Abstract: 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.