Analysis of an optical target tracking system implementing a snake-based segmentation

Abstract

A complete target tracking processor including a segmentation method based on active contours or "snakes" has been optically implemented and its behavior has been evaluated. The purpose of this segmentation method is to darken the background in order to avoid its disruptive influence on tracking. The optical processor includes a Joint Transform correlator (JTC) with improved performance since with the proposed method the target will appear with a maximum of details in the reference. The complete processor is very complex and depends on the technology of several photonic devices. It consists of three nematic liquid crystal spatial light modulators, a preprocessor using an optical high pass filter, two different channels separated by the polarization direction of the light, a holographic edge enhancement filter, an electromechanical iris diaphragm, a camera and several photodetectors. The complete optical tracking processor must exhibit a negligible response time compared to a tracking cycle. To meet this requirement, research was conducted into two directions -- the algorithms and the components. A new snake optimization criterion is introduced and two different tracking algorithms are compared. The parameters of each component are studied and optimized, and their influences on the performance of the optical processor in terms of speed and tracking accuracy are analyzed. A complete system level demonstration of target tracking is presented as a conclusion.