Capacity Of Hand Vein Image And Separability

Dorsal Hand vein recognition is a biometric tecnology for identification by analyzing the image of the human dorsal hand vein. In recent years, social information security has been more and more emphasized, dorsal hand vein recognition as a new biometric identification methods has been developing rapidly. But there are still two problems restricting the study of the dorsal hand vein recognition, one of which is the study of hand vein recognition is still not widely used, although the hand vein recognition has been successfully used in instances of business, but there are still doubts about the feasibility of the biometric identification with is based on the dorsal hand vein, another problem is the lack of large-scale dorsal hand vein image library. In the current study of the dorsal hand vein recognition field, the tests are mostly carried out on a small data libruary.Therefore, this article focuses on these two problems, the main work is as follows:(1) This paper conducted some researches about the feasibility of the biometric identification with dorsal hand vein. Firstly, the basic features of the image, entropy, of the dorsal hand vein is compared with other biometric feathers (face, iris, fingerprint). The result prove that the dorsal hand vein images can contain enough information. Then, by using a coding method based on GLCM, the potential capacity of the dorsal hand vein is analyzed. The results showed that the dorsal hand vein has enormous potential capacity. The dorsal hand vein can be used in biometrics.(2) This paper introduces a novel synthesis method of DHV images using principal component analysis (PCA), which will be applied to enlarging the existing DHV image database to a large and new one. Based on the method of PCA, new data are synthesized with the feature extracted from the existing real data. The feature space grows up with the sample sets, the number of which is decided by our experiments. Extensive experiments show that the synthesized DHV databases can be very large. The experimental results show that the synthesized database has a good recognition rate, which indicates the proposed method performs well and would be applicable in the simulation test.