Vitaly Kober

Accuracy of location measurement of a noisy target in a nonoverlapping background

Accuracy of target position estimation, defined as the variance of location errors, is evaluated when a noisy target is embedded on a nonoverlapping background. It is shown, with some assumptions, that the generalized matched filter minimizes this variance. We also investigate the performance of various correlation filters in terms of location accuracy. Computer simulations are made to compare the results obtained with the generalized matched filter with those of other filters.

Implementation of bipolar real-valued input scenes in a real-time optical correlator: application to color pattern recognition

Preprocessing of the input scene can be used to improve the performance of an optical correlator. The result of the preprocessing algorithms can be bipolar real-valued images, with positive and negative values, for input to the optical correlator. We present different procedures for the implementation of the bipolar real-valued signal with spatial light modulators that work in either amplitude-only or phase-only regimes. We apply these techniques to the implementation of bipolar realvalued preprocessed input images in color pattern recognition.

Nonlinear synthetic discriminant function filters for illumination-invariant pattern recognition

Novel nonlinear adaptive composite filters for illumination-invariant pattern recognition are presented. Pattern recognition is carried out with space-variant nonlinear correlation. The information about objects to be recognized, false objects, and a background to be rejected is utilized in an iterative training procedure to design a nonlinear adaptive correlation filter with a given discrimination capability. The designed filter during recognition process adapts its parameters to local statistics of the input image. Computer simulation results obtained with the proposed filters in nonuniformly illuminated test scenes are discussed and compared with those of linear composite correlation filters with respect to discrimination capability, robustness to input additive and impulsive noise, and tolerance to small geometric image distortions.

Design of correlation filters for recognition of linearly distorted objects in linearly degraded scenes.

Generalized correlation filters are proposed to improve recognition of a linearly distorted object embedded in a nonoverlapping background when the input scene is degraded with a linear system and additive noise. Several performance criteria defined for the nonoverlapping signal model are used for the design of filters. The derived filters take into account information about an object to be recognized, disjoint background, noise, and linear degradations of the target and the input scene. Computer simulation results obtained with the proposed filters are discussed and compared with those of various correlation filters in terms of discrimination capability, location errors, and tolerance to input noise.

COLOR PATTERN RECOGNITION WITH CIRCULAR COMPONENT WHITENING

Polychromatic object recognition based on circular whitening preprocessing of red-green-blue components and multichannel matched filtering is described. Computer simulations and experimental results are provided to facilitate recognizing a color target among objects of similar shape but with different color contents. Experimental results are obtained with an optical correlator with two spatial light modulators, one to introduce the scene and the second one to introduce the filter.

Adaptive composite filters for pattern recognition in linearly degraded and noisy scenes

New adaptive correlation filters for reliable recognition of geometrically distorted objects in blurred and noisy scenes are proposed. The filters are based on modified synthetic discriminant functions. The information about objects to be recognized, false objects, disjoint background, additive noise, and expected degradations of targets and input scenes are utilized in an iterative training algorithm. The algorithm is used to design a correlation filter with a specified discrimination capability. Computer simulation results obtained with the proposed adaptive filters in test scenes are discussed and compared with those of various correlation filters in terms of discrimination capability and location errors.

Optimal filter approximation by means of a phase-only filter with quantization.

Approximate filters based on a phase-only filter for reliable recognition of objects are proposed. Good light efficiency and discrimination capability close to that of the optimal filter can be obtained. Computer simulation results are presented and discussed.

Color component transformations for optical pattern recognition

Several elementwise component transformations performed over primary color image components (RGB) before optical multichannel correlations are proposed to improve real-time multispectral pattern recognition. The first transformation is deduced from the theory of the optimal filter for object location and recognition extended to multispectral images. Several modifications of this transformation are studied. We investigate these transformations in terms of noise robustness and discrimination capability. Computer simulation with noisy input images for various kinds of correlation filter are presented to illustrate improvement of color pattern recognition by using the proposed transformations. Experimental results are also presented.

Impulse noise removal with gradient adaptive neighborhoods

A new effective algorithm of impulse noise suppression is pro- posed. Conventional filtering schemes usually utilize a fixed shape for the moving window, such as a rectangle or circle. In contrast, the pro- posed algorithm exploits a signal-dependent shape for the moving win- dow. We suggest a simple adaptive algorithm of impulse noise detection in monochrome images that takes into account the size of signal gradient neighborhoods and image statistics. Experimental results show superior performance of the proposed algorithm compared to that of conventional algorithms in terms of both subjective and objective criteria.

Motion stereo based on adaptive correlation matching

A new algorithm to compute precise depth estimates for motion stereo is described. Input data is obtained from a single CCD camera and a moving belt. It is shown that the problem of matching among multiple motion stereo images can be effectively carried out by use of adaptive correlation matching. Experimental results with real stereo images are presented to demonstrate the performance of the algorithm.