基于欧氏骨架的手势识别系统

引进最新骨架提取算法,设计并实现了一种以手势的欧氏骨架为基准的手势识别系统,系统由通用视频采集模块和ARM开发板硬件组成.利用动态前景检测算法结合YCbCr肤色识别模型,分割出手势区域;借助欧氏距离变换和Delta—中轴骨架提取算法获得手势区域的欧氏骨架,并提取骨架的关键点和欧氏距离等几何参数,以此建立手势识别的几何模型.实验测试正确识别率高达94%,每帧图片处理时间小于25 ms,表明该系统实时、有效.

Gesture recognition system based on Euclidean skeleton

A gesture recognition system is developed based on the lasted skeleton extracting algorithm and Euclidean distance transform. The system consists of generic video acquisition module and ARM. Firstly,gestures region are segmented by dynamic forescene detection algorithm combined with skin color detection model of YCbCr;secondly,the Delta-medial axis skeleton extraction algorithm and Euclidean distance transform are applied to obtain Eudidean skeleton so as to extract geometric parameters,and key points on the skeleton. Then a gesture recognition model is built by these geometric parameters. Test results show that the overall recognition rate reaches 94%, processing time of each frame image is less than 25 ms,which shows that the system is real-time and effective.