Research On Algorithm Of Finger Vein Image Recognition

Finger-vein recognition technology is a new kind of contact-less biometric identification technology, which assumes high reliability, convenient validation, high recognition accuracy and identification of living body and so on. Different from the traditional identification technology, it has reached a better balance between usability and reliability. But there are still some problems of the technology from image acquisition, preprocessing to feature extraction and matching. As a result the objective of this paper is to research and solve the aforesaid problems.This paper has analyzed the basic principle of the finger vein recognition technology, designed the procedure of finger vein image processing, and studied the principle and realization of image acquisition, zone location of vein image, extraction of finger vein pattern, extraction and matching of the feature of finger vein. The main research work is as follows.(1)Aiming at the image acquisition, in this paper we design the project of image acquisition of the finger vein, which based on analyzing various characteristics of the way of image acquisition of the finger vein and the standard of hardware configuration on the image acquisition. Verified by experiments, the vein image we get can meet the needs of subsequent research better.(2)Aiming at the zone location of vein image, in this paper we introduce the conception of ROI (Region of Interest), and put forward to a capture method of two-dimensional projection and positioning of finger vein image based on ROI region, which based on the research of the biological features and positioning method of finger vein image. Verified by experiments, this method can overcome the positioning problems caused by the feature of finger contour curve and achieve a content success of positioning.(3)Aiming at the extraction of finger-vein pattern, we adopt a comprehensive extraction method of finger-vein pattern which integrates processes of the detection of Valley-shape direction, enhancement of image fuzzy,segmentation of image threshold, noise removal according to the classification and image thinning, which based on the research on various extraction methods of finger-vein pattern. This paper analyzes each procedure in detail and determines the optimal threshold segmentation method by experiments to obtain a content experimental effect.(4)Aiming at the extraction and matching of the feature of finger vein, we analyze and summary up the problems of real-time and fault tolerance and anti-denaturing that feature extraction and matching algorithm have faced, and study and adopt the NMI (Normalized Moment of Inertia) feature in physics to puts forward to the algorithm of extraction and matching based on NMI features, which based on the research of the two classical algorithm. Verified by five experiments, the method can solve the aforesaid problems better to get a content experimental effect.(5)In this paper, we design and develop a system of finger vein recognition based on PC. By setting parameters and options, the system not only demonstrates the experimental results of sections of image processing in the paper, but also verifies the impact on the image processing effects of related parameters or algorithms, at the same time it also can realize the extraction and matching of characteristic value of finger vein.