一种面向移动终端的WEB微手势识别

摘要

WEB applications mobile terminal micro-gesture recognition,a method is proposed to adapt to micro-gesture rec?ognition WEB mobile terminal platform. Haar features is used to recognize the palm of your hand position,which can identify by skin color segmentation,morphology processing methods for the hand image,and then according to the information analysis of the current gesture of palms and fingers represent commands,thus it can response corresponding operation behavior. A Web framework of RTC is used to collect 200 groups of color of skin different,different sizes,hand gesture image,1000 micro gesture recognition experiments are conducted. The experimental results show that the gesture recognition success rate is more than 90%,the average time is less than 180 ms,and it can identify dynamic gesture instructions on Android devices.%针对移动WEB终端微手势识别的应用需求,提出了一种适应移动WEB终端平台的微手势识别方法.应用Haar特征对手掌位置进行识别,通过肤色分割、形态学处理等方法获取手部图像,然后根据手掌和手指的信息分析获得当前手势所代表的命令,从而响应相应的操作行为.采用Web RTC框架采集了200组肤色不同、大小不同、手形不同的手势图像,进行1000次微手势识别实验.实验结果表明,该方法的手势图像识别成功率大于90%,平均耗时小于180ms,并且可以在Android设备上识别动态手势指令.