Researches On Fingerprint Identification Related Algorithms

Along with the development of the computer image processing and pattern recognition technology, biometrics identification teconology, the method to determine the identity of a person using his unique physiological features has been developed rapidly. Fingerprint, as a unique, stable, easy to sample and highly reliable physiological feature of human being, has become the most effective means on personal identity, and is the only biological characteristics admissible in court too. Currently, the development of the automatic fingerprint identification system algorithm is relatively mature, but the low-quality fingerprint identification technology is still unsatisfactory. In this paper, some in-depth researches on the key areas of fingerprint recognition technology have been conducted and the following main work and innovations are executed:1. An improved ZS method is proposedA critical and Indispensable step of an Automatic Fingerprint Identification System is fingerprint image thinning. Many algorithms have been devised and applied trying to obtain one pixel wide and8neighbors connected thinning results. The ZS algorithm is a parallel thinning algorithm which has fast converge speed, but it also has some problems such as distortion of the fingerprint ridge, incapability of gaining one-pixel wide skeleton. In this paper, a method of improving the thinning result through adding eliminate structure templates is proposed. The algorithm will solve the problems without extra time cost. as a result, the more effective fingerprint minutiae points could be extracted. 2. A Gabor filter based fingerprint image enhancement algorithmon is proposedMany fingerprint images are with poor quality, according to statistics, about10%of the fingerprint images are low quality images. In order to make sure that the system could process the poor quality fingerprint images, the image enhancement step is necessary. Gabor wavelet filter has some good features which makes it a good image enhancement tools. In this paper, through making good use of some properties Gabor filter have both in frequency and spatial domain, an algorithm is proposed to determine the parameters and the filter frequence bandwidth, and finally achieve Gabor filter enhancement in frequency domain. The Experiments show that the proposed filter could achieve a good fingerprint image enhancement result.3. A global point pattern matching method based on accelerating genetic algorithm is implementedOne of the most important phases is the fingerprint match. In this paper, accelerating genetic algorithm is presented in order to avoid the low efficiency and premature convergence of Standard Genetic Algorithm. The proposed algorithm could make the matching process converge rapidly to the global optimal solution through narrowing down the population size and the number of best individuals step by step, reducing the probability of mutation, and trying to select individuals with higher fitness. The experimental results show that the improved algorithm could enhance the accuracy of fingerprint matching with less time cost.