Izidor Gertner

The discrete Zak transform application to time-frequency analysis and synthesis of nonstationary signals

Algorithms to compute coefficients of the finite double sum expansion of time-varying nonstationary signals and to synthesize them from a finite number of expansion coefficients are presented. The algorithms are based on the computation of the discrete Zak transform (DZT). Fast algorithms to compute DZT are presented. Modifications to the algorithms which increase robustness are given. >

Multi-sensor fusion: an Evolutionary algorithm approach

Modern decision-making processes rely on data coming from different sources. Intelligent integration and fusion of information from distributed multi-source, multi-sensor network requires an optimization-centered approach. Traditional optimization techniques often fail to meet the demands and challenges of highly dynamic and volatile information flow. New methods are required, which are capable of fully automated adjustment and self-adaptation to fluctuating inputs and tasks. One such method is Evolutionary algorithms (EA), a generic, flexible, and versatile framework for solving complex problems of global optimization and search in real world applications. The evolutionary approach provides a valuable alternative to traditional methods used in information fusion, due to its inherent parallel nature and its ability to deal with difficult problems. However, the application of the algorithm to a particular problem is often more an art than science. Choosing the right model and parameters requires an in-depth understanding of the morphological development of the algorithm, as well as its recent advances and trends. This paper attempts to give a compact overview of both basic and advanced concepts, models, and variants of Evolutionary algorithms in various implementations and applications particularly those in information fusion. We have brought together material scattered throughout numerous books, journal papers, and conference proceedings. Strong emphasis is made on the practical aspects of the EA implementation, including specific and detailed recommendations drawn from these various sources. However, the practical aspects are discussed from the standpoint of concepts and models, rather than from applications in specific problem domains, which emphasize the generality of the provided recommendations across different applications including information fusion.

A parallel algorithm for 2-D DFT computation with no interprocessor communication

A parallel algorithm is proposed for the two-dimensional discrete Fourier transform (2-D DFT) computation which eliminates interprocessor communications and uses only O(N) processors. The mapping of the algorithm onto architectures with broadcast and report capabilities is discussed. Expressions are obtained for estimating the speed performance on these machines as a function of the size N*N of the 2-D DFT, the bandwidth of the communications channel, the time for an addition, the time T(F/sub N/) for a single processing element to perform an N-point DFT, and the degree of parallelism. For single I/O channel machines that are capable of exploiting the full degree of parallelism of the algorithm, attainable execution times are as low as the time T(F/sub N/) plus the I/O time for data upload and download. An implementation on a binary tree computer is discussed. >

Reflective-memory multiprocessor

Reflective memory may be thought of as hardware-supported replication of data on multiple computers. This simple mechanism has been extended further, in software, to support memory semantics required by multiprocessors. The paper describes the reflective memory implementation of the Encore Infinity-a distributed shared memory multiprocessor. A brief overview the system is provided and comparisons are made with other cached architectures. Spinlocks are employed in this system to manage the herculean cache coherency problems which are a natural result of any system which employs massive replication. A brief introduction to the recovery procedure used to return crashed nodes to normal operation is also described.<<ETX>>

Image representation using Hermite functions

We present a mathematical technique for analyzing images based on two-dimensional Hermite functions that are translated in both space and spatial frequency. Although the translated functions are not orthogonal, they do constitute a frame and hence can be used for image expansion. The technique has the practical advantage that fast algorithms based on the Zak transform (ZT) can be used to compute expansion coefficients. We describe properties of the ZT that are relevant to image representation and which allow us to use it both to compute expansion coefficients efficiently and to reconstruct images from them. Finally, we use a Hermite function frame to decompose and reconstruct a texture image.

A novel motion-based active video indexing method

In this paper, a novel motion-based active video indexing scheme is proposed for videos acquired from active visual systems. With the foveate wavelet transform as the representation of active videos, the description for the object of primary interest during smooth pursuit can be generated in the foveal area based on the behavior of the motion field estimated by a two-pass motion estimation MRF-MAP algorithm. Afterwards, a vector of 36 numbers is generated as the index for the object. Furthermore, the attention shift camera motion can be detected by use of this motion field, which is employed as another important indicator of the video contents. The efficacy of this active object-based video indexing scheme is demonstrated through extensive experiments.

Gradient-based genetic algorithms in image registration

Two modifications of Genetic algorithm (GA) are proposed that employ gradient analysis of the fitness function and are integrated with the main genetic procedure. Combination of the relative weighted error factor and adaptive size of the mutation pool accelerates convergence of the iterative process and indicates When the global optimum um solution is found. Local gradient correction of the initial pool during interactions refines the search procedure. Computational experiments show that both modifications can increase efficiency of GA when they are applied to an image registration problem.

Zak-Gabor representation of images

A mathematical approach to image analysis and synthesis in the combined frequency- position space is presented. The formalism is based on the Weil map - Zak transform, which provides one of the most fundamental tools for studies of nonstationary processes and images as such. Algorithms suitable for computation of Gabor expansion coefficients are presented.

Discrimination, tracking, and recognition of small and fast moving objects

In military applications it is of utmost importance to detect and identify targets from Infra-Red (IR) sequences. Usually the targets of interest are of small size and moving with great velocity. In addition, the IR sequences are extremely noisy due to rampant systemic noise incurred by the sensing instrument and the noise from the environment. In this paper, we developed a method which can effectively detect and identify the targets of interest form the noisy IR sequences using the manipulation of both the temporal, spectral, and spatial bandpass filtering. With this method, first a bandpass filtering in the spectral domain is conducted in order to remove noises, especially the systemic ones. Next the candidate target locations are declared through a spectral bandpassing from the temporal pixel process, where by taking advantage of the fact that the targets of interest are fast moving the background and random noises are largely removed. The estimated targets of interest for each IR frame are further refined after a post- processing in the spatial domain. The final targets of interest are then declared after a consistency check-up across adjacent IR frames by use of an adaptive Hough transform. Experimental results based on this proposed method have suggested desirable performance.

Reducing the computational cost of local search in the hybrid evolutionary algorithm with application to electronic imaging

This article focuses on the efficiency problems associated with the use of local search in the hybrid evolutionary algorithm. A two-phase cyclic local search is proposed that alternates the random search and the downhill simplex method (DSM), and helps prevent the algorithm from converging to a sub-optimal solution in multidimensional optimization. The algorithm utilizes a novel micro-model of image local response, in order to reduce the number of fitness evaluations during the local DSM search, with the application to the global optimization problem arising in electronic imaging. The problem is stated as the search for the feasible transformation parameters that minimize the difference between two images. Image local response is defined as the variation of the fitness function that occurs because of a small variation of the parameters, and is computed over a small pixel area. The computed response coefficients specify a contraction transformation applied to the vector of the regular DSM coefficients that control the movement and the shape of the simplex. The transformation adjusts the length of the vector, making the step size of the simplex adaptive to the local properties of the fitness landscape. The computational experiments with two-dimensional grayscale images provide the experimental support and justification of the analytical model of image local response and its utilization for the reduction of the computational cost of local search, without the loss of the quality of the final solution.