The Research Of Finger Vein Recgnition System

The biological recognition technology is a product of the combination of human biological characteristics and high-tech, which is a very active research field in recent years. The theory of finger vein recognition is that the finger is irradiated by a near-infrared light of certain wavelength to form a vein image, then after image processing to complete the similarity matching of two image. Because the vein recognition technology has the advantages of living identify, non-contact, internal characteristics and high recognition rate etc, it could be widely applied in the places of the entrance guard system, financial institutions and security entrance etc, which is called the biological recognition technology of second generation. In this paper, the finger vein recognition system is studied, including the finger vein image acquisition,image processing and image recognition, which is the concrete research content as follows:According to the theory of finger vein imaging, selected the appropriate equipment to build image acquisition device, and illustrated the hardware realization of acquisition device. Proposed a light source array of the equilateral triangle which uses the mathematical model of near infrared LED light and Sparrow`s criterion to optimize two light source distance, and obtained the optimal distance of uniform illumination. Through the simulation and experiment proved that the light source array can get high quality original finger vein image.Processed the finger vein image: First, the region of interest is extracted from the sampled finger vein image in order to remove the background region. Second,did normalized processing to unified image size and gray value. Third,proposed guided image filter algorithm to enhance image, and introduced the image enhancement model to obtain the prominent image of the vein texture structure. Forty, aimed at the existed problems of the classical threshold segmentation method, proposed a image segmentation method which is based on the detection of valley-shape direction to get a satisfactory segmentation effect. Fifth, dealt with the noise, including the burr noise,speckle noise and hollow noise. Finally, Hilditch serial thinning algorithm is employed to extract the vein skeleton, and got the finger vein texture image of single pixel, good connectivity, pseudo vein less, smooth.Studied Housdorff distance recognition algorithm which decomposes the overall collection into separate elements to analyzed, used the endpoint and branch point to represent the whole picture. Because relied too much on the feature points influenced by noise to result in low recognition rate and poor stability. Proposed Hu invariant moments algorithm which uses seven invariant moment for matching. Proved that Hu invariant moments algorithm has good resistance to degeneration, fault tolerance,accuracy through the experiments.Contrasted by Hausdorff distance recognition algorithm it has a higher stability and a relatively satisfactory recognition effect.