Reserch Of Palm Vein Recognition System Based On LabVIEW

Biometric recognition is a research focus in the field of identityauthentication. The vein recognition technology gets more and more attentionbecause of its higher security, reliability，compared with fingerprint identification.Vein identification technology recognition is based on hand vein image. The keytechnologies of identity authentication are involve vein image acquisition,preprocessing, feature extraction and pattern recognition. In this dissertation, thewhole process is studied, which include image acquisition system, imagepreprocessing and feature extraction based on the analysis of tissue biology. Themain works are listed as follows.1、 The optimization of the light source for image acquisition has beenresearched from theoretical and experimental. It was found that vein and othertissues has the maximum contrast in810nm.This showed that near infrared sourceof810nm wavelength is more suitable for the vein image acquisition. Imageacquisition system developed based on LED ring light source, CMOS camera and780nm longpass filter.2、Using the vein image acquisition system, image acquisition module wasdeveloped based on LabVIEW software platform. DLL files are called by DynamicLink Call Function to acquire experimental object’s hand vein image.3、In the vein image preprocessing, the original noise of the vein image wasfiltered through the median filter. Sobel operator was used to extract palm contour.The palm contour edge was used to locate the palm center point. An area of640*640pixels was selected as the ROI image based on the center point. Gamma correctionmethod was adopted because the extraction ROI image had not high light contrastand the gradation distribution is not uniform. This method can improve uniformity ofillumination and the contrast of the vein image.4、Splitting the vein image with NiBlack algorithm will lead to broken lines andother unordered lines. To solve this problem, a texture feature extraction method based on Gabor wavelet transform was proposed to extract vein characteristics.Fourier transform of the vein image was performed. And the result was multiplied bythe Fourier transform of the Gabor wavelet nuclear function. And inverse Fouriertransform of the product was performed. The mean and variance’s feature vectors ofthe transforming result was extracted, which were stored in the vein texture databaseas the characteristics of the vein image.