Research On Vein Feature Extraction And Matching Based On Radon And Wavelet Theory

Biological feature recognition is a kind of identity authentication technology, it carries out recognition according to human body’s inherent physiological features or the behavior features. The combination with the computer technology, make biological recognition technology widely used in a lot of fields. As a new method of biological feature recognition, vein recognition, although with a late start, has become one of the mainstream biometric identification methods after several years of development.Using biological features to authenticate identity, the key issue is the research of feature extraction and matching algorithms. Taking vein of human back of the hand is treated as the subject, a research of feature extraction and matching of vein based on Radon transform and wavelet theory is carried out.Firstly, pre-process methods, such as contrast-limited adaptive histogram equalization, method of threshold image and inflation and corrosion, are used to realize the vein feature extraction and matching work.Then, the vein images are projected onto the one-dimensional space by using Radon transform, and we calculat the direction of texture features which match the vein images coarsely. Then, the translation and the proportion of invariant wavelet transform coefficients can be got from the translation and the proportion of invariant adaptive wavelet transform; wavelet transform coefficients of projection data from different angles composite the characteristics matrix which are decomposited to obtain decomposition coefficients in multiple scales. The average energy is calculated though the decomposition coefficients on different scales, and constitute the feature vector. Experiments show that the method not only has the translation, scale and rotation invariance, and reflect the image at different scales of energy distribution. After the feature extraction, support vector machine can be used to match.Comparison with the Log-Polar showes that the method can be used to describe and recognite the texture feature.