Research On Recognition Algorithm Of The Human Finger Vein Image

In twenty-first Century, the issue of information security has attracted public attention, but the traditional identity recognition technology(such as personal identity card, driving license) cannot meet the requirements of information security, so biological recognition technology came into being. At present, Fingerprint recognition is the most widely used biometric technology. But with the continuous development of science and technology, the human fingerprint is not difficult to fake, and even a few yuan can be spent on the Taobao Internet to buy Silicone fingerprint film to clone fingerprint. This clone fingerprint can also be easy to deceive the fingerprint recognition instrument, which is a great security defect. So in view of the shortcomings of fingerprint recognition, its development has encountered a bottleneck. While the human finger vein due to its unique(the medical profession has been verified), growth in the flesh, not easy to forge etc. many advantages, has become a hotspot of the research of many scholars.This paper mainly focuses on the following points:(1)the principle of finger vein image acquisition is introduced in this paper: Because the deoxyhemoglobin in blood can absorb near-infrared light with a certain wavelength, the finger-veins can be imaged under the near-infrared light. In the image pre-processing stage, An improved algorithm based on ROI(region of interest) for two-dimensional projection and positionin of the finger vein image is proposed.It is proved that the algorithm can overcome the localization problem caused by the characteristic of the finger profile curve, and the result is satisfactory.(2)In this paper, several classic finger vein image segmentation algorithm Based on the mean method, Niblack method, OTSU method and iterative method, and carried out simulation experiments. The experimental results show that these algorithms can not extract the vein of high quality, the segmentation result is not very ideal, and there are a lot of noise and features. Therefore, an improved algorithm based on the combination of the maximum curvature and the valley shape segmentation is proposed, and gets a good segmentation result. Then, the filtering, denoising and thinning are performed.(3)In this paper, two kinds of key nodes in the skeleton image of the finger vein are extracted, which are the cross-point and the endpoint. Then the modified Hausdroff distance(MHD) is used to calculate their similarity. Experiments show that: using the two types of minutiae as feature point recognition at the same time, can be effectively used for personal recognition. The advantages, the level, the structure of the fusion algorithm and the selection of the mode source based on Multi-modal Biometrics' s Identify are also introduced in this paper. Finally, multi-modal biometric system construction method based on the fingerprint and finger vein is analysed.