基于DM642的双目视觉系统和角点检测算法研究

Research on binocular vision system and corner detection based on DM642 platform

本文分析了平行双目视觉测量技术中的角点检测算法,针对DM642嵌入式平台进行了相应改进.为了提高算法的简便性,减少人工干预,提出了自适应的Harris角点检测阈值设定算法,并针对DM642在算法实现时做了优化.经过实验,该方法可以准确地自动提取角点,比传统Harris角点检测算法更加适合嵌入式平台应用.

In this paper the corner detection algorithm in parallel binocular vision technology is analyzed. Improvements are made for its adaptation on DM642 embedded platform. To simplify the algorithm and reduce possible manual intervention, we propose an adaptive thresholdinging algorithm for Harris Corner detection so that it is more suitable for usage on embedded systems. Experiments clearly proves that our algorithm can detect corners accurately.