Study On Technology Of Palmvein Recognition

With the development of information society, information security has become more and more important, where identity recognition was an important issue. With the advancement of science and technology community, identity recognition device needed more convenient miniaturization, higher and higher level of security. Traditional authentication recognition technology has not kept pace of social development and has not met people's needs. Moreover, emerging biometric technology has been emerging. The palm vein recognition technology has been a higher level of security, better stability, with a living body identification feature and so on. This paper has studied the palm vein recognition of related technologies.In terms of image acquisition, this paper has designed a circular near-infrared LED array, to make the light uniformly and improve the quality of image. In addition, CCD camera has fitted with 850 nm filter to reduce visible interference. Human palm vein images have been acquired and image library has been composed for subsequent image experiments.In terms of image processing, this paper has designed a creative approach ROI(region of interest) interception and has identified critical points based on hand-type images to extract the area full of palm vein. This paper has proposed an enhanced processing algorithm and compared histogram equalization, Niblack algorithm and the modified algorithm. Moreover, it has normalized and enhanced the ROI image. Image pre-processing has achieved good results, for the subsequent feature extraction foundation.In terms of image feature extraction and matching, this paper has described the extraction algorithm based on Hu moments, one-way 2D-Gabor features and the latter has been improved. Then it has designed a two-stage recognition algorithm based on Hu moments and multi-directional 2D-Gabor. Image feature matching has been contrasted with Hamming distance. The method for calculating the Hamming distance has been modified to compensate for the palm key offset impact. Finally, it has conducted recognition experiments based on collected palm vein image library. Compared to the Hu moments recognition algorithm and multi-directional 2D-Gabor recognition algorithm, the proposed algorithm has performed better.