Finger Vein Characteristic Extracted Based On WINCE

The rapid development of information age technology, security and confidentiality ofinformation is particularly important, compared with traditional identification techniques. Atpresent, the application of biological recognition technology has iris identification, facerecognition, fingerprint recognition, finger vein identification and other biological recognitiontechnology. Finger vein recognition as the second generation of the authentication technologyhas living identification, internal characteristics, at the same time has such of advantages:high security level, high precision, high speed, non-contact, has been used more and more.Modify camera.cpp file under Platform Builder5.0, making the S3C2440A previewchannel of the image size in pixel for output320*240. compile camera.cpp generationcamera.dll, and re-compile WINCE5.0mirror, and download file NK.bin to mini2440development board, and under in VS2005develop MFC smart device application, throughloading camera driver get BMP image which pixels is320*240.This paper analyzes the basis of the results of domestic and international finger veinrecognition technology, from local to get finger vein image preprocessing, through Robertedge detection to realize the background of the place, use the method of histogramequalization of image enhancement, use Gaussian filter for image filtering, finger veinimage segmentation adopt NiBlack method(where r=12, k=0.001), and image morphologicalprocessing to eliminate some noise, according to the area threshold method to achieve fillempty spots and large area of dark spots, the refinement of finger vein image use the methodof binarization refinement algorithm, and cut burr.Studied the finger vein recognition based on the vein structure, recognition technologybased on the geometric characteristics of the vein, vein recognition technology based onmoment invariants and finger vein recognition technology based on NMI(NormalizedMoment of Inertia), because the finger vein recognition technology has low complexity and itcan overcome the image acquisition angle offset, location, shift and other factors, has a goodrobustness, so in this paper select this method to achieve finger vein recognition based on theARM platform under WINCE5.0embedded systems, and gives the results of simulation in thevein image acquisition angle offset, position shift experiment, and in accordance with the selection of experimental results match the NMI difference threshold of0.01, the results showthat the threshold can accurately identify the finger vein image.