Research And Implementation Of Color Target Recognition And Tracking System Based On Robot

Vision system is the most important apparatus for human being and people acquire most of information from outside world by their vision systems. Computer vision was to use computer to simulate human eyes system. It could get information from image sequence, and recognize object around it. Color image is the best approach to the world, as there are more advantages in color image processing than in gray-level image.The mechanism on machine color vision as well as the relative theory and the application of machine vision in color target recognition are summarized in this thesis. Many problems on the machine recognition theory and the system implementation for segmentation are studied deeply and systematically. Some new methods for color segmenting, target recognition and tracking, shape extracting are proposed. The experimental system for color targets recognition and tracking based on the P3-DX mobile robot is built.The main content of this paper is as follows:1. It puts forward a histogram-based adaptive threshold method to segment and extract object from color image. The algorithm is built on the base of color image sub-way segmentation method. According to the hue and saturation, it can segment color image based on a histogram-based bi-threshold iterative. Meanwhile ,it captures some feature points to measure illumination variant, real-time update segmentation threshold. The method can improve the performance of color target segmentiation.2. A recognition and tracking algorithm adapted to the moving target is described. Based on the method, a tracking window method is adopted to track the target, and locate the position of target centroid by adjusting the window self-adaption. It proved that the algorithm could provid performance of fast and accurate.3. Color target recognition and tracking system are developed and based on P3-DXmobile robot. it can recognize and track color target real-time and stability Experimental results show that all the above algorithm is effective and fast operating.