Multi-passageway Information Processing Mechanism and Application in Vision Tracing

Inspired by principle of human vision system, the multi-passageway parallel processing mechanism was introduced and used in the research on robot vision tracing. After the analysis of human vision system, a new object tracking approach simulating human vision is presented in this paper. This method uses color, shape and moving information, which are three main features of moving objects in a multi-passageway parallel processing way, and the result is obtained by synthesizing the output information of each passageway. A new contour extraction method is proposed by the concept of recept field (RF), and the model of ´on´ RF is used in object´s shape information processing. Some tracking experiments using this approach have been done on a multifunctional experiment platform. Comparing with traditional serial processing mechanism, the results of tracking experiments showed that this method could improve the image processing speed and precision in a great degree