A Static Gesture Recognition Algorithm Based on DAG-SVMs

Abstract

In this paper, a static gesture recognition method with Kinect depth sensor to collect various gesture depth images was proposed. Based on the depth probability statistics of the original depth histogram, the neighborhood statistical parameters were added to construct the two-dimensional depth histogram so as to extract gesture images. In addition, the idea of integrating edge feature Hu moment and edge length moment as new features of static gesture is presented, and the improved DAG-SVMs algorithm is used to train gesture recognition so as to improve the recognition rate of the algorithm. Experimental results show that recognition rate of the recognition method achieves 97.96%