An improved method of large angle palm recognition rate based on UNet depth prediction and projection transformation

Abstract

In the development of modern science and technology innovation, identity authentication technology in daily life and work in the field of application more and more, among which fingerprint recognition, facial recognition, voice recognition, vein recognition, iris recognition, etc., are not easy to forge the characteristics, so it is the main content of research scholars in various countries. Especially for palmar vein recognition technology, because this recognition technology has the characteristics of stability and uniqueness, the characteristic area of palmar vein is large, so more and more research topics are proposed. In the traditional sense of the hand vein recognition method, although has obtained the high accuracy, but need to manually during the recognition image design and gathering more features, need to study during the data preprocessing high quality hand vein image, so how to make use of artificial intelligence algorithm is optimized, are the major risks to the present study. In this paper, based on the understanding of the development status of palm vein recognition technology and the basic principle of Unet depth prediction and projection transformation, a palm vein recognition method based on feature fusion network is proposed. The final results show that compared with the traditional palm-vein recognition method, the proposed algorithm has stronger features, expression ability and generalization ability.