Optical correlator based target detection, recognition, classification, and tracking

Abstract

A dedicated automatic target recognition and tracking optical correlator (OC) system using advanced processing technology has been developed. Rapidly cycling data-cubes with size, shape, and orientation are employed with software algorithms to isolate correlation peaks and enable tracking of targets in maritime environments with future track prediction. The method has been found superior to employing maximum average correlation height filters for which the correlation peak intensity drops off in proportion to the number of training images. The physical dimensions of the OC system may be reduced to as small as 2 in. × 2 in. × 3 in. (51 mm × 51 mm × 76 mm) by modifying and minimizing the OC components.

Pattern recognition with composite correlation filters designed with multi-objective combinatorial optimization

“…In this approach, the coordinates of the maximum of the filter output are taken as estimates of the target coordinates in the observed scene. Correlation filters possess a good mathematical basis and they can be implemented by exploiting massive parallelism either in hybrid opto-digital correlators [11,12] or in high-performance hardware such as graphics processing units (GPUs) [13] or field programmable gate arrays (FPGAs) [14] at high rate. Additionally, these filters are capable to reliably recognize a target in highly cluttered and noisy environments [8,15,16].…”

Image registration under translation and rotation in two-dimensional planes using Fourier slice theorem

“…The problem of translation-and rotation-invariant recognition of an object via optical processor has been studied for a long period of time and several techniques have been proposed [1][2][3][4][5][6][7][8][9][10]. Optical correlation is shift-invariant and therefore can be advantageously used for such problems [11].…”

Object Tracking in Nonuniform Illumination Using Space-Variant Correlation Filters

“…An attractive option for target recognition is given by correlation filtering. Correlation filters possess a good mathematical background, and they can be implemented by exploiting massive parallelism either in hybrid opto-digital correlators [2,3] or in digital hardware such as graphic processing units (GPU) [4]. A correlation filter is a linear system where the coordinates of the system output maximum are estimates of the target coordinates in the observed scene [5,6].…”

“…In this sense, the problem of target tracking can be addressed by applying correlation filters to multiple frames. Recently, several proposals have been suggested to perform target tracking with the help of correlation filters [3]. In this work, we propose a reliable system for recognition and tracking of a moving target in nonuniformly illuminated scenes using a filter bank of space-variant correlation filters.…”