Svm-Based Fingerprint Recognition Technology

Now automatic fingerprint identification technology as a unique, invariant and reliability of biometric technology has been widely used. The current fingerprint identification technology is not very mature, need to address the question of how to improve the recognition rate and reduce the computation time.This paper studies the basic method of automatic fingerprint identification, focus on the fingerprint image preprocessing, feature extraction and minutiae's matching.Preprocessing focused on fingerprint image segmentation, the fingerprint pattern extraction, image enhancement, binarization and thinning. Experimental analysis by discussing them with the fingerprint image segmentation using gradient-based image segmentation algorithm, pattern extraction method based on gradient RAO extraction method, fingerprint image enhancement sub-image normalization, histogram equalization, image smoothing, frequency Gabor Four steps to complete filter, binarization process using a local adaptive threshold method, using a synthetic image refinement refinement refinement algorithm. Focused on feature extraction part of the minutiae feature points extraction and removal of false. Partial fingerprint matching fingerprint minutiae matching problem as a binary classification problem, first of all to support the model matching method used to strike the two fingerprint minutiae matching image collection, and structure matching feature vectors, and then samples of the training database match vector obtained using support vector machine, and concluding with a test sample library classifier trained on the simulation, experimental results prove the validity and reliability.