Research on the Hand Image Biometric Feature Detection and Recognition

The hand gesture and action of the operator should be recognized and understood by detecting hand biological structure feature in the man-machine interactive assembly process. A new detecting method on hand contour feature detection and recognition is presented based on the skin color segmentation and hand image biological feature in the man-machine interactive assembly. The Otsu algorithm, maximum inter class variance algorithm, is optimized by using grayscale average variance instead of grayscale average. The hand image skin color segmentation is completed by detecting the skin similarity with adapt threshold. The finger shape fitting is completed with B-spline which combined the skin color information and hand image biological characteristics. And the fingertips, finger root and the palm are located by using convex hull detection and curvature criterion. At the same time, the finger local characteristics are enriched by combining with knuckles image detection. The hand image multi biometric features detection and recognition are realized effectively. It shows that the method is effective and feasible in experiment.