Show all abstracts
View Session
- Front Matter: Volume 7000
- Holography-Interferometry
- Industrial Applications
- Wavelet Methods
- Compression Technologies
- Advanced Camera Systems
- Wavelet Applications and Denoising
- Watermarking
- Advanced Imaging
- Medical Imaging
- Interaction between Optics and Digital Image Processing
- Poster Session
Front Matter: Volume 7000
Front Matter: Volume 7000
Show abstract
This PDF file contains the front matter associated with SPIE Proceedings Volume 7000, including the Title Page, Copyright information, Table of Contents, Introduction, and the Conference Committee listing.
Holography-Interferometry
Superresolution microscopy using common-path phase-shifting interferometry
Show abstract
In this contribution, we present a new approach aimed to achieve superresolution in digital holographic
microscopy that overcomes the limitations imposed by the Abbe's diffraction limit. Based on time
multiplexing and off-axis illumination, a common-path interferometric architecture allows the synthesis of an
enlarged aperture that expands the cutoff frequency of the conventional imaging system. Complex object
amplitude distribution is recovered using an extremely simple phase-shifting configuration instead of image
plane off-axis holographic recording. As a consequence, the setup becomes easy-to-configure (less
requirements and lower number of optical elements) and it is useful for practical implementation in
microscopes while only simple modifications are required (no higher magnifications are needed to resolve the
interference pattern at the CCD plane). Experimental results validating the capabilities of the presented
approach when a low numerical aperture commercial microscope objective is used as imaging lens are
included using both a synthetic object (high resolution USAF test target) and a biosample (red blood cells) as
objects under test.
Segmentation of macroscopic object digital holographic reconstructions using extracted depth information
Show abstract
The nature of digital hologram's allows for the implementation of segmentation process' using volumes of reconstructions
as their input, where each reconstruction in the volume is a reconstruction at a different focal plane.
Our segmentation technique utilizes extracted focus and shape information. In the case of digital hologram's
encoding macroscopic objects, this information is generally obtained using data extraction algorithms applied to
a volume of reconstructions. We have developed a three stage segmentation algorithm for macroscopic objects
encoded in digital hologram's. This algorithm uses a depth-from-focus technique applied to a set of numerical
reconstructions from a single perspective of the digital hologram to extract focus and shape information in the
form of a depth map and a maximum focus map. First we estimate the degree of focus at each coordinate. We
then calculate a depth map of the scene and segment all object coordinates from background coordinates in the
second stage. Finally in the third stage, we perform the segmentation of a digital hologram's reconstruction into
independent objects or object regions. A segmentation image is created through applying histogram and image
processing algorithms to the depth map. We present results of applying our technique to digital hologram's
containing single and multiple macroscopic objects.
Holographic data storage using phase-only data pages
Show abstract
The application of phase-only input data pages has several advantages with respect to conventional amplitude
modulated holographic storage: It avoids the saturation of the storage material by providing a smooth Fourier
plane, improves the response in associative read-out, increases the light efficiency of the recording object wave
and provides the opportunity of data encryption. However, if the information is carried by the phase of object
wave front recovery of the data from the reconstructed beam is problematic with simple intensity sensitive
devices as a CCD camera. To solve this problem we propose a compact phase to amplitude data page conversion
method and apply it to the output of a Fourier holographic data storage system. The phase to amplitude
conversion uses a birefringent crystal to generate two equal intensity copies of the reconstructed data page that
are geometrically shifted by an integer number of pixels with respect to each other each other. The interference
of these two phase modulated images is projected on the detector field of the camera. The interference pattern
contains low and high intensity pixels if the phases of the interfering pixels are opposite and identical
respectively. Using proper data coding, the original data matrix recovered from the intensity pattern of the CCD.
Fourier plane homogeneity, bit error rate and positioning tolerances of the proposed holographic storage method
are investigated by computer modeling and a comparison is provided with amplitude modulated data pages.
Quantitative phase restoration in differential interference contrast (DIC) microscopy
Show abstract
Phase contrast imaging is a specific technique in optical microscopy that is able to capture the minute structures of
unlabeled biological sample from contrast generated in the variations of the object's refractive index. It is especially
suitable for living cells and organisms that are hardly visible under conventional light microscopy as they barely alter the
intensity and only introduce phase shifts in transmitted light. Optical phase imaging provides sensitivity to measure
optical path length (OPL) differences down to nanometers, which has great potential in biomedical applications from
examining both topological and three-dimensional biophysical properties of cells and organisms.
Conventional DIC microscopy with partially coherent light source is a very powerful technique for phase imaging, and is
able to yield higher lateral resolution compared to other interferometric phase imaging methods. However, it is
inherently qualitative and the information obtained is a phase-gradient image rather than a true linear mapping of OPL
differences. This hinders its further application as it is difficult to infer the results directly. Work has been done
previously to obtain the quantitative phase information out of DIC. However, some of these methods not only involves
costly hardware modification but also complicated computation. Here we investigate another approach that combines the
correlation of light intensity and phase with polarization-modulated differential interference contrast (DIC) microscopy.
The required hardware modification is simple, and numerically solving the relationship of light propagation in a series of
through-focus DIC images allows phase information to be restored from phase gradients in two-dimensional planes.
Quantitative phase information in three-dimensional space can then be reconstructed from 3D rendering of the calculated
phase images. Initial results of application in biological cells are also demonstrated.
Industrial Applications
New solutions for industrial inspection based on 3D computer tomography
Show abstract
In recent years the requirements of industrial applications relating to image processing have significantly increased.
According to fast and modern production processes and optimized manufacturing of high quality products, new ways of
image acquisition and analysis are needed. Here the industrial computer tomography (CT) as a non-destructive
technology for 3D data generation meets this challenge by offering the possibility of complete inspection of complex
industrial parts with all outer and inner geometric features. Consequently CT technology is well suited for different kinds
of industrial image-based applications in the field of quality assurance like material testing or first article inspection.
Moreover surface reconstruction and reverse engineering applications will benefit from CT. In this paper our new
methods for efficient 3D CT-image processing are presented. This includes improved solutions for 3D surface
reconstruction, innovative approaches of CAD-based segmentation in the CT volume data and the automatic geometric
feature detection in complex parts. However the aspect of accuracy is essential in the field of metrology. In order to
enhance precision the CT sensor can be combined with other, more accurate sensor systems generating measure points
for CT data correction. All algorithms are applied to real data sets in order to demonstrate our tools.
Automated inspection of microlens arrays
Show abstract
Industrial inspection of micro-devices is often a very challenging task, especially when those devices are produced
in large quantities using micro-fabrication techniques. In the case of microlenses, millions of lenses are produced
on the same substrate, thus forming a dense array. In this article, we investigate a possible automation of
the microlens array inspection process. First, two image processing methods are considered and compared:
reference subtraction and blob analysis. The criteria chosen to compare them are the reliability of the defect
detection, the processing time required, as well as the sensitivity to image acquisition conditions, such as varying
illumination and focus. Tests performed on a real-world database of microlens array images led to select the blob
analysis method. Based on the selected method, an automated inspection software module was then successfully
implemented. Its good performance allows to dramatically reduce the inspection time as well as the human
intervention in the inspection process.
Flaw detection and segmentation in textile inspection
Show abstract
We present a new method to automatically segment local defects in a woven fabric that does not require any additional
defect-free reference for comparison. Firstly, the structural features of the repetition pattern of the minimal weave repeat
are extracted from the Fourier spectrum of the sample under inspection. The corresponding peaks are automatically
identified and removed from the fabric frequency spectrum. Secondly, we define a set of multi-scale oriented bandpass
filters, adapted to the specific structure of the sample, that operate in the Fourier domain. The filter design is the key part
of the method. Using the set of filters, local defects can be extracted. Thirdly, the filtered images obtained at different
scales are inverse Fourier transformed, binarized and merged to obtain an output image where flaws are segmented from
the fabric background. The method can be applied to fabrics of uniform color as well as to fabrics woven with threads of
different colors. It is Euclidean motion invariant and texture adaptive and it is useful for automatic inspection both online
and off-line. The whole process is fully automatic and can be implemented either optical or electronically. A variety
of experimental results are presented and discussed.
Wavelet Methods
The wavelet transform on the two-sphere and related manifolds: a review
Show abstract
In a first part, we discuss several properties that seem desirable for any type of wavelet, such as smoothness,
orthogonality, local support, Riesz stability, or vanishing moments. Then we review the construction of the
spherical continuous wavelet transform based on the stereographic projection. Next we turn to the discrete
wavelet transform. We review the various existing constructions and compare them in the light of the requirements
listed above. Finally, we briefly describe the continuous wavelet transform on a two-sheeted hyperboloid
and give some hints concerning the case of a general conic section.
A Hilbert-space approach to the complete investigation of Vaidyanathan's procedure applied to the design of unitary filterbanks for the generation of orthogonal wavelet bases
Show abstract
The decomposition of signals using the classical Daubechies wavelets can equivalently be described as a decomposition
using a two channel filterbank with the scaling function corresponding to the low pass and the mother
wavelet to the high pass channel.
This classical two channel approach was extended to also comprise filterbanks with more than two channels,
corresponding to one scaling function and two or more mother wavelets.1 Although these newly found
wavelets fitted in very well with the theory of the existing Daubechies wavelets, the frequency selectivity of the
corresponding filterbank was in general not satisfying.
We presented a new method2 to improve the selectivity of the corresponding filterbank by introducing additional
elementary building blocks in the design process, leaving the low pass filter and the polynomial degree of
the impulse responses of the other filters of the filterbank untouched.
In this paper we take the new method one step further and not only show how the improvement could be
achieved but also what possibilities there are to modify the design and what patterns they adhere to. Furthermore
we introduce a figure of merit of the total filterbank and its individual filters and show the improvements to the design process.
Wavelet-constrained stereo matching under photometric variations
Show abstract
We propose a new method to address the problem of stereo matching under varying illumination conditions.
First, a spatially varying multiplicative model is developed to account for photometric changes induced between
both images in the stereo pair. The stereo matching problem based on this model is then formulated as a
constrained optimization problem in which an appropriate convex objective function is minimized under convex
constraints. These constraints arise from prior knowledge and rely on various properties of both disparity and
illumination fields. In order to obtain a smooth disparity field while preserving discontinuities around object
edges, we consider an appropriate wavelet-based regularization constraint. The resulting multi-constrained
optimization problem is solved via an efficient block iterative algorithm which offers great flexibility in the
incorporation of several constraints. Experimental results demonstrate the efficiency of the proposed method
to recover illumination changes and disparity map simultaneously, making stereo matching very robust w.r.t.
such changes.
Compression Technologies
Sparse spike coding : applications of neuroscience to the processing of natural images
Show abstract
If modern computers are sometimes superior to cognition in some specialized tasks such as playing chess or
browsing a large database, they can't beat the efficiency of biological vision for such simple tasks as recognizing
a relative or following an object in a complex background. We present in this paper our attempt at outlining the
dynamical, parallel and event-based representation for vision in the architecture of the central nervous system.
We will illustrate this by showing that in a signal matching framework, a L/LN (linear/non-linear) cascade
may efficiently transform a sensory signal into a neural spiking signal and we apply this framework to a model
retina. However, this code gets redundant when using an over-complete basis as is necessary for modeling the
primary visual cortex: we therefore optimize the efficiency cost by increasing the sparseness of the code. This
is implemented by propagating and canceling redundant information using lateral interactions. We compare the
eciency of this representation in terms of compression as the reconstruction quality as a function of the coding
length. This will correspond to a modification of the Matching Pursuit algorithm where the ArgMax function is
optimized for competition, or Competition Optimized Matching Pursuit (COMP). We will particularly focus on
bridging neuroscience and image processing and on the advantages of such an interdisciplinary approach.
Heuristic dynamic complexity coding
Show abstract
Distributed video coding is a new video coding paradigm that shifts the computational intensive motion estimation
from encoder to decoder. This results in a lightweight encoder and a complex decoder, as opposed
to the predictive video coding scheme (e.g., MPEG-X and H.26X) with a complex encoder and a lightweight
decoder. Both schemas, however, do not have the ability to adapt to varying complexity constraints imposed by
encoder and decoder, which is an essential ability for applications targeting a wide range of devices with different
complexity constraints or applications with temporary variable complexity constraints. Moreover, the effect of
complexity adaptation on the overall compression performance is of great importance and has not yet been investigated.
To address this need, we have developed a video coding system with the possibility to adapt itself to
complexity constraints by dynamically sharing the motion estimation computations between both components.
On this system we have studied the effect of the complexity distribution on the compression performance.
This paper describes how motion estimation can be shared using heuristic dynamic complexity and how
distribution of complexity affects the overall compression performance of the system. The results show that the
complexity can indeed be shared between encoder and decoder in an efficient way at acceptable rate-distortion performance.
Priority-based error protection for the scalable extension of H.264/SVC
Show abstract
Using the scalable video coding (SVC) extension of the H.264/AVC video coding standard, encoding a video sequence
yields a quality, resolution, and frame-rate scalable bit-stream. This means that a version of the video sequence with a
lower resolution, quality and/or frame rate can be obtained by extracting selected parts of the scalable bit-stream, without
the need for re-encoding. In this way, easy adaptation of the video material to the end-users' device (computational
power and display resolution) and channel characteristics is possible. In this paper, the use of unequal error protection
(UEP) for error-resilient transmission of H.264/SVC-encoded video sequences is studied. By using unequal error
protection, graceful degradation of the video quality is achieved when the targeted packet loss probability is exceeded. In
contrast, using equal error protection (EEP), an immediate and dramatic drop in quality is observed under these
conditions.
Compression of confocal microscopy images: a comparative study
Show abstract
In the present communication an extensive study of lossless and lossy compression of laser confocal microscopy images
is conducted. For the lossy (irreversible) case, both objective and subjective results are given. It is shown that the
JPEG2000 compression standard outperforms all others, even the recently introduced HD Photo, in both the reversible
and irreversible cases. More specifically, the JPEG2000 lossless performance is approximately 15% better than the next
best image data compression algorithm tested (namely the PNG) and the JPEG2000 lossy compression objective
performance results are at least 3dB better than that of HD Photo or JPEG.
Image quality assessment through a logarithmic anisotropic measure
Show abstract
Typically, in many situations, image quality assessment requires a reference or "ground truth" image to provide a
quantitative measure. Popular quality measures such as the Peak Signal Noise Ratio (PSNR) or simply the Root Mean
Squared Error (RMSE) are simple to calculate, but it is well known that they are not always well correlated with the
perceived visual quality. Provided that a reference image is not always available, other blind quality assessment
methods have been proposed to achieve a measure of the image quality assessment. In this paper, a new self-contained
logarithmic measure that not requires the knowledge of a ground-truth image is introduced. This new measure is based
on the use of a particular type of the high-order Rényi entropies. This method is based on measuring the anisotropy of
the image through the variance of the expected value of the
pixel-wise directional image entropy. Thus, a new
logarithmic quality measure (LQM) is applied to a set of test images and compared with PSNR and other recently
proposed quality metrics to reveal advantages and differences with them.
Advanced Camera Systems
Smart camera with embedded co-processor: a postal sorting application
Show abstract
This work describes an image acquisition and processing system based on a new co-processor architecture designed
for CMOS sensor imaging. The platform permits to configure a wide variety of acquisition modes (random
region acquisition, variable image size, multi-exposition image) as well as high-performance image pre-processing
(filtering, de-noising, binarisation, pattern recognition). Furthermore, the acquisition is driven by an FPGA, as
well as a processing stage followed by a Nexperia processor. The data transfer, from the FPGAs board to the
Nexperia processor, can be pipelined to the co-processor to increase achievable throughput performances. The co-processor
architecture has been designed so as to obtain a unit that can be configured on the fly, in terms of type
and number of chained processing (up to 8 successive pre-defined pre-processing), during the image acquisition
process that is dynamically defined by the application. Examples of acquisition and processing performances are
reported and compared to classical image acquisition systems based on standard modular PC platforms. The
experimental results show a considerable increase of the performances. For instance the reading of bar codes
with applications to postal sorting on a PC platform is limited to about 15 images (letters) per second. The new
platform beside resulting more compact and easily installable in hostile environments can successfully analyze
up to 50 images/s.
Panoramic lens applications revisited
Show abstract
During the last few years, innovative optical design strategies to generate and control image mapping have been
successful in producing high-resolution digital imagers and projectors. This new generation of panoramic lenses
includes catadioptric panoramic lenses, panoramic annular lenses, visible/IR fisheye lenses, anamorphic wide-angle
attachments, and visible/IR panomorph lenses. Given that a wide-angle lens images a large field of view on a limited
number of pixels, a systematic pixel-to-angle mapping will help the efficient use of each pixel in the field of view.
In this paper, we present several modern applications of these modern types of hemispheric lenses. Recently,
surveillance and security applications have been proposed and published in Security and Defence symposium.
However, modern hemispheric lens can be used in many other fields. A panoramic imaging sensor contributes most to
the perception of the world. Panoramic lenses are now ready to be deployed in many optical solutions. Covered
applications include, but are not limited to medical imaging (endoscope, rigiscope, fiberscope...), remote sensing (pipe
inspection, crime scene investigation, archeology...), multimedia (hemispheric projector, panoramic image...). Modern
panoramic technologies allow simple and efficient digital image processing and the use of standard image analysis
features (motion estimation, segmentation, object tracking, pattern recognition) in the complete 360° hemispheric area.
Ultra-miniature omni-directional camera for an autonomous flying micro-robot
Show abstract
CSEM presents a highly integrated ultra-miniature camera module with omni-directional view dedicated to autonomous
micro flying devices. Very tight design and integration requirements (related to size, weight, and power consumption)
for the optical, microelectronic and electronic components are fulfilled. The presented ultra-miniature camera platform is
based on two major components: a catadioptric lens system and a dedicated image sensor. The optical system consists of
a hyperbolic mirror and an imaging lens. The vertical field of view is +10° to -35°.The CMOS image sensor provides a
polar pixel field with 128 (horizontal) by 64 (vertical) pixels. Since the number of pixels for each circle is constant, the
unwrapped panoramic image achieves a constant resolution in polar direction for all image regions. The whole camera
module, delivering 40 frames per second, contains optical image preprocessing for effortless re-mapping of the acquired
image into undistorted cylindrical coordinates. The total weight of the complete camera is less than 5 g. The system's
outer dimensions are 14.4 mm in height, with a 11.4 mm x 11.4 mm foot print. Thanks to the innovative PROGLOGTM, a
dynamic range of over 140 dB is achieved.
A CAPD based time-of-flight ranging pixel with wide dynamic range
Show abstract
An in depth view is given of time-of-flight distance sensor and read-out circuit implementations. The sensor
is a current assisted photonic demodulator, using majority currents to accelerate generated minority carriers
towards the detecting junctions, and combines an efficient mixer with a sensitive detector. Single ended and
differential configurations of the structure are illustrated. Measurements show a quantum efficiency of 58%, a
DC demodulation contrast of nearly 100% and a contrast of over 40% maintained up till 60 MHz. An improved
embodiment, having an extra drain tap to tune sensitivity, is also presented. With this structure automatic gain
control and improved dynamic range can be achieved. Measurements demonstrate a controllable sensitivity from
maximal to below 1% by changing a sensor control voltage. Secondly design aspects of the read-out circuit are
discussed. A circuit with improved background light suppression and dynamic range is explained, called variable
transimpedance amplifier. Special low pass filters for in-pixel averaging are presented, having a tuneable cut-off
frequency of 10 Hz to 10 kHz and using only 100 μm2. Circuit simulations show a dynamic range of 87 dB, as
well as a background light tolerance up to 10 μA. Combining the circuit with a drain tap sensor yields a dynamic
range of 127 dB. A standard CMOS 0.35 μm proof of principle prototype pixel, measuring 50 μm x 50 μm and
having a 60% fill factor, is presented. Properties, as well as an example image, of a 32x32 CAPD imager are
included.
Detection of activity pattern changes among elderly with 3D camera technology
Show abstract
This paper provides an overview of the medical scales which are currently in practice at the geriatrics department
of the hospital for assessing independence and mobility of elderly patients. Several shortcomings and issues related
to the scales are identified. It is shown how a 3D camera system could be used for the automatic assessment
of several items of the scales. this automated assessment is overcoming many of the issues with the existing
methods. An analysis of the automatically identified activity features of a typical patient is used to compare the
data derived from our system with data obtained with accelerometer readings.
Wavelet Applications and Denoising
Video modelling in the DWT domain
Show abstract
The main issue this paper addresses is to obtain the information sources characterising the video sequences represented in
DWT domain and to discuss their relevance for practical applications. From the statistical point of view, this means to
establish whether the DWT coefficients can be approximated by random variables and, if so, to compute the
corresponding probability density functions (pdf). The corpus considered in experiments is composed of 10 video
sequences, belonging to different movies, each of them about 25 minutes long, DivX coded at a very low rate.
Fringe pattern analysis using a one-dimensional modified Morlet continuous wavelet transform
Show abstract
This paper proposes the use of a modified Morlet wavelet in order to demodulate fringe patterns in conjunction with the
one-dimensional continuous wavelet transform (1D-CWT). Our investigations demonstrate that the modified Morlet
wavelet produces better results compared to the conventional Morlet wavelet when used in fringe pattern analysis. This
novel technique offers superior performance in analysing fringe patterns from objects that exhibit large height variations.
This new technique has been used in conjunction with the direct maximum ridge extraction algorithm and an
improvement in performance is observed. The algorithm has been tested using both computer-generated and real fringe
patterns; and was found to be suitable for fringe pattern demodulation and robust in operation.
Discrete multiscale wavelet shrinkage and integrodifferential equations
Show abstract
We investigate the relation between discrete wavelet shrinkage and integrodifferential equations in the context of
simplification and denoising of one-dimensional signals. In the continuous setting, strong connections between
these two approaches were discovered in 6 (see references). The key observation is that the wavelet transform can be understood
as derivative operator after the convolution with a smoothing kernel. In this paper, we extend these ideas to the
practically relevant discrete setting with both orthogonal and biorthogonal wavelets. In the discrete case, the
behaviour of the smoothing kernels for different scales requires additional investigation. The results of discrete
multiscale wavelet shrinkage and related discrete versions of integrodifferential equations are compared with
respect to their denoising quality by numerical experiments.
A wavelet transform based multiresolution edge detection and classification schema
Show abstract
In this work a new multiresolution method to detect and classify edges appearing in images has been proposed. The edge
detection and classification schema is based on the analysis of the data obtained by a multiresolution image analysis
using Mallat and Zhong's wavelet. Multiresolution analysis allows to detect edges of different relevance at different
scales, as well as to obtain other important aspects of the detected edge. The Discrete Wavelet Transform proposed by
Mallat and Zhong has been used for detection and classification purposes. The classification schema developed is based
on a simple polynomial-fitting algorithm. Analyzing properties of the fitted polynomial we are able to classify several
edge types. The robustness of the proposed method has been tested with different geometrical contour types appeared in
the literature. A real edge type has also been introduced: the 'noise', that allow us to implement a novel noise-filtering
algorithm simply by eliminating the points belonging to this class. The proposed classification schema could be
generalized to real edge types: shadows, corners, etc.
Watermarking
Capacity evaluation for MPEG-4 AVC watermarking
Show abstract
Nowadays, robust watermarking clearly identified its functionality within the multimedia production chain, from the content creation to the end-user consumption: property right identification and copy-maker tracking. In the quest for the speed required by today's real-time applications, compressed-domain watermarking becomes a hot research topic. This study evaluates the watermarking capacity in the MPEG-4 AVC domain in order to establish whether and to what extent compressed domain watermarking is viable. In this respect, the additive watermarking techniques are modelled by discrete noisy channels with non-causal side information at the transmitter. The study considers several attacks (linear and non-linear filtering, geometric) and computes the capacity of the corresponding channels. The experimental results are obtained out of processing a natural video corpus of 10 video sequences belonging to different movies, each of them about 25 minutes long (35000 frames in each video sequence).
A new watermarking method based on the use of the hyperanalytic wavelet transform
Corina Nafornita,
Ioana Firoiu,
Jean-Marc Boucher,
et al.
Show abstract
Watermarking using pixel-wise masking in the wavelet domain proves to be quite robust against common signal
processing attacks. Initially, in a system proposed by Barni et al., embedding is made only in the highest resolution level;
there are two disadvantages to this technique: the watermark information can be easily erased by a potential attacker and
embedding in the DWT is susceptible to geometric attacks, such as shifting. To enhance this watermarking method, we
use a modified perceptual mask that models the human visual system behavior in a better way, previously proposed by
the authors. The texture content is appreciated with the local standard deviation of the original image, which is further
compressed in the wavelet domain. Since the approximation image of the coarsest level contains too little information,
we appreciate the luminance content using a higher resolution level approximation sub-image. To increase the capacity
of the watermarking scheme the embedding is made in the HWT domain, using two strategies: in the real parts of the
HWT coefficients and in the absolute value of the HWT coefficients of the original image. The implementation of the
HWT is made using a new technique, recently proposed by the authors. Moreover, we make use of all the levels except
the coarsest one, for attack resilience. We use three types of detectors that take advantage of the hierarchical
decomposition. Tests were made for different attacks (JPEG compression, median filtering, resizing, cropping, gamma
correction, blurring, shifting and addition of white Gaussian noise), that prove the effectiveness of perceptual
watermarking in the HWT domain.
Image watermarking in the Hermite transform domain with resistance to geometric distortions
Show abstract
This paper proposes a novel perceptual watermarking scheme operating in a Hermite transform domain. To achieve an
acceptable level of watermark invisibility, masking properties of the Human Vision system (HVS) are exploited in the
extraction of relevant local image features (texture, smooth regions, edges) for watermark embedding purpose. Many
other works suggest the use of wavelets or contourlets. In our case, image features are extracted efficiently from the
Hermite transform image representation which agrees with the Gaussian derivative model of the human visual
perception. The resulting weighing mask is used to adapt the watermark strength to image regions during the embedding
process.
In order to ensure watermark resistance to global affine geometric attacks (rotation, scaling, translation and shearing) the
design of the watermarking scheme is modified, mainly, by incorporating a normalization procedure. Image
normalization, a means to achieve invariance to geometric transformations, is well known in computer vision and pattern
recognition areas. In this new design, both watermark embedding and detection are carried out in the Hermite transform
domain of moment-based normalized images.
A sequence of tests is conducted on various images. Many removal attacks (JPEG compression, additive noise and
filtering) as well as geometric attacks are applied from the Checkmark benchmark. Experimental results show the
effectiveness of the whole scheme in achieving its goals in terms of watermark invisibility and robustness.
A novel adaptive multi-resolution combined watermarking algorithm
Gui Feng,
QiWei Lin
Show abstract
The rapid development of IT and WWW technique, causing person frequently confronts with various kinds of authorized
identification problem, especially the copyright problem of digital products. The digital watermarking technique was
emerged as one kind of solutions. The balance between robustness and imperceptibility is always the object sought by
related researchers. In order to settle the problem of robustness and imperceptibility, a novel adaptive multi-resolution
combined digital image watermarking algorithm was proposed in this paper. In the proposed algorithm, we first
decompose the watermark into several sub-bands, and according to its significance to embed the sub-band to different
DWT coefficient of the carrier image. While embedding, the HVS was considered. So under the precondition of keeping
the quality of image, the larger capacity of watermark can be embedding. The experimental results have shown that the
proposed algorithm has better performance in the aspects of robustness and security. And with the same visual quality,
the technique has larger capacity. So the unification of robustness and imperceptibility was achieved.
Advanced Imaging
Adaptive illumination source for multispectral vision system applied to material discrimination
Show abstract
A multispectral system based on a monochrome camera and an adaptive illumination source is presented in this paper. Its
preliminary application is focused on material discrimination for food and beverage industries, where monochrome,
color and infrared imaging have been successfully applied for this task. This work proposes a different approach, in
which the relevant wavelengths for the required discrimination task are selected in advance using a Sequential Forward
Floating Selection (SFFS) Algorithm. A light source, based on Light Emitting Diodes (LEDs) at these wavelengths is
then used to sequentially illuminate the material under analysis, and the resulting images are captured by a CCD camera
with spectral response in the entire range of the selected wavelengths. Finally, the several multispectral planes obtained
are processed using a Spectral Angle Mapping (SAM) algorithm, whose output is the desired material classification.
Among other advantages, this approach of controlled and specific illumination produces multispectral imaging with a
simple monochrome camera, and cold illumination restricted to specific relevant wavelengths, which is desirable for the
food and beverage industry. The proposed system has been tested with success for the automatic detection of foreign
object in the tobacco processing industry.
Extended dynamic range imaging in shearography
Show abstract
Extended dynamic range (EDR) imaging is a post-processing technique commonly associated with photography. A set of
images of the same scene which are recorded by the camera using different shutter settings are integrated into a single
higher dynamic range image. Image processing software is used to identify pixels in the image which are correctly
exposed, underexposed or overexposed and pixels with the optimum intensity are rescaled with respect to their exposure
time and fused to form the EDR image. An intensity rescaling processing is then required as the EDR images generated
generally exceed the contrast ratio of conventional displays and printers. In optical metrology, speckle interferometry
and holography techniques require a well-modulated intensity signal to extract the phase information and of these
techniques, shearography is most sensitive to different object surface reflectivities as it uses self-referencing from a
sheared image. In this paper the authors demonstrate real-time EDR imaging in shearography extending an 8-bit camera
to 16-bits. Experimental results from a difficult surface reflectivity sample, a wooden panel painting containing gold and
dark earth colour paint, are presented. A pre-processing step that could be incorporated in the image capture routine was
identified and implemented. This allowed the existing library of 8-bit image processing routines to be used without
modification.
Minimum image entropy technology applied to the real-time autofocusing system of space optical remote sensors
Linghua Guo,
Jianquan Li,
Yi Li,
et al.
Show abstract
For space remote sensors with high resolution, large caliber, and long focal length, in-orbit automatic focusing technique
is a significant application technology. Minimum image entropy (MIE) technology applied to real-time autofocusing
system of space optical remote sensor possesses creativity and engineering significance. MIE's theoretical analysis,
algorithm's computer simulation, and "Experimental Optical System" experiment have successfully validated MIE's
validity as the criterion for optical remote sensor's auto-focusing. Related data indicate that for a diffraction-limited
optical system with the f-ratio of F#=39, the detecting sensitivity for checking-focus can be better than 0.1 mm, by means
of MIE.
On a method to eliminate moving shadows in video sequences
Show abstract
We present a simple computational model that works in the RGB colour space to detect moving shadow pixels in video
sequences of indoor scenes, illuminated in each case by an incandescent source. A channel ratio test for shadows cast on
some common indoor surfaces is proposed that can be appended to the developed scheme so as to reduce the otherwise
high false detection rate. The core method, based on a Lambertian hypothesis, has been adapted to work well for near-matte
surfaces by suppressing highlights. The results reported, based on an extensive data analysis conducted on some
of the crucial parameters involved in the model, not only bring out the subtle details of the parameters, but also remove
the ad hoc nature of the chosen thresholds to a certain extent. The method has been tested on various indoor video
sequences; the results obtained indicate that it can be satisfactorily used to mark or eliminate the strong portion of the
foreground shadow region.
Medical Imaging
Preprocessing and analysis of microarray images from integrated lensless bio-photonic sensors
Show abstract
A new integrated lensless bio-photonic sensors is being developed.
It replaces the ordinary slide supporting the DNA spots, and the
complex, large and expensive hybridisation and the scanner reading
system, by a sandwich of well defined chemical and optical layers
grafted onto a CCD sensor. The upper layer of the new biochip
performs the biological function.
Due to the architecture of the biochip leading to a lensless imaging
of the spots directly on the sensor pixels, the images produced will
have novel characteristics beyond the analysis capacity of reading
software packages of microarray analysis. In this framework,
specific image processing and statistical data analysis algorithms
have been developed in order to assess and to quantify these images.
Imaging of laser-induced thermo-elastic stress in biotissues with shadowgraph
Show abstract
This work presents the results of shlieren visualization of thermo-elastic stress, which is induced by laser irradiation in the
bulk of tissue or tissue-like gel. This thermo-elastic stress results from water absorption of the pulses of an Er-fiber laser
radiation, which produced thermal effect in irradiated zone of tissue. Imaging of laser-irradiated zone of tissue with
shadowgraph and digital image processing showed that thermal stress-zone is located in the tissues at the depth of
absorption of laser radiation. Digital streak camera allowed us to observe frame-by-frame growth of thermal stress in
irradiated tissue. Imaging of tissue with shadowgraph presented data on stress-waves, bubbles formation during a laser
pulse repetition and generation of hydrodynamic flow after a laser pulse.
A structured light system to guide percutaneous punctures in interventional radiology
Show abstract
Interventional radiology is a new medical field which allows percutaneous punctures on patients for tumoral
destruction or tissue analysis. The patient lies on a CT or MRI table and the practitioner guides the needle
insertion iteratively using repetitive acquisitions (2D slices). We aim at designing a guidance system to reduce the
number of CT/MRI acquisitions, and therefore decrease the irradiation and shorten the duration of intervention.
We propose a system composed of two calibrated cameras and a structured light videoprojector. The cameras
track at 15Hz the needle manipulated by the practitioner and a software displays the needle position with respect
to a preoperative segmented image of the patient. To register the preoperative image in the camera frame, we firstly reconstruct the patient skin in 3D using the structured light. Then, the surfacic registration between the
reconstructed skin and the segmented skin from the preoperative image is performed using the Iterative Closest
Point (ICP) algorithm. Ensuring the quality of this registration is the most challenging task of the system.
Indeed, a surfacic registration cannot correctly converge if the surfaces to be registered are too smooth.
The main contribution of our work is the evaluation on patients of the conditions that can ensure a correct
registration of the preoperative skin surface with the reconstructed one. Furthermore, in case of unfavourable
conditions, we propose a method to create enough singularities on the patient abdomen so that the convergence
is guaranteed. In the coming months, we plan to evaluate the full system during standard needle insertion on
patients.
Interaction between Optics and Digital Image Processing
Analysis of polarization vortices generated from a polarization diffractive mask
Show abstract
In this work we analyze the diffraction pattern generated by a four quadrant polarization mask. Different quadrants
transmit different linear polarizations. We show that the diffraction pattern, in the absence of analyzer, shows the
intensity distribution characteristic from a square aperture, but the state of polarization is not uniform. In particular, we
demonstrate that an array of singular points (polarization vortices) is being generated. When an analyzer is placed behind
the polarization diffractive mask, the intensity of the diffracted field is changed as the analyzer rotates. Experimental
verification is performed with a twisted nematic liquid crystal display.
The assessment of phase only filters in imaging systems by the classical optical merit functions
Show abstract
Spatial light modulators (SLM) have found a wide range of applications in many fields of optical imaging and
measurement systems. We implement different phase only filters in the pupil plane of an imaging system. The phase
only filter is divided in five equally spaced annuli. Each annulus has a different phase transmission and inside each
annulus the phase is constant. We analyse first the influence of linear decreasing or increasing phase, second we use one
phase annulus with a phase shift of π in different positions over the pupil and finally an alternating phase between 0 and
π over the pupil.
Merit functions of the different filters are calculated. The radial and axial point spread function (PSF) or the 3D line
spread function show that in some cases these phase only filters will shift the best image plane. The experimental results
show the close correlation to the calculated shift of the best image plane.
The strong side lobes that appear in the merit functions lead to the conclusion that the image quality will be influenced
as well. This can be confirmed by the calculation and the measurement of the image intensity. So in order to get more
information about the expected image it is necessary to study the 3D modulation transfer function (MTF). With the MTF
one can see that the contrast decreases for the image obtained with each filter in comparison with the image obtained
with the clear pupil.
The conclusion of our work is, that it is necessary to study the influence of all merit functions in order to design an
optimum filter for a given application.
Image capture and processing for a microoptical compound-eye sensor
Show abstract
An artificial compound-eye imaging system has been developed consisting of one planar array of microlenses
positioned on a spacing structure and coupled to a commercial CMOS optoelectronic detector array of different pitch,
providing different viewing directions for the individual optical channels. Each microlens corresponds to one channel,
which can be related to one or more pixels due to the different fill factors of the microlens array and the image sensor.
Also alignment problems resulting from the matching of the microlens focal spots and the pixels during the assembly
and the possible residual rotation between the artificial compound-eye objective and the pixel matrix are considered. We
have written a program to automatically select the illuminated pixels of the sensor which correspond to each channel in
order to form the final image. This calibration method is based on intensity criterions besides the geometric disposition
of the microlens array. An image capture program that uses only the channels selected by the calibration is also
presented. This program additionally implements image post-processing methods adapted to the microoptical
compound-eye sensor. They are applied to the captured images in real time and allow increasing the contrast of the
captured images. One of the methods used is the Wiener filter that is computed by taking into account an approximation
of the multichannel imaging process of microoptical compound-eye sensors. Experimental results are presented, which
show a noticeable increase in the frequency response when the Wiener filter is used, partially compensating the
characteristic low spatial resolution of the artificial compound eyes.
Spatial transform in non-conventional ultra-high resolution image-carrying fiber bundle
Show abstract
Restricted by the device fabrication technology, traditional scanning imaging system can not get high spatial resolution, and it's difficult to get a line array whose sensor number is over 6000. A kind of non-conventional fiber bundle is designed to couple information. Using existing small scale plane arrays to receive the signals of the fiber bundle's output can acquire ultra-high spatial resolution image. However, owing to the special structure of the bundle, the image processing is very complicated. Double module extremum filter algorithm is proposed in this paper. This algorithm can eliminate the blind pixels caused by broken fibers, locate the center of each fiber in the output terminal of fiber, recover
the input images completely, and achieve the spatial transform.
Poster Session
Scale and rotation invariant 3D object detection using spherical nonlinear correlations
Show abstract
Three-dimensional object recognition with scale and rotation changes is addressed. The recognition method is described
in terms of correlations between spherical surfaces. Tridimensionality is codified into range images. We used the phase
Fourier transform of a range image which gives information about the orientation of the 3D object surfaces. A 3D object
orientation map containing information about all possible rotations of the object is obtained. This map distribution is
calculated using the amplitude of the phase Fourier transform for different views of the object. From that 3D object
description it is possible to achieve detection and rotation estimation by performing a correlation between unit spheres
even when only partial information is presented. In addition, a scale change of a rotated 3D object implies a change of
the intensity in the unit sphere. We define correlations between the reference unit sphere and a certain target-patch
placed on the surface of the unit sphere. Various experiments are carried out to confirm the correct detection. We also
validate the method when other false targets were used. In addition to tolerance to scale and rotation, high discrimination
against false targets is also achieved.
Analysis of non-ideal behavior of CAPD based time-of-flight pixels
Show abstract
An overview is given of the output of two main pixels based on a Current Assistant Photonic Demodulator (CAPD) and
their usability in a ranging camera based on the Time-Of-Flight (TOF) principle. The first structure is based on an active
pixel with an integrating capacitor, while the second structure is based on a transimpedance amplifier. A measurement
system capable of interfacing with a one-dimensional array of CAPD pixels has been designed and is briefly discussed.
The linearity of both CAPD pixels is measured, compared to each other and discussed. Several sources of non-idealities
present in the CAPD output have been investigated, including differences between rise and fall time and duty cycle
imperfections of the high frequency binary correlating signal, as well as CAPD bandwidth. The possible influences of
such non-idealities of the correlating wave on the measurements as well as on the calculated distance are discussed.
Resulting non-idealities in the raw CAPD output values include clipping and non-linearity, yielding offsets and non-linearities
in the calculated phase difference between the original and the reflected signal. Theoretic values are shown
and compared against measurements.
Finally, inter-pixel properties of a one-dimensional array of CAPD structures are presented, as well as a sample range
image acquired by scanning.
Applying SIMD to optical character recognition (OCR)
Show abstract
Optical Character Recognition (OCR) techniques are widely used in data/text entry, process automation. Decades of
research efforts have made the accurate recognition of typewritten text largely accepted as a solved problem. Driven by
practical usage demands, the low complexity and high performance implementation techniques of OCR systems are
studied. Recent research shows that it may not be possible even for a simple OCR to run on a portable device without a
specialized digital signal processor. In this paper, we present a highly data-parallelized implementation of OCR for
typewritten text onto the linear processor array of the Xetal chip. Besides the preprocessing stage, the most computation
intensive part of OCR recognizing individual characters is highly parallelized onto the Single Instruction Multiple Data
(SIMD) engine of the Xetal chip, which can process a VGA-resolution text frame within one tenth of a second. In
addition, different parallelization schemes are explored to make trade-off between the degree of parallelism and the costs
of preprocessing to reorganize data to feed the SIMD engine and post-processing to assemble and collect results. The
exploration of parallelized OCR application brings additional performance gain when mapped onto the linear processor
array of the Xetal chip.
Image processing with JPEG2000 coders
Show abstract
In the note, several wavelet-based image processing algorithms are presented. Denoising algorithm is derived
from the Donoho's thresholding. Rescaling algorithm reuses
sub-division scheme of the Sweldens' lifting and
a sensor linearization procedure exploiting system identification algorithms developed for nonlinear dynamic
systems. Proposed autofocus algorithm is a passive one, works in wavelet domain and relies on properties of
lens transfer function. The common advantage of the algorithms is that they can easily be implemented within
the JPEG2000 image compression standard encoder, offering simplification of the final circuitry (or the software
package) and the reduction of the power consumption (program size, respectively) when compared to solutions based on separate components.
An iris recognition algorithm based on DCT and GLCM
G. Feng,
Ye-qing Wu
Show abstract
With the enlargement of mankind's activity range, the significance for person's status identity is becoming more and
more important. So many different techniques for person's status identity were proposed for this practical usage.
Conventional person's status identity methods like password and identification card are not always reliable. A wide
variety of biometrics has been developed for this challenge. Among those biologic characteristics, iris pattern gains
increasing attention for its stability, reliability, uniqueness, noninvasiveness and difficult to counterfeit. The distinct
merits of the iris lead to its high reliability for personal identification. So the iris identification technique had become hot
research point in the past several years.
This paper presents an efficient algorithm for iris recognition using gray-level co-occurrence matrix(GLCM) and Discrete
Cosine transform(DCT). To obtain more representative iris features, features from space and DCT transformation domain
are extracted. Both GLCM and DCT are applied on the iris image to form the feature sequence in this paper. The
combination of GLCM and DCT makes the iris feature more distinct. Upon GLCM and DCT the eigenvector of iris
extracted, which reflects features of spatial transformation and frequency transformation. Experimental results show that
the algorithm is effective and feasible with iris recognition.
Automatic detection and recognition of traffic signs in stereo images based on features and probabilistic neural networks
Show abstract
Considering the problem of automatic traffic sign detection and recognition in stereo images captured under motion
conditions, a new algorithm for traffic sign detection and recognition based on features and probabilistic neural networks
(PNN) is proposed in this paper. Firstly, global statistical color features of left image are computed based on statistics
theory. Then for red, yellow and blue traffic signs, left image is segmented to three binary images by self-adaptive color
segmentation method. Secondly, gray-value projection and shape analysis are used to confirm traffic sign regions in left
image. Then stereo image matching is used to locate the homonymy traffic signs in right image. Thirdly, self-adaptive
image segmentation is used to extract binary inner core shapes of detected traffic signs. One-dimensional feature vectors
of inner core shapes are computed by central projection transformation. Fourthly, these vectors are input to the trained
probabilistic neural networks for traffic sign recognition. Lastly, recognition results in left image are compared with
recognition results in right image. If results in stereo images are identical, these results are confirmed as final recognition
results. The new algorithm is applied to 220 real images of natural scenes taken by the vehicle-borne mobile
photogrammetry system in Nanjing at different time. Experimental results show a detection and recognition rate of over
92%. So the algorithm is not only simple, but also reliable and high-speed on real traffic sign detection and recognition.
Furthermore, it can obtain geometrical information of traffic signs at the same time of recognizing their types.
Invariant approach to the character classification
Show abstract
Image moments analysis is a very useful tool which allows image description invariant to translation and rotation, scale
change and some types of image distortions. The aim of this work was development of simple method for fast and
reliable classification of characters by using Hu's and affine moment invariants. Measure of Eucleidean distance was
used as a discrimination feature with statistical parameters estimated. The method was tested in classification of Times
New Roman font letters as well as sets of the handwritten characters. It is shown that using all Hu's and three affine
invariants as discrimination set improves recognition rate by 30%.
Fast deconvolution with non-invariant PSF for 3-D fluorescence microscopy
Show abstract
The 3-D fluorescence microscope is a powerful method for imaging and studying living cells. However, the data
acquired with conventional 3-D fluorescence microscope are not quantitatively significant for spatial distribution or
volume evaluation of fluorescent areas in reason of distortions induced on data by the acquisition process.
Theses distortions must be corrected for reliable measurements. The knowledge of the impulse response
characterizing the instrument permits to consider the backward process retrieving the original data. One realize a
deconvolution opposed to the convolution process induced by the microscope, projecting the 'object' space in the
'image' space. However, when the response of the system is not invariant in the observation field, the classical
algorithms using Fourier Transform for computations are not usable.
The contribution of this work is to present several approaches making it possible to use the Fourier Transform
in non-invariance conditions and to simulate it's application in the 3-D fluorescence microscope problems.
Gray level image reconstruction using Jacobi-Fourier moments
Show abstract
It is well known that, different types of Jacobi radial functions can be obtained from generic formulas changing two real parameters
α and β. In this paper a comparative gray level image reconstruction analysis for the cases α = β = 2; α = 3, β=2; α=β = 3; and α = β = 4 is done. As a reconstruction measurement is implemented the normalized image reconstruction error (NIRE). The test images analyzed in this work are acquired from flowers.
High-speed smart camera with embedded feature extractions and profilometry measurements
Show abstract
Nowadays, high-speed imaging offers high investigation possibilities for a wide variety of applications such as motion
study, manufacturing developments. Moreover, due to the electronic progresses, real-time processing can be
implemented in the high-speed acquisition systems. Important information can be extracted in real-time from the image
and then be used for on-line controls. Therefore we have developed a high-speed smart camera with high-speed CMOS
sensor, typically 500 fps with a 1.3 Mega-pixels resolution. Different specific processing have been implemented inside
an embedded FPGA according to the high-speed data-flow. The processing are mainly dedicated to feature extraction
such as edge detection, or image analysis, and finally markers extraction and profilometry. In any case, the data
processing allows to reduce the large data flow (6.55 Gbps) and to propose a transfer on a simple serial output link as
USB 2.0. This paper presents the high-speed smart camera and focuses two processing implementations: the marker
extraction and the related profilometry measurement. In the marker extraction mode, the center of mass is determined for
each marker by a combination of image filtering. Only the position of the center is transferred via the USB 2.0 link. For
profilometry measurements, a simplify algorithm has been implemented at low-cost in term of hardware resources. The
positions of the markers or the different object's profiles can be determined in real-time at 500 fps with full resolution
image. A higher image rate can be reached with a lower resolution (i.e. 500 000 profiles for a single row image).
Improvement of recording density by using a Wiener filter in multiplexed hologram recording
Show abstract
To improve the recording density of Polytopic multiplexed recording, we studied the spatial filtering effect from
the view points of the filter size and signal-to-noise-ratio (SNR) of the reproduced signal. The relationship
between SNR and spatial filter size is evaluated by using three dimensional model and Fresnel diffraction analysis.
By applying the Wiener filter to the reproduced signal, we can improve SNR for small spatial filter size.
Wavefront coding using coma aberration for dual field of view IR systems
Show abstract
Wavefront coding (WFC) is a powerful hybrid optical-numerical technique for increasing the depth of
focus of imaging systems. There is a low cost solution that uses decentred lenses inducing coma as an
adjustable and removable phase element. This coding has been proven useful for IR systems. However
these systems usually have several configurations with multiple fields of view. Unless the detector is
uncooled, the f/number of the system is maintained for all configurations thus entrance pupil size changes
for each one. As a result, the coding coma changes accordingly. This paper presents an approach to
maintain the same amount of coma for dual field of view (DFOV) systems.
Shape measurement of transparent objects with polarization imaging
Show abstract
This paper deals with a 3D measurement system based on the "Shape from Polarization" method applied for transparent
objects.
The method is an application of polarization imaging techniques and its principle is as follows: after being reflected, an
unpolarized light becomes partially linearly polarized. By analyzing its polarization parameters and by knowing the
refractive index of the object to be controlled, the surface normals can be evaluated. Finally, the 3D shape is obtained by
integrating the normals field.
Section 1 introduces the opportunity to use polarisation imaging, section 2 recalls the measurement method and describes
the angle measurement ambiguities naturally appearing and discusses how to overcome them, section 3 introduces the
measurement setup and a part of its calibration and section 4 describes some examples on transparent objects.
Study of the holographic phase stitching technique
Show abstract
Stitching technology is used to enlarge the tested area in interference and fringe projection system. By using this method
we can test the profile of object which is larger than testing aperture. In the traditional digital holographic technique,
only small objects can be recorded and reconstructed. In this study, a holographic phase stitching technique is proposed
to solve this problem. Differ from the interference and fringe projection stitching, holographic phase stitching has its
characteristic in the holographic diffraction process. In this paper, taking plane image measurement as an example, the
theory of holographic phase stitching is presented. Stitching errors in the holographic phase stitching are deeply analyzed
and relevant 2×2 stitching simulation is performed using a phase grating.
A dynamic three-dimensional display technique based on liquid crystal spatial light modulator
Show abstract
A dynamic three-dimensional (3D) display method based on Liquid Crystal on Silicon (LCoS) spatial light modulator is
introduced in this paper. Sixty viewing-angles are set at regular intervals of six degrees along a complete circle in the
horizontal plane, and phase holograms of three-dimensional virtual object are calculated under these viewing-angles by
use of computational tomographic method as well as spatial coordinate transformation. Multi- kinoforms are calculated
for each angle by adding proper pseudorandom phase in each object plane, which can avoid losing spectrum information
and be able to reduce the speckle noise of reconstructed image. These kinoforms are written to a 3D electro-holographic
display system based on LC-R2500 spatial light modulator (SLM) by a proper loading sequence. A 650 nm diode laser
with power of 150 mW is used in the system. A transparent screen is taken as a display media, which could show
three-dimensional images reconstructed from holograms loaded on LC-R2500. A pair of Fresnel lenses is used to
improve the viewing effect and three-dimensional images floating in space can be viewed by multiple observers.
Advances in LCoS SLM characterization for improved optical performance in image processing
Show abstract
Electrically addressed spatial light modulators (SLM) are widely used in optical image processors to display not only
input images but also a huge variety of optical components such as lenses, complex filters and other diffractive elements.
These components are fully programmable and dynamically controllable by computer thus bringing flexibility and new
degrees of freedom to current optical and digital image processors. A good characterization is the most important step in
the SLM initialization. The quality and effectiveness of the optical component addressed to the SLM strongly depends on
the knowledge of the device response. This work deals with the spatial and temporal characterizations of reflective zerotwist
liquid-crystal on silicon (LCoS) SLM. The signal is spatially modified before addressing it to the LCoS SLM to
compensate for the distortions internally introduced by the device. For time varying optical components, the signal is
also modified before addressing it to the LCoS SLM to compensate for the distortions internally introduced by the device
when phase variations of 2π are required at high rate. Experimental results and applications in image processing are
shown.
3D imaging of radioactivity in man measured with a whole body counter
Show abstract
In the Medical Physics Department of the University of Insubria, Varese, Italy, a whole body counter is in use, for
clinical and radioprotection measurements. It consists of a scanning bed, four opposite (anterior-posterior and laterallateral)
NaI(Tl) detectors and a shielding based on the shadow-shield principle. By moving the bed on which the patient
lies in the supine position, the longitudinal profiles of the counts measured by each probe along the patient axis are
obtained. Making the assumption that radioactivity is distributed in N voxel sources located in N selected positions in
the patient, this distribution is calculated by solving an over-determined system of linear equations. The solution can be
calculated using different methods. An iterative method and a regularization technique are presented. The algorithm
proposed allows the evaluation of the distribution of the radioactivity in 3D in voxels with dimensions ranging from 15
to 20 mm, depending on the size of the patient. The 3D distribution of the radioactivity and the knowledge of the time
of the intake allow the assessment of the effective dose.
Partial volume correction of whole body PET images using the wavelet transform
Show abstract
A general approach for partial volume correction of positron emission tomography (PET) images is introduced. The method is based on the merging of functional information from PET images and anatomical information using high resolution anatomical images. In order to decompose the PET and high resolution images
the "á trous" algorithm was implemented. Results obtained with simulated and real patients images show a significant partial volume reduction and image enhancement. The relative errors in the partial volume corrected image are always less than 3,6% with respect to 16% of the original image.
Determining locus and periphery of optic disk in retinal images
Show abstract
Diabetes can be recognized by features of retina. Automatic retina feature extraction improves the speed of diabetes
diagnosis. The first step in extracting the features is to localize the optic disk. Methods with low accuracy in localizing
the optic disk include area with maximum lightness or the largest area containing pixels with maximum gray levels. A
more accurate method is to find the physical position of blood vessel that passes through optic disk. This paper presents a
fast and accurate algorithm for localizing the optic disk. The process of localization consists of finding the target area,
Optic Disk center and Optic Disk boundaries. Optic Disk boundaries are recognized by our algorithm with %90
accuracy.
Algorithms improvement in image processing for optical observations of artificial objects in geostationary orbit with the TAROT telescopes
Show abstract
TAROT (Télescope á Action Rapide pour les Objets Transitoires - Rapid Action Telescope for Transient Objects)
is a network of robotic ground based telescopes. Since 2002, we use them for a survey of artificial objects
(satellites, debris) in the geostationary orbit. The objects are detected, their orbit is computed, and follow-up
observations are planned. We are currently implementing new, more efficient, image processing algorithms in
order to improve the processing speed, the sensitivity, and to decrease the rate of false detections. We present
our new implemented methods, as well as the results obtained.
Feature-constrained surface reconstruction approach for point cloud data acquired with 3D laser scanner
Show abstract
Surface reconstruction is an important task in the field of 3d-GIS, computer aided design and computer graphics (CAD
& CG), virtual simulation and so on. Based on available incremental surface reconstruction methods, a feature-constrained
surface reconstruction approach for point cloud is presented. Firstly features are extracted from point cloud
under the rules of curvature extremes and minimum spanning tree. By projecting local sample points to the fitted tangent
planes and using extracted features to guide and constrain the process of local triangulation and surface propagation,
topological relationship among sample points can be achieved. For the constructed models, a process named consistent
normal adjustment and regularization is adopted to adjust normal of each face so that the correct surface model is
achieved. Experiments show that the presented approach inherits the convenient implementation and high efficiency of
traditional incremental surface reconstruction method, meanwhile, it avoids improper propagation of normal across sharp
edges, which means the applicability of incremental surface reconstruction is greatly improved. Above all, appropriate k-neighborhood
can help to recognize un-sufficient sampled areas and boundary parts, the presented approach can be used
to reconstruct both open and close surfaces without additional interference.
Atmospheric correction for inland water based on Gordon model
Show abstract
Remote sensing technique is soundly used in water quality monitoring since it can receive area radiation information
at the same time. But more than 80% radiance detected by sensors at the top of the atmosphere is contributed by
atmosphere not directly by water body. Water radiance information is seriously confused by atmospheric molecular and
aerosol scattering and absorption. A slight bias of evaluation for atmospheric influence can deduce large error for water
quality evaluation. To inverse water composition accurately we have to separate water and air information firstly. In this
paper, we studied on atmospheric correction methods for inland water such as Taihu Lake. Landsat-5 TM image was
corrected based on Gordon atmospheric correction model. And two kinds of data were used to calculate Raleigh
scattering, aerosol scattering and radiative transmission above Taihu Lake. Meanwhile, the influence of ozone and white
cap were revised. One kind of data was synchronization meteorology data, and the other one was synchronization
MODIS image. At last, remote sensing reflectance was retrieved from the TM image. The effect of different methods
was analyzed using in situ measured water surface spectra. The result indicates that measured and estimated remote
sensing reflectance were close for both methods. Compared to the method of using MODIS image, the method of using
synchronization meteorology is more accurate. And the bias is close to inland water error criterion accepted by water
quality inversing. It shows that this method is suitable for Taihu Lake atmospheric correction for TM image.
Mapping of alpine grassland cover in western China from normalized Landsat TM image
Show abstract
Grassland cover near Lake Qinghai in western China was mapped into nine percentage classes from a TM-derived
Normalised Difference Bareness Index (NDBI) image based on 178 in situ samples collected within 1 m2 sites. Their
ground coordinates logged with a GPS unit were used to locate their pixel values on the NDBI image. A new method, in
which the in situ samples and their pixel NDBI values were independently ranked prior to the establishment of their
linear regression relationship, was applied to converting the NDBI image into a map of grass coverage. This relationship
enabled the NDBI image to be translated into a map of grassland cover with a meaningful spatial pattern. Assessed
against visually interpreted results, grassland cover was mapped at an overall accuracy of 80%. In order for this method
to generate satisfactory results, image pixel NDBI values have to be normalized so that they have the same standard
deviation as that of the ground samples. This proposed method should be applicable to any grassland where grassland
cover varies subtly at the pixel scale of the image used.
Example based learning of image stitching for an omni-directional camera using a variational optical flow methodology
Show abstract
Omni-Directional vision plays an important role in autonomous and remotely controlled vehicles providing the
critical ability of peripheral situational awareness. We introduce an omni-directional system which is able to build
a high resolution uniform panoramic image from four different wide angled cameras. In order to build a uniform
panoramic image, we developed a state of the art stitching algorithm using a variational optical flow estimation
methodology. Optical flow is traditionally considered as the apparent 2D image motion captured by a single
camera in different time samples. In this paper on the other hand, we consider optical flow as the 2D motion
registering the overlap regions of images taken from different cameras at the same time instant. Since the rigid
geometry between the cameras is fixed, the optical flow registering the different views is fixed for distant scenes.
We use this fact in order to formulate a functional which requires that the same optical flow registers properly
all the provided scene examples taken in the learning process. Our minimization functional incorporates in the
data term all the available information as provided by the scene examples. We mathematically show that the
variety of scene examples helps to overcome the aperture problem inherent in traditional optical flow problems.
We demonstrate the robustness and accuracy of our method on synthetic test cases and on real images captured
by our omni-directional commercial product.
Quantitative retrieval of chlorophyll-a by remote sensing based on TM data in Lake Taihu
Show abstract
Based on TM (ETM) data and in-situ measurements of chlorophyll-a concentration (Chl-a) in Lake
Taihu, analysis was conducted to decide the correlation between Chl-a and the ratios of different
reflectance corrected by the 6S model. The results show Chl-a is closely related to TM3/(TM1+TM4)
and the inverse model to infer Chl-a in Lake Taihu can be written as
Ln(Chl-a)=-9.247*(TM1+TM4)/(TM2+TM3) -27.903*TM3/(TM1+TM4) +24.518. However, the accuracy of this model can not be assured due to the complexity of spectral reflectance strongly dependant on water quality in Lake Taihu. Thus we developed a further 2-layer BP neural net model based on 4 input nodes, 7 hide nodes and 1 output node to for calculating Chl-a in the lake. The derived results reveal that the BP model has much higher accuracy than the linear model. A test was made based on 16 samples and suggests that the maximum relative error (RE) of BP model was only 35.43%. Of all the samples, 15 ones had a RE of less than 30% from the BP model.. However, there were only 3 samples with RE less than 30% from the results derived from the linear model. The comparison shows that the BP model has high availability for inferring Chl-a of surface water having complex spectral reflectance.
Retrieval of land cover information under thin fog in Landsat TM image
Yuchun Wei
Show abstract
Thin fog, which often appears in remote sensing image of subtropical climate region, has resulted in the low image
quantity and bad image mapping. Therefore, it is necessary to develop the image processing method to retrieve land
cover information under thin fog. In this paper, the Landsat TM image near the Taihu Lake that is in the subtropical
climate zone of China was used as an example, and the workflow and method used to retrieve the land cover information
under thin fog have been built based on ENVI software and a single TM image. The basic step covers three parts: 1)
isolating the thin fog area in image according to the spectral difference of different bands; 2) retrieving the visible band
information of different land cover types under thin fog from the near-infrared bands according to the relationships
between near-infrared bands and visible bands of different land cover types in the area without fog; 3) image post-process.
The result showed that the method in the paper is easy and suitable, and can be used to improve the quantity of
TM image mapping more effectively.
Landsat TM image feature extraction and analysis of algal bloom in Taihu Lake
Yuchun Wei,
Wei Chen
Show abstract
This study developed an approach to the extraction and characterization of blue-green algal blooms of the study area
Taihu Lake of China with the Landsat 5 TM imagery. Spectral feature of typical material within Taihu Lake were first
compared, and the most sensitive spectral bands to blue-green algal blooms determined. Eight spectral indices were then
designed using multiple TM spectral bands in order to maximize spectral contrast of different materials. The spectral
curves describing the variation of reflectance at individual bands with the spectral indices were plotted, and the TM
imagery was segmented using as thresholds the step-jumping points of the reflectance curves. The results indicate that
the proposed multiple band-based spectral index NDAI2 (NDAI2 = (B4-B1)*(B5-B3)/(B4+B5+B1+B3) performed
better than traditional vegetation indices NDVI and RVI in the extraction of blue-green algal information. In addition,
this study indicates that the image segmentation using the points where reflectance has a sudden change resulted in a
robust result, as well as a good applicability.
A novel multi-focus image fusion algorithm based on feature extraction and wavelets
Show abstract
Focusing cameras is an important problem in computer vision and microscopy. Due to the limited depth of
field of optical lenses in CCD devices, there are sensors which cannot generate images of all objects with equal
sharpness. Therefore, several images of the same scene have different focused parts. One way to overcome this
problem is to take different in-focus parts and combine them into a single composite image which contains the
entire focused scene. In this paper we present a multi-focus image fusion algorithm based on feature extraction
and wavelets. Classical wavelet synthesis is known to produce Gibbs phenomenon around discontinuities. The
approach of wavelet on the interval transform is suitable to orthogonal wavelets and does not exhibit edge
effects. Since Canny filter's operator is a Gaussian derivative, a well known model of early vision, we used
it to get salient edges and to build a decision map who determines which information to take and at what
place. Finally, quality of fused images is assessed using both traditional and perception-based quality metrics.
Quantitative and qualitative analysis of the results demonstrate higher performance of the algorithm compared
to traditional methods.
SONAR images despeckling using a Bayesian approach in the wavelet domain
Show abstract
During acquisition, the SONAR images are corrupted by multiplicative noise (speckle).The aim of an image denoising
algorithm is then to reduce the noise level, while preserving the image features. There is a great diversity of wavelet
based estimators used like denoising systems. The corresponding denoising methods have three steps: the computation of
the forward Wavelet Transform (WT); the filtering of the wavelet coefficients; and the computation of the inverse
wavelet transform of the result obtained. In the following, the Dual Tree Complex Wavelet Transform (DT-CWT) will
be associated with a variant of a maximum a posteriori bishrink filter because its explicit input-output relation permits a
sensitivity analysis. The bishrink filter has a high sensitivity with some parameters, especially in the homogeneous
regions. The main idea of this paper is to reduce this sensitivity by diversification. In this respect the regions with
different homogeneity degrees are identified and in each of them the WT of the acquired image is filtered using a number
of different variants of bishrink filters in accordance with its homogeneity.
Multi-sensor image fusion with the steered Hermite transform
Show abstract
The steered Hermite Transform is presented as an efficient tool for multi-sensor image fusion. The fusion algorithm is
based on the Hermite transform, which is an image representation model based on Gaussian derivatives that mimic some
of the most important properties of human vision. Moreover, rotation of the Hermite coefficients allows efficient
detection and reconstruction of oriented image patterns in reconstruction applications such as fusion and noise reduction.
We show image fusion with different image sensors, namely synthetic aperture radar (SAR) and multispectral optical
images. This case is important mainly because SAR sensors can obtain information independently of weather conditions;
however, the characteristic noise (speckle) present in SAR images possesses serious limitations to the fusion process.
Therefore noise reduction is a key point in the problem of image fusion. In our case, we combine fusion with speckle
reduction in order to discriminate relevant information from noise in the SAR images. The local analysis properties of
the Hermite transform help fusion and noise reduction adapt to the local image orientation and content. This is especially
useful considering the multiplicative nature of speckle in SAR images.
The location-searching of image-carrying fiber bundles based on Kalman filter
Show abstract
To obtain ultra-high spatial resolution image, a kind of non-conventional fiber bundle, whose input is line array while
output plane array, has been designed. The input terminal of the fiber bundle is set at the focal plane of scanning
imaging system. A plane array is set behind the output terminal to receive the information coupled by the fiber bundle.
The coordinates of each fiber in the output terminal and position mapping relationship between output and input of
each fiber in the fiber bundles must be ascertained when the fiber bundle is used in scanning image system. However,
limited by the fiber bundle manufacture technology, the fiber bundle output structure is not strictly aligned. There is
position error in both horizontal and vertical directions in every row and column, especially at the boundary of bundle,
which causes great difficulties in restoring the object image. An algorithm based on Kalman Filter, aiming at the
coordinates error, is proposed in this paper. By searching fiber center's position one by one in the output array, the
coordinates table of the output is established, and the position mapping relationship between output and input is built.
With the table, we can fully acquire the information of the scanning image system and restore the object image.
Digital holographic microscope with low spatial and temporal coherence of illumination
Show abstract
In this paper we present a newly developed digital transmission holographic microscope. The microscope enables using
arbitrarily low coherent illumination (both spatially and temporally) in conjunction with the off-axis holography. The
setup of the microscope, its function and the object wave reconstruction procedure are described. The optical sectioning
effect, similar to a confocal microscope, resulting from the use of low spatially coherent light source is demonstrated.
The microscope has been tailored for studies of living cell dynamics. Time-lapse phase reconstruction series of live cells
activities were carried out. The different behavior related to changes in the cell cycle is demonstrated.