摘要
设计了一种基于智能摄像机的分布式并行处理的高速视觉系统,并应用该系统跟踪快速运动的乒乓球;同时提出了一套针对灰度图像的快速、有效的乒乓球识别跟踪算法.该算法能够在变化的背景下跟踪运动的乒乓球,具有较小的运算量及较好的鲁棒性.为了进一步提高视觉系统的实时性,还运用了动态窗口.实验结果表明,本系统能够快速识别跟踪运动的乒乓球,对于640×480像素的8位灰度图像,采集与处理速度最快可以达到6ms.
A high-speed vision system based on smart camera and distributed parallel processing is designed and applied to tracking the fast-moving ping pong ball. At the same time, a set of quick and efficient target recognition and tracking algorithm based on grey image is proposed. This algorithm is applied to distinguishing the moving ping pong ball from the changing background with less computation and good robustness, In order to improve the real-time performance further, dynamic window is used. The experimental results demonstrate that the system can distinguish and track a moving ping pong ball quickly. It takes only 6 ms to capture and process a grey frame with a size of 640×480 pixels.
出处
《机器人》
EI
CSCD
北大核心
2009年第3期229-234,241,共7页
Robot
基金
国家自然科学基金资助项目(60672039)
关键词
高速视觉系统
筛选框
动态窗口
high-speed vision system
search region
dynamic window
