为满足人机交互系统中手势识别与理解的需要,提出了一种基于TMS320DM642的实时手势识别系统。整个系统由手势图像采集、手势图像处理和匹配识别组成。实验结果表明,该系统通过手势图像预处理、肤色分割、特征提取及识别等算法的实现,能正确识别待测的手势图像,其速度可以满足后端处理的需要,可为下一步人机交互系统中手势的理解和识别打下良好的基础。

To satisfy the demands of hand gesture recognition in human-computer interaction system, a design method of real-time gesture recognition system based on TMS320DM642 is proposed. The system is composed by these function module groups:the gesture image acquisition modules and the gesture image processing and recognition modules. The experimental results show that the system works frequency is fast enough to realize the processing and recognition function via hand gesture pretreatment, skin segmentation, feature extraction and recognition. It is appropriate for the hand gesture recognition in human-computer interaction system.