Research On Automatic Fingerprint Identification In Entrance Guard System

With the development,the society increasingly needs highly effective,the reliable personal identification system to satisfy the needs of social economy activity and the social security.The fingerprint entrance guard system is the new technical product of an organic combination which includes the computer control technology,the fingerprint recognition technology,the smart card technology and the electronic lock technology.Along with the improvement of security requirements of all walks of life, the fingerprint entrance guard system's application is more widespread.Therefore, carrying on the researching of fingerprint recognition technology has the high practical significance and the theory significance.However due to the fingerprint recognition technology and other factors' restriction,for the large-scale fingerprint entrance guard system,stemming from to the match precision and the timely consideration,mostly uses "ID+ fingerprint"(one to one) recognition mode.The entrance guard system which is depending upon the fingerprint as the recognition mode(one to many) is still a very difficult research assignment completely.The paper mainly has deeply researched the fingerprint image pretreatment,the feature extraction,the fingerprint classification and the recognition.In the fingerprint image pretreatment part,the paper thorough analysis each step of pretreatment.Because the fingerprint image has good direction and frequency uniformity,the paper propose that uses the Gabor filter to eliminate the noise and enhance real trace.The algorithm has made a good effect.In feature extraction part,according to the different characteristic information that fingerprint classification and recognition based on,extract different characteristic point separately.The paper proposed that take the central point as the reference point, transform the real detail characteristic to the value under the polar coordinate,thus effectively overcome the distortion which is brought by the fingerprint translation and revolving,The algorithm enhance the recognition robustness.In classifies and the recognition part,based on the analysis and research of artificial neural networks model,the paper proposes that divides the fingerprint automatic diagnosis process into two steps.Firstly uses the BP neural network to carry on the classification,divides he large-scale fingerprint storehouse into certain small fingerprint storehouse t,and based on this,extracts the fingerprint type according to the BP sorter o Secondly in the corresponding small fingerprint storehouse,send the pole coordinate value of the detail characteristic point and the type of the characteristic point to the BP neural network to carry on the recognition. Finally output recognition result.The experiment proves that this algorithm has the good recognition effect,enhances system's recognition timeliness effectively.