HIGH SPEED GESTUTURE RECOGNITION SYSTEM USING MASSIVELY PARARLLEL PROCESSING VISION

本研究的目的是开发一个高速手势识别系统。为了实现这一目标，我们做了以下工作：（1）设计并研制了人机接口视觉芯片系统;（2）对高速视频图像进行了分析，研究了各种人体动作;（3）系统化了高速视觉的图像处理和手势识别算法;（4）在系统上对算法进行了评估，并实现了应用算法。2000年我们研制了视觉芯片系统，并在系统上实现了算法，用一个主动视觉系统实现了二维跟踪。2001年，我们利用视觉芯片系统开发了立体系统，实现了三维人体姿态跟踪，并系统化了高速姿态识别的图像处理算法。

The aim of this research is to develop a high speed gesture recognition system. To realize the aim, the followingworks were done :(1) to design and develop a vision chip system for human interface,(2) to analyze high speed video, images to various human actions,(3) to systemize image processing and gesture recognition algorithm for high speed vision,(4) to evaluate the algorithms on the system, and implement application algorithms.In 1999 we designed a vision chip system for human interface, analyzed high speed video images, and proposed gesture recognition algorithms for high speed vision. In 2000 we developed the vision chip system, implemented the algorithmon the system, and realized 2D trackingusing one active vision system. In 2001 we developed stereo system using vision chip systems, realized 3D human gesture tracking using the stereo system, and systemize the image processing algorithm for high speed gesture recognition.