of the world. This paper gives a comparison of SAR imaging and camera imaging from the viewpoint of epipolar geometry. The imaging model and epipolar geometry of the two sensors are analyzed in detail. Their difference is illustrated, and their unification is particularly demonstrated. We hope these may benefit researchers in field of computer vision or SAR image processing to construct a computer SAR vision, which is dedicated to compensate and improve human vision by electromagnetically perceiving and understanding the images.
A line segment based registration method for Terrestrial Laser Scanning point cloud data
Author(s):
Jun Cheng;
Ming Cheng;
Yangbin Lin;
Cheng Wang
Show Abstract
This paper proposed a 3d line segment based registration method for terrestrial laser scanning (TLS) data. The 3D line segment is adopted to describe the point cloud data and reduce geometric complexity. After that, we introduce a framework for registration. We demonstrate the accuracy of our method for rigid transformations in the presence of terrestrial laser scanning point cloud.
An automatic and overlap based method for LiDAR intensity correction
Author(s):
Qiong Ding
Show Abstract
LiDAR provides intensity data that reflect the material characteristics of objects. However, intensity values need to be corrected before they can be reliably used for applications because of the error during data acquisition. This study proposed an automatic and overlap based method for intensity correction. Firstly, a radar equation based method was employed for removal of main errors. Then, nearest neighbor algorithm was used to find out homologous points of overlap regions and assumption was made that homologous points should have same intensity. Finally, an improved model was utilized to eliminate overlap discrepancies. This method can be considered as a potential aid to enhance the accuracy of LiDAR intensity data and improve the automation of data process.
Detecting traffic hot spots using vehicle tracking data
Author(s):
Zhimin Xu;
Zhiyong Lin;
Cheng Zhou;
Changqing Huang
Show Abstract
Vehicle tracking data for thousands of urban vehicles and the availability of digital map provide urban planners unprecedented opportunities for better understanding urban transportation. In this paper, we aim to detect traffic hot spots on urban road networks using vehicle tracking data. Our approach first proposes an integrated map-matching algorithm based on the road buffer and vehicle driving direction, to find out which road segment the vehicle is travelling on. Then, we estimate travel speed by calculating the average the speed of every vehicle on a certain road segment, which indicates traffic status, and create the spatial weights matrices based on the connectivity of road segments, which expresses the spatial dependence between each road segment. Finally, the measure of global and local spatial autocorrelation is used to evaluate the spatial distribution of the traffic condition and reveal the traffic hot spots on the road networks. Experiments based on the taxi tracking data and urban road network data from Wuhan have been performed to validate the detection effectiveness.
Combinative hypergraph learning on oil spill training dataset
Author(s):
Binghui Wei;
Ming Cheng;
Cheng Wang;
Jonathan Li
Show Abstract
Detecting oil spill from open sea based on Synthetic Aperture Radar (SAR) image is a very important work. One of key issues is to distinguish oil spill from “look-alike”. There are many existing methods to tackle this issue including supervised and semi-supervised learning. Recent years have witnessed a surge of interest in hypergraph-based transductive classification. This paper proposes combinative hypergraph learning (CHL) to distinguish oil spill from “look-alike”. CHL captures the similarity between two samples in the same category by adding sparse hypergraph learning to conventional hypergraph learning. Experimental results have demonstrated the effectiveness of CHL in comparison to the state-of-the-art methods and showed that our proposed method is promising.
Application of LiDAR’s multiple attributes for wetland classification
Author(s):
Qiong Ding;
Shengyue Ji;
Wu Chen
Show Abstract
Wetlands have received intensive interdisciplinary attention as a unique ecosystem and valuable resources. As a new technology, the airborne LiDAR system has been applied in wetland research these years. However, most of the studies used only one or two LiDAR observations to extract either terrain or vegetation in wetlands. This research aims at integrating LiDAR’s multiple attributes (DSM, DTM, off-ground features, Slop map, multiple pulse returns, and normalized intensity) to improve mapping and classification of wetlands based on a multi-level object-oriented classification method. By using this method, we are able to classify the Yellow River Delta wetland into eight classes with overall classification accuracy of 92.5%
A sea-land segmentation algorithm based on graph theory
Author(s):
Zewen Huang
Show Abstract
As the key technology for sea-based target detection and recognition, sea-land segmentation directly effects the execution efficiency of target detection algorithm. Therefore we proposed an innovative sea-land segmentation method in this paper. Aiming at the problem of over-segmentation, we improved the graph-based image segmentation model based on difference in texture between sea and land. Firstly, by introducing the conception of average weight, we achieved the extraction of texture feature. Secondly, we redefined the difference between two components. Experiments results show the proposed method outperforms traditional sea-land segmentation approaches.
Thinking on sharing 3D GIS data by web service following OGC standard
Author(s):
Lvewei Wu
Show Abstract
Sharing based on web service is a well-understood method in GIS. This paper mainly study on how to share 3D GIS data in this way. Based on the analysis of current situation, this paper concludes the existing problem and put forward the idea of sharing 3D by web service following OGC standard. Besides, feasibility and practicality of the idea are verified by relevant experimental evidence. In the end, a framework conforming to the idea above is present, providing reference to the sharing 3D GIS based on web service.
A simplified method to estimate atmospheric water vapor using MODIS near-infrared data
Author(s):
Xinming Wang;
Xiaoping Gu;
Zhanping Wu
Show Abstract
Atmospheric water vapor plays a significant role in the study of climate change and hydrological cycle processes. In order to acquire the accurate distribution of atmospheric water vapor which is varying with time, location, and altitude, it is necessary to monitor it at high spatial and temporal resolution. Unfortunately, it is difficult to map the spatial distribution of atmospheric water vapor due to the lack of meteorological instrumentation at adequate spatial and temporal observation scales. This paper introduces a simplified method to retrieve Precipitable Water Vapor (PWV) using the ratio of the apparent reflectance values of the 18th and 19th band of Moderate Resolution Imaging Spectroradiometer (MODIS). Compared to the EOS PWV products of the same time and area, the PWV estimated using this simplified method is closer to the radiosonde results which is considered as the true PWV value. Results reveal that this simplified method is applicable over cloud-free atmospheric conditions of the mid-latitude regions.
Economic load dispatch using improved gravitational search algorithm
Author(s):
Yu Huang;
Jia-rong Wang;
Feng Guo
Show Abstract
This paper presents an improved gravitational search algorithm(IGSA) to solve the economic load dispatch(ELD) problem. In order to avoid the local optimum phenomenon, mutation processing is applied to the GSA. The IGSA is applied to solve the economic load dispatch problems with the valve point effects, which has 13 generators and a load demand of 2520 MW. Calculation results show that the algorithm in this paper can deal with the ELD problems with high stability.
A RANSAC-ST method for image matching
Author(s):
Fengman Jia;
Zhizhong Kang
Show Abstract
Facing challenges of external environmental noise, it is necessary to find a robust, accurate and fast image-matching method. This paper proposed a method combining SIFT (Scale Invariant Feature Transform) algorithm and RANSACST (RANdom Sampling Consensus with Statistical Testing). RANSAC-ST algorithm is the improvement of RANSAC, which uses a strategy for best model determination in terms of the statistical characteristics of a deterministic mathematical model for hypothesis testing. It will generate a statistical histogram of all hypothesis fundamental matrices, and then the fundamental matrix whose convergence degree reaches the threshold is regarded as the best model. Experimental results show that with the proposed algorithm, the robustness and computation efficiency of correspondence matching can be effectively improved.
Improved phase congruency based interest point detection for multispectral remote sensing images
Author(s):
Min Chen;
Qing Zhu;
Jun Zhu;
Zhu Xu;
Duoxiang Cheng
Show Abstract
One of the biggest challenges in multispectral image interest point detection is the variation of radiation. Many methods have been proposed to address this problem. However, the detection performance is still unstable. In this paper, a robust point detector is proposed. Firstly, image illumination space is constructed by using a parameters adaptive method. Secondly, a phase congruency based interest point detection algorithm is adopted to compute candidate points in illumination space. Then, all interest point candidates are mapped back to the original image and a non-maximum suppression step is added to find final interest points. Finally, the feature scale values of all interest points are calculated based on the Laplacian function. The experimental results show that the proposed method performs better than other traditional methods in feature repeatability rate and repeated features number for multispectral images.
A class-oriented model for hyperspectral image classification through hierarchy-tree-based selection
Author(s):
Zhongqi Tang;
Guangyuan Fu;
XiaoLin Zhao;
Jin Chen;
Li Zhang
Show Abstract
With the development of hyperspectral sensors over the last few decades, hyperspectral images (HSIs) face new challenges in the field of data analysis. Due to those high-dimensional data, the most challenging issue is to select an effective yet minimal subset from a mass of bands. This paper proposes a class-oriented model to solve the task of classification by incorporating spectral prior of the target, since different targets have different characteristics in spectral correlation. This model operates feature selection after a partition of hyperspectral data into groups along the spectral dimension. In the process of spectral partition, we group the raw data into several subsets by a hierarchy tree structure. In each group, band selection is performed via a recursive support vector machine (R-SVM) learning, which reduces the computational cost as well as preserves the accuracy of classification. To ensure the robustness of the result, we also present a weight-voting strategy for result merging, in which the spectral independency and the classification effectivity are both considered. Extensive experiments show that our model achieves better performance than the existing methods in task-dependent classifications, such as target detection and identification.
Are ground-level visual attributes useful for high resolution remote sensing image classification?
Author(s):
Shuai Liu;
Hao Sun;
Shilin Zhou
Show Abstract
This paper extends the ground-level visual attributes to high resolution remote sensing imagery to demonstrate the useful-ness of visual attributes for remote sensing tasks such as image classification. Visual attributes have been introduced as the semantic properties that transcend the categories. We train predictors from the largest ground-level attributes datasets, SUN, for 102 visual attributes, which is well defined in SUN. We first form an attribute-based representation for the remote sensing imagery with the output of trained attribute predictors. We then evaluate the classification performances of the attribute-based representation against traditional features. Extensive experiments on the ground-level baseline dataset scene 15 and remote sensing dataset UCMLU shows that ground-level visual attributes outperform the traditional low-level features in the classification problem, and the combination of ground-level visual attribute and low-level features obtains best classification rate. Moreover, we demonstrate that attribute-based representation is much more semantically powerful than the low-level features.
Supervised nonparametric sparse discriminant analysis for hyperspectral imagery classification
Author(s):
Longfei Wu;
Hao Sun;
Kefeng Ji
Show Abstract
Owing to the high spectral sampling, the spectral information in hyperspectral imagery (HSI) is often highly correlated and contains redundancy. Motivated by the recent success of sparsity preserving based dimensionality reduction (DR) techniques in both computer vision and remote sensing image analysis community, a novel supervised nonparametric sparse discriminant analysis (NSDA) algorithm is presented for HSI classification. The objective function of NSDA aims at preserving the within-class sparse reconstructive relationship for within-class compactness characterization and maximizing the nonparametric between-class scatter simultaneously to enhance discriminative ability of the features in the projected space. Essentially, it seeks for the optimal projection matrix to identify the underlying discriminative manifold structure of a multiclass dataset. Experimental results on one visualization dataset and three recorded HSI dataset demonstrate NSDA outperforms several state-of-the-art feature extraction methods for HSI classification.
A new 3D shape precision measurement system calibration method based on non-diffraction grating structured light projection
Author(s):
Ya Zhu;
Liping Zhou;
Wenlong Li;
Jianghong Gan;
Long Xu
Show Abstract
Phase calculation-based fringe projection techniques are widely used in three-dimensional shape measurement fields to obtain the 3D shape data of the object’s surface. One important step of the phase calculation is calibration, which determines the relationship between the image phase and depth information. The traditional calibration methods are too complex and require many parameters. In this paper, model of 3D shape precision calibration method based on non-diffraction grating structured light fringes projection is proposed, which is consist of camera model, fringe phase obtaining, height-phase relationship model. This method is simple, convenient and there is no approximation in it, which can satisfy the precision measurement.
An approach of DSM generation from multi-view images acquired by UAVs
Author(s):
DaTian Hu;
Mingyao Ai;
Qingwu Hu;
Jiayuan Li
Show Abstract
In the past few years, for its lower-cost, safer and high-resolution images, unmanned aerial vehicles (UAVs) demonstrated great potential for photogrammetric measurements in numerous application fields. Nevertheless, these images are often affected by large rotation, big viewpoint change as well as small overlaps, in which case traditional procedure are not able to orientate images or generate reliable Digital Generation Models (DSM). This paper introduces the whole procedure of the DSM generation, which comprehensively utilizes advantage of both computer vision and multi-image matching algorithms in extracting points and generating a dense DSM. Experiment shows that, based on this procedure, it can quickly extract points from the high-resolution images acquired by UAVs with high location accuracy.
3D scene reconstruction based on 3D laser point cloud combining UAV images
Author(s):
Huiyun Liu;
Yangyang Yan;
Xitong Zhang;
Zhenzhen Wu
Show Abstract
It is a big challenge capturing and modeling 3D information of the built environment. A number of techniques and technologies are now in use. These include GPS, and photogrammetric application and also remote sensing applications. The experiment uses multi-source data fusion technology for 3D scene reconstruction based on the principle of 3D laser scanning technology, which uses the laser point cloud data as the basis and Digital Ortho-photo Map as an auxiliary, uses 3DsMAX software as a basic tool for building three-dimensional scene reconstruction. The article includes data acquisition, data preprocessing, 3D scene construction. The results show that the 3D scene has better truthfulness, and the accuracy of the scene meet the need of 3D scene construction.
Research on precise modeling of buildings based on multi-source data fusion of air to ground
Author(s):
Yongqiang Li;
Lubiao Niu;
Shasha Yang;
Lixue Li;
Xitong Zhang
Show Abstract
Aims at the accuracy problem of precise modeling of buildings, a test research was conducted based on multi-source data for buildings of the same test area , including top data of air-borne LiDAR, aerial orthophotos, and façade data of vehicle-borne LiDAR. After accurately extracted the top and bottom outlines of building clusters, a series of qualitative and quantitative analysis was carried out for the 2D interval between outlines. Research results provide a reliable accuracy support for precise modeling of buildings of air ground multi-source data fusion, on the same time, discussed some solution for key technical problems.
Reconstruction of individual trees based on LiDAR and in situ data
Author(s):
Jianbo Qi;
Donghui Xie;
Wuming Zhang
Show Abstract
This paper presents a method to reconstruct individual trees from Terrestrial Laser Scanning (TLS) data obtained in leafoff conditions of an experiment plot. It firstly used the point clouds to build the branch structures of trees with a global optimization method. Computer generated needles and shoots were added to the previously constructed branches according to the leaf area (LA) of each individual tree, in consideration of clumping effect of small-scale structures. The LA was determined by the proportion of crown volume in this plot with LAI measured. In this way, several larix trees with different shapes and heights were reconstructed, which is basis of 3D forest scene reconstruction.