Showing papers by "Alan C. Bovik published in 1993"

AM-FM energy detection and separation in noise using multiband energy operators

Abstract: This paper develops a multiband or wavelet approach for capturing the AM-FM components of modulated signals immersed in noise. The technique utilizes the recently-popularized nonlinear energy operator Psi (s)=(s)/sup 2/-ss to isolate the AM-FM energy, and an energy separation algorithm (ESA) to extract the instantaneous amplitudes and frequencies. It is demonstrated that the performance of the energy operator/ESA approach is vastly improved if the signal is first filtered through a bank of bandpass filters, and at each instant analyzed (via Psi and the ESA) using the dominant local channel response. Moreover, it is found that uniform (worst-case) performance across the frequency spectrum is attained by using a constant-Q, or multiscale wavelet-like filter bank. The elementary stochastic properties of Psi and of the ESA are developed first. The performance of Psi and the ESA when applied to bandpass filtered versions of an AM-FM signal-plus-noise combination is then analyzed. The predicted performance is greatly improved by filtering, if the local signal frequencies occur in-band. These observations motivate the multiband energy operator and ESA approach, ensuring the in-band analysis of local AM-PM energy. In particular, the multi-bands must have the constant-Q or wavelet scaling property to ensure uniform performance across bands. The theoretical predictions and the simulation results indicate that improved practical strategies are feasible for tracking and identifying AM-FM components in signals possessing pattern coherencies manifested as local concentrations of frequencies. >

Locally monotonic regression

Abstract: The concept of local monotonicity appears in the study of the set of root signals of the median filter and provides a measure of the smoothness of the signal. The median filter is a suboptimal smoother under this measure of smoothness, since a filter pass does necessarily yield a locally monotonic output; even if a locally monotonic output does result, there is no guarantee that it will possess other desirable properties such as optimal similarity to the original signal. Locally monotonic regression is a technique for the optimal smoothing of finite-length discrete real signals under such a criterion. A theoretical framework in which the existence of locally monotonic regression is proved and algorithms for their computation are given. Regression is considered as an approximation problem in R/sub n/, the criterion of approximation is derived from a semimetric, and the approximating set is the collection of signals sharing the property of being locally monotonic. >

Visual pattern image sequence coding

Abstract: Visual pattern image sequence coding (VPISC) is a new digital image sequence (video) coding process that possesses significant advantages relative to other technologies; in particular, it is extremely efficient in terms of the computational effort required. It is designed to exploit properties of the human visual system (HVS), and thus yields high visual fidelity. Visual quality criteria are deliberately chosen over information-theoretic ones on the grounds that, in images intended for human viewing, visual criteria are the most meaningful ones. VPISC yields impressive compression comparable to other recent methods, such as motion-compensated vector quantization. VPISC divides the images into spatiotemporal cubes, which are then independently matched with one of a small, predetermined set of visually meaningful three-dimensional space-time patterns. The pattern set is chosen to conform to specific characteristics of the HVS. Also introduced are two modifications of VPISC: adaptive and foveal VPISC. These are spatiotemporally nonuniform implementations that code different portions of the image sequence at different resolutions, according to either a fidelity criterion (for AVPISC) or a foveation criterion (for FVPISC). >

Theorems for discrete filtered modulated signals

Abstract: Useful approximations to the responses of discrete linear systems and certain discrete nonlinear systems are developed for input complex AM-FM signals of the form s(m)=a(m) exp (j phi (m)). These are used to derive limits on simple AM-FM demodulation mechanisms related to the Teager-Kaiser operator. >

Three‐dimensional distribution of damaged cells in cryopreserved pancreatic islets as determined by laser scanning confocal microscopy

Abstract: The technique of serial optical sectioning by confocal microscopy, in conjunction with off-line digital image analysis, was used to quantify the radial distribution of damaged cells in rat pancreatic islets following cryopreservation. The process consists of imaging frozen-thawed islets of Langerhans using laser scanning confocal microscopy (LSCM). The three-dimensional (3-D) distribution and analysis of the two populations of viable and damaged cells was visualized via acridine orange/propidium iodide (AO/PI) fluorescent staining. In preparation for cryopreservation, isolated and cultured rat pancreatic islets were brought to a 2 M concentration of dimethyl sulphoxide (DMSO) by serial addition at decreasing temperatures. Ice was nucleated in the islet suspension at -10 degrees C, and individual specimens were frozen to -70 degrees C at cooling rates of 1, 3, 10 and 30 degrees C/min in a programmable bulk freezer and subsequently stored in liquid nitrogen. After rapid thawing and serial dilution to remove DMSO, individual islets were prepared with AO/PI stains for imaging on the LSCM. Serial sections of the islets, 2-7 microns in thickness, were obtained and processed to obtain high-contrast images. Analysis algorithms consisted of template masking, grey-level thresholding, median filtering and 3-D blob colouring. The radial distribution of damaged cells in the islets was determined by isolating the cell and computing its distance from the centroid of the 3-D islet volume. An increase in the number of blobs corresponding to single and/or aggregates of damaged cells was observed progressively with distance from the centre towards the periphery of the islet. This pattern of freeze-induced killing of cells within the islet was found to occur consistently in the numerous individual specimens processed.

Nonlinear regression for image enhancement via generalized deterministic annealing

Abstract: We introduce new classes of image enhancement techniques that are based on optimizing local characteristics of theimage. Using a new optimization technique for nonconvex combinatorial optimization problems, Generalized DeterministicAnnealing (GDA), we computefuzzy nonlinear regressions of noisy images with respect to characteristic image sets definedby certain local image models. The image enhancement results demonstrate the powerful approach of nonlinear regression andthe low-cost, high-quality optimization of GDA. 1. INTRODUCTION According to [7], the goal of image enhancement is "to process a given image so that the result is more suitable than the original (input) image for a specific application." Global approaches to the image enhancement problem, such as histogram modification, linear filtering, and Fourier transform modification, often fail to preserve important local features inthe image. Local methods, based on constrained optimization, attempt to preserve an image property, such as continuity,while at the same time producing an output image that is "close" to the input image [6,9]. Unfortunately, the optimizationmethods suffer two main drawbacks. First, the methods do not provide a flexible mathematical framework for different imagemodels needed for different applications. Secondly, the computational cost of obtaining the image enhancement results usingconstrained combinatorial optimization is impractical for time-dependent image processing. In this paper, an approach toimage enhancement is described that exploits several image models based on well-defmed local image characteristics. Also,using generalized deterministic annealing (GDA), high quality image enhancement results are obtained inexpensively.In this paper, we first define fuzzy nonlinear regression and then introduce the regression models employed for imageenhancement. A brief discussion of the optimization algorithm used, GDA, is given, and a non-heuristic implementation isprovided. Finally, we give several examples that demonstrate image enhancement using nonlinear regression via GDA.

Analysis of volumetric changes in rat pancreatic islets under osmotic stress using laser scanning confocal microscopy.

Abstract: Analysis of the volumetric changes in rat pancreatic islets undergoing shrinkage and/or swelling due to osmotic stress is essential for understanding the mechanism of mass transport between cells and their environment and for optimizing cryopreservation protocols. Addition and removal of cryoprotective additives is an integral component of all cryopreservation processes. We have used laser scanning confocal microscopy (LSCM) of acridine orange/propidium iodide (AO/PI) stained Islets of Langerhans to analyze the effects of osmotic stress induced by exposure to varying concentrations of the cryoprotectant dimethyl sulfoxide (DMSO), on the islet volume at two temperatures 23 degrees C and 15 degrees C. Experiments were conducted by mounting a single islet onto a unique freeze-thaw-perfusion stage on which the system temperature and the chemical composition of the solutions can be precisely controlled. The bathing medium of the islet was rapidly changed from isotonic saline to the desired DMSO osmolality to produce a defined osmotic stress, and the islet was imaged simultaneously using 488 nm argon laser. Three to seven serial sections were obtained through each islet at increments varying between 15 microns and 20 microns. The three-dimensional (3-D) image was segmented into islet and non-islet regions using a combination of median filtering, gray level thresholding and region labeling, and the islet volume was computed by counting voxels. Further, a special analysis algorithm was applied to identify shape changes both locally and globally throughout the islet volume.

Audio-visual speech recognition for a vowel discrimination task

Abstract: Dept. of Electrical and Computer Engineering Dept. of Electrical and Computer EngineeringOld Dominion University The University of Texas at AustinNorfolk, VA 23529 Austin, TX 78712ABSTRACTAmong the various methods which have been proposed to improve the robustness and accuracy of AutomaticSpeech Recognition (ASR) systems, lipreading has received very little attention. In this paper, we providemotivation for the use of lipreading. A novel speaker dependent lipreading system is developed, which useshidden Markov modeling, a well known and highly successful technique for audio-based ASR. It is used

Selective laser sintering for the creation of solid models from 3D microscopic images.

Abstract: An exciting application of using Selective Laser Sintering (SLS) to produce solid models of microscopic specimens imaged with a laser scanning confocal microscope is presented. The SLS model, fabricated by sintering together successive layers of a fine powder with a computer controlled scanning laser, is a true 3-D magnification of the microscopic sample. The 3-D models produced are accurate representations of the data and can be handled and manipulated to evaluate surface details and morphology. The models give an extremely powerful method for 3-D data visualization and tactilization. Three dimensional digital image processing is performed on microscopic images to prepare the data for the selective laser sintering process. The image processing techniques for preparing 3-D images of both translucent and opaque specimens for laser sintering are presented. For translucent specimens, the image processing removes noise and "fills in" inclusions and cavities within the specimen data in order to produce a structurally sound model. When imaging an opaque specimen an image of the upper surface alone is acquired. Image processing is used to remove noise and to Fill the volume under the specimen surface. Sample models of a dandelion (Taraxicum officinale Compositae) pollen grain and the surface of a U.S. penny are presented. These models present examples of both the translucent and opaque data types.

Fabrication of macroscopic solid models of three-dimensional microscopic data by selective laser sintering

Abstract: SUMMARY We discuss and give examples of the use of selective laser sintering to fabricate solid macroscopic models of microscopic specimens that have been imaged with a confocal microscope. The digital image processing necessary to create structurally sound models of both translucent and opaque specimens is presented. The fabricated models offer the ultimate in data visualization since they can be physically handled and manipulated to investigate the shape and features of the specimen. Such a powerful visualization tool is useful in both research and educational environments.

Fast combinatorial optimization using generalized deterministic annealing

Abstract: Generalized Deterministic Annealing (GDA) is a useful new tool for computing fast multi-state combinatorial optimization of difficult non-convex problems. By estimating the stationary distribution of simulated annealing (SA), GDA yields equivalent solutions to practical SA algorithms while providing a significant speed improvement. Using the standard GDA, the computational time of SA may be reduced by an order of magnitude, and, with a new implementation improvement, Windowed GDA, the time improvements reach two orders of magnitude with a trivial compromise in solution quality. The fast optimization of GDA has enabled expeditious computation of complex nonlinear image enhancement paradigms, such as the Piecewise Constant (PICO) regression examples used in this paper. To validate our analytical results, we apply GDA to the PICO regression problem and compare the results to other optimization methods. Several full image examples are provided that show successful PICO image enhancement using GDA in the presence of both Laplacian and Gaussian additive noise.

Demodulation of AM-FM Signals in Noise Using Multiband Energy Operators

Abstract: with time-varying amplitude a and instantaneous frequency o i = 4 , using the operator Y(s) = ($2-ss' developed by Teager [I] and Kaiser [2], shown to be highly effective for detecting AM-FM modulations [31. For signals of the form (l), "(3) 5 u 2 q 2 and Y(i = a 2 0 t , with small approximation error under realistic conditions [31. This motivates the energy separaiwn algorithm @SA): SO) = a(t) cos[~i ) l , (1)

Shape change analysis and shape modeling using three dimensional biomedical images

Abstract: A method for computing material deformations from multidimensional images of a biological specimen undergoing shape changes is presented. The specimen data are segmented within the images and the domain of the specimen is parametrized. The parametrization is a material coordinate system for the specimen. Deformations of the material coordinate system are computed by minimizing an energy functional that is a linear combination of a brightness continuity constraint and a shape-change constraint based on differential geometric properties of the parametrization. The resulting parametrization completely describes the position and shape-change of the specimen for the entire image sequence. This approach was applied to images of a motile human white blood cell. >