基于肤色分割与边缘检测的手势识别技术研究

Because the skin-color has a good character of clustering in the Ycbcr color space, a algorithm for gesture segmentation which combines the information of skin-color and edge is presented.Firstly, the area of hand gesture is segmented from the complex background in the Ycbcr color space based on skin-color,at the same time, getting the edge of the gesture by detecting the image,and then ombining these with and-operation to remove the area which looks like the skin in background to get the binary Image of gesture.Then, use the normalized Fourier descriptors of the contour curve to extracting the feature. Finally,BP neural network was adopted as the classifier to recognize hand gestures. The results showed that, for complex gestures background image, the algorithm performs better, and is of low computational complexity, fast computing speed,high recognition rate and real-time.