Hand Static Gesture Recognition Using HMM Based on Curvature Gradient of Occluding Contour

Today، the connection between human and computer is possible using mouse، keyboard and etc. These devices have some limitations like application speed. Making easy the interaction between human and computer is final objective of technology. In this paper، we propose one method for gesture recognition using curvature of B-Spline curves. First، the image of hand is extracted from different frames and some numbers of points are selected on the contour of hand in equal distances. Then، B-spline curves for groups of 4 points are calculated. Next، the slope of curvature for B-spline curves is calculated and used as feature vector of HMM classifier. The proposed method is inariant to rotation، movement and size of image because of using the slope of curvature. The results on 15 video sequences (with 150 frames in average) give recognition rate of 92. 21%.