Image tracking in real-time: a transputer emulation of some early mammalian vision processes

By emulating some of the very 'early' vision processes believed to occur close to the photo-receptors of the mammalian retina, dramatic savings can be made in the computational efforts associated with some aspects of image processing. This paper describes the design and implementation of part of a low-cost visiual tracking system that realises the 'smooth pursuit' reflex. It uses no special signal processing, vector processing nor floating-point hardware. Its effectiveness relies only on the parallel replication of fast scaler integer processors with low latency communications. The current (unoptimized) implementation uses 10 T425 transputers, processes 544 x 544 byte-pixel images at camera frame rates and provides 100 Hz tracking feedback signals to the camera pan-and-tilt motors with a latency of no more than one-fith of a second. The system will track any object that occupies the (majority of the) centre of the field of view moving against any background - no special lighting conditions or artificial scenery are required. The design and implementation follow naturally parallel structrues and are expressed at all levels in occam.