Person recognition by hand shape based on skeleton of hand image

A new approach to the biometrical identification of a person based on the analysis of his hand shape is suggested in this paper. The shape is compared for reference and test hands. The reference hand is a preset model of the hand, while the test hand is the binary image obtained using a web-camera. In comparing the hans, the model’s adjustment to the test image is carried out for the best coincidence of their silhouettes. In contrast to the existing algorithms, the propsed method allows one to compare hands with partly occluded fingers, which are often observed under real conditions. The paper also describes a prototype of the system which allows real-time person recognition.