Showing papers by "Alan C. Bovik published in 1994"

Anisotropic diffusion pyramids for image segmentation

Abstract: We introduce the Anisotropic Diffusion Pyramid (ADP), a structure for multiresolution image processing. We also develop the ADP for use in region-based segmentation. The pyramid is constructed using the anisotropic diffusion equations, creating an efficient scale-space representation. Segmentation is accomplished using pyramid node linking. Since anisotropic diffusion preserves edge localization as the scale is increased, the region boundaries in the coarse-to-fine ADP segmentation are accurately delineated. An application to segmentation of remotely sensed data is provided. The results of ADP segmentation are compared to Gaussian-based pyramidal segmentation. The examples show that the ADP has a superior ability to subdivide the image into integral groupings, minimizing the error in boundary localization and in pixel intensity. >

Conditions for positivity of an energy operator

Abstract: The authors present necessary and sufficient conditions such that the output from the Teager-Kaiser (1989) energy operator /spl lsqb/s/spl dot/(t)=ds(t)/dt/spl rsqb/ /spl Psi//sub c//spl lsqb/s(t)/spl rsqb/=s/spl dot//sup 2/(t)/spl minus/s(t)s/spl uml/(t) for continuous-time signals s(t) and the output from the corresponding discrete-time energy operator /spl Psi//sub d//spl lsqb/s(n)/spl rsqb/=s/sup 2/(n)/spl minus/s(n+1)s(n/spl minus/1) be non-negative everywhere. These operators have been shown to be effective for AM and FM demodulation in several useful classes of signals, such as speech and image signals. >

In‐vivo analysis of angiogenesis and revascularization of transplanted pancreatic islets using confocal microscopy

Abstract: A technique to measure angiogenesis and revascularization in pancreatic islets transplanted at the renal subcapsular site in the rat has been developed. In-vivo imaging of the microcirculation of transplanted pancreatic islets was conducted using a confocal scanning laser microscope (CSLM) to achieve optical sectioning through the graft in order to perform a computer reconstruction of the three-dimensional neovascular morphology. Individual islets were harvested by enzymatic digestion of excised pancreas from Fischer 344 rats. Isolated islets were cultured for 24 h, and approximately 300-350 islets were transplanted at the renal subcapsular site of the left kidney in an anaesthetized rat. Six to 14 days post-transplantation, the animal was anaesthetized and prepared for in-vivo imaging of the microvasculature on a Zeiss LSM-10. Optical contrast of the microvasculature was enhanced by the administration of fluorescein-labelled dextran into the circulating blood. The transplant site was identified and serial sections were obtained through the vascular bed at varying z-intervals. Complementary fluorescence video images were also obtained via a silicon intensifier tube camera mounted on the CSLM. At completion of the imaging procedure, the kidney was returned into the body cavity, the area was sutured and the animal was allowed to recuperate for subsequent examinations. Image processing algorithms, such as grey-level thresholding, median filtering, skeletonization and template matching, were applied to compute the vessel density and diameters and extrapolated to measure 3-D vessel lengths and the tortousity index of the neovasculature.

On the statistical optimality of locally monotonic regression

Abstract: Locally monotonic regression is a recently proposed technique for the deterministic smoothing of finite-length discrete signals under the smoothing criterion of local monotonicity. Locally monotonic regression falls within a general framework for the processing of signals that may be characterized in three ways: regressions are given by projections that are determined by semimetrics, the processed signals meet shape constraints that are defined at the local level, and the projections are optimal statistical estimates in the maximum likelihood sense. the authors explore the relationship between the geometric and deterministic concept of projection onto (generally nonconvex) sets and the statistical concept of likelihood, with the object of characterizing projections under the family of the p-semi-metrics as maximum likelihood estimates of signals contaminated with noise from a well-known family of exponential densities. >

Multi-component AM-FM image models and wavelet-based demodulation with component tracking

Abstract: In this paper we utilize multi-component AM-FM functions to model multi-partite nonstationary images that are locally coherent, yet globally wideband. We also detail an approach for simultaneously estimating the modulating functions associated with each of the multiple components. Components are isolated by a multiband bank of Gabor wavelets, and estimates of the modulating functions are derived from each channel. We use simple one-dimensional Kalman filters to track each identified component across the channels. >

Demodulation of images modeled by amplitude-frequency modulations using multidimensional energy separation

Abstract: Locally narrowband images can be modeled as 2D spatial AM-FM signals with several applications in image texture analysis and computer vision. In this paper we formulate such an image demodulation problem, and present a solution based on the multidimensional energy operator /spl Phi/(f)=/spl par//spl nabla/f/spl par//sup 2/-f/spl nabla//sup 2/f. We discuss some interesting properties of this multidimensional operator and develop multidimensional energy separation algorithms to estimate the amplitude envelope and instantaneous frequencies of 2D spatially-varying AM-FM signals. Experiments are also presented on applying this 2D energy demodulation algorithm to estimate the instantaneous amplitude contrast and spatial frequencies of image textures bandpass filtered via Gabor filters. The attractive features of the multidimensional energy operator and the 2D energy separation algorithm are their simplicity, efficiency, and ability to track instantaneously-varying spatial modulation patterns. >

Semi-automatic morphological measurements of 2-D and 3-D microvascular images

Abstract: The technique of serial optical sectioning by laser scanning confocal microscopy (LSCM), in conjunction with off-line digital image analysis, was used to quantize the morphological changes occurring during angiogenesis and revascularization of pancreatic islets transplanted at the renal subcapsular site in rats. The process consisted of in-vivo imaging of the microvasculature which was optically enhanced by the administration of a fluorescent probe into the circulating blood. Serial two-dimensional (2-D) optical sections were obtained through the vascular bed at varying z-intervals in order to perform a computer reconstruction of the complete three-dimensional (3-D) morphology. Image processing algorithms such as gray level thresholding, median filtering, skeletonization, region labeling and template matching were applied to compute the vessel density, lengths and diameters of the neovasculature, and the tortuosity index. >

Vergence control using a hierarchical image structure

Abstract: This paper presents a fast and reliable vergence control method using a hierarchical image structure for active vision systems. In this method, the sign pyramid is generated from the Laplacian pyramid. The search for the common fixation point and the disparity value of that point is performed by sign-correlation matching and a coarse-to-fine search strategy. This method works well for a very large range of disparities and is not sensitive to calibration problems common to stereo images. >

Motion-compensated visual pattern image sequence coding

Abstract: This paper presents an improved motion compensated version of the Visual Pattern Image Sequence Coding (VPISC) paradigm. It is a high performance video codec that is easily implemented in software. Software video codecs are not only cheaper, but are more flexible system solutions because they enable multi-vendor computers to exchange encoded video information without requiring on-board protocol compatible video codec hardware. The codec is intended for real-time desktop computer applications like multimedia delivery, and local area network (as well as point-to-point) televideo conferencing. We describe a version of motion compensated VPISC (MCVPISC) that has achieved bit rates of 0.025 bpp or better for MPEG test sequences (source coding has further reduced the bit rate by about half). The computational complexity of the encoder is less than 3 integer operations per pixel. The decoder is bounded between 0.016 and 0.125 logical and integer operations per pixel.© (1994) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Stereo vision using Gabor wavelets

Abstract: The analysis of video images in stereo can extend machine vision to interpret the 3-D structure of a scene. Applications of stereo vision include robotics, industrial automation, autonomous land rovers and automated cartography. The simplest stereo paradigm, binocular stereo vision, provides man and many animals the capability to see the depth from two images without ambiguity. Thus, it is interesting to study the biological solution to stereopsis. In this paper, a biologically motivated model of stereopsis based on a coarse-to-fine matching algorithm using multiband Gabor wavelets is presented. This approach generates a dense disparity map by phase difference computation between stereo image pairs without complex feature extraction. Results of the algorithm for both synthetic and natural stereo images are presented. >

Multiscale stereopsis via Gabor filter phase response

Abstract: The most difficult task in stereopsis is to resolve the correspondence problem from a pair of stereo 2-D images. By observing that the perspective projections of 3-D objects can be described as a phase shift issue, we solve the correspondence problem by employing the phase response via multichannel Gabor filters as matching primitive. Our stereo algorithm is biologically motivated and based on a multiscale, coarse-to-fine strategy. This approach produces a dense disparity map without complex feature extraction and interpolation. Results of the algorithm for both synthetic and natural stereo images are presented. >

Piecewise and local class models for image restoration

Abstract: In this paper, we present a new approach to image restoration based on a flexible constraint framework that encapsulates structural assumptions about the uncorrupted image. Piecewise and local class (PALC) models are defined and utilized to restore images degraded by linear blurring and additive noise. The restoration process is accomplished by iteratively deconvolving the solution image while simultaneously optimizing local image characteristics defined by the PALC models. Solution images to this ill-posed, combinatorial problem are computed using the novel generalized deterministic annealing (GDA) optimization technique. The results demonstrate high quality image restoration as measured by local feature integrity, improvement in signal-to-noise ratio, and reduction of restoration artifacts, especially in the presence of heavy-tailed additive noise. >

The Texas active vision testbed

Abstract: The Texas active vision testbed (TAVT) represents the most recent stage of an ongoing design process that weighed a large number of alternatives, for both hardware and software to create a flexible and precise tool for active vision research. In this paper, we describe this platform and our ongoing research employing this tool. The platform is used in the development of depth recovery algorithms to extend conventional active vision techniques, such as variable baseline stereo and active focus control, as well as support research in the active control of a vergent stereo geometry. >

Texas active vision testbed

Abstract: The Texas active vision testbed (TAVT) represents the most recent stage of an ongoing design process that weighed a large number of alternatives, for both hardware and software to create a flexible and precise tool for active vision research. In this paper, we describe this platform and our ongoing research employing this tool. The platform is used in the development of depth recovery algorithms to extend conventional active vision techniques, such as variable baseline stereo and active focus control, as well as support research in the active control of a vergent stereo geometry.<>

Image segmentation via piecewise constant regression

Abstract: We introduce a novel unsupervised image segmentation technique that is based on piecewise constant (PICO) regression. Given an input image, a PICO output image for a specified feature size (scale) is computed via nonlinear regression. The regression effectively provides the constant region segmentation of the input image that has a minimum deviation from the input image. PICO regression-based segmentation avoids the problems of region merging, poor localization, region boundary ambiguity, and region fragmentation. Additionally, our segmentation method is particularly well-suited for corrupted (noisy) input data. An application to segmentation and classification of remotely sensed imagery is provided.© (1994) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Multiresolution feature extraction based on multifrequency decomposition

Abstract: Edges and regions are important features for various computer vision and pattern recognition applications. We propose simple schemes for generating edge and region features based on multifrequency decomposition. These methods involve two steps. First, a multiresolution image structure is obtained by multifrequency decomposition. Second, multiresolution features are generated from this multiresolution image structure by some simple operations. Since these features have a hierarchical structure, we can use the coarse-to-fine strategy to improve the reliability and computational efficiency for matching and for other applications. >

DEMODULATION OF IMAGES MOD MODULATIONS USING MULTIDIlL ELED BY AMPLITUDE-FREQUENCY :ENSIONAL ENERGY SEPARATION

Abstract: Locally narrowband images can be modeled as 2D spatial AM-FM signals with several applications in image texture analysis and computer vision. In this paper we formulate such an image demodulation problem, and present a solution based on the multidimensional energy operator @(f) = llVf112 - fV2 f. We discuss some interesting properties of this multidimensional operator and develop multidimensional energy separation algorithms to estimate the amplitude envelope and instantaneous frequencies of 2D spatially-varying AM-FM signals. Experiments are also presented on applying this 2D energy demodulation algorithm to estimate the instantaneous amplitude contrast and spatial frequencies of image textures bandpass filtered via Gabor filters. The attractive features of the multidimensional energy operator and the 2D energy separation algorithm are their simplicity, efficiency, and ability to track instantaneously-varying spatial modulation patterns.