A Study Of Hand Vein Image Recognition Method

As the new development trend of Biometric Identification, Hand Vein Recognition has been the hot field. Because of its uniqueness, reliability especially lower cost and friendly human-computer interactive, it will be proved to widely apply.In this thesis, we mainly make research on the image pre-processing, minutiae extracting and matching. We also verify the algorithm effectiveness based on the experimental platform structured by our Laboratory .Each procedure is studied deeply and mainly including the following aspect:In this thesis, according to the features of hand vein images, in order to accelerating the speed of the algorithm with highly accuracy, we directly process on the grey-image including image quality assessment, image segmentation, locally enforcement. we propose an approach based on vein ridgeline tracking by which to solve the problem caused by starting point sensitive, overall value of domain and gray unbalanced. By using the technique of locally value setting solving the problem about the character point selecting, at the same time by locally filtered image enforce solving the problem caused by allover value setting and non-equilibrium grey. The algorithm is carried into execution with high efficiency.In the stage of minutiae extraction, hand vein image is searched to find such minutiae as ridge bifurcations and endings, and their positions, local ridge directions of this features points, and minutiae directions are recorded as a vector. All vectors of minutiae from a vein image can form a vector matrix, which is just what we are interested in when matching. Before matching, such distortions as positions as position shift and rotation are adjusted.At last, we test our system using a samples library constructed according to age and gender with considerable hand back vein image and the results can be satisfactory.