The Fingerprint Recognition Algorithm Based On Image Processing Research

In recent years, most systems are required to use reliable solutions to determine each person's identity, to ensure that only provide services to users on the system. Computer systems, mobile phones, bank ATM, houses, and personal email system need to verify the identity of the customer. If no reliable secure identification method, the system is easy to be deceived, unable to protect the rights and interests of clients. Due to the traditional identification using "user ID and password" way of recognition, gives the user memory burden, as a result of the ID or password forget, lost, the inconvenience to use. Fingerprint recognition technology can be a very good solution to this problem, fingerprint identification is based on a finger pointing at the people on the surface of the details of the distribution of feature points on pattern recognition.This paper mainly studies the fingerprint identification of three main parts, the first single pixel wide refining extract fingerprint diagram, and then extract the fingerprint feature points, finally adopt the way of the fingerprint feature point matching complete fingerprint identification work. Fingerprint image preprocessing work includes: separation of foreground and background of fingerprint image, the calculation of direction field, the calculation of gradient field, intelligent, fingerprint skeleton extraction, skeleton elaboration and refinement of post-processing, etc. The extraction of refined diagram, extraction of fingerprint feature points, including singular points and two probability higher endpoint and bifurcation point. Specific neighborhood judging method is extracted, the extract to deal with the fake fingerprints, based on the fingerprint, the type of feature points and feature points in the direction of the field around the distribution of the feature points of information vector. Fingerprint matching based on feature points matching method,respectively, for database template and fuzzy recognition fingerprint feature points matching, according to the threshold value to judge whether two fingerprint image is derived from the same fingerprint.In this paper,the algorithm is tested by MATLAB.Fingerprint is from Institute of Automation, after algorithm testing, in this paper, the automatic fingerprint recognition algorithm is adaptable for several typical low quality fingerprint, also for the fingerprints of different grain collector to collect.