Research And Implement On Synthetic Fingerprint Generation

Fingerprint recognition is one of the most widely used biometric identification methods. In the course of the study of Automated Fingerprint Identification Algorithm,we need to establish large-scale fingerprint database for algorithm testing and optimization. Synthesis of fingerprint method can create large-scale database in relatively simple conditions.Considering this problem,a synthetic fingerprint generation method which called SFINGE is introduced and implemented in this paper. Orientation field generated by this method, the density map generation of steps to obtain the fingerprint ridge line structure, that is, generate a master fingerprint. Then by adjusting the ridge line width,fingerprint deformation distortion,plus scratches,plus noise,plus the background to obtain the final steps such as fingerprint image compression.This paper describes each step of this method in detail, then the fingerprint generated by several key steps to put forward and introduced a variety of improved methods, which improved gabor filtering method is used to solve the problem of real fingerprint ridge line and valley line width is not necessarily the same. Combination orientation field model avoids the zero-pole model on the lack of distortion in the regional orientation field generation, making the direction of the entire fingerprint region field simulation closer to the real fingerprints. Perlin noise model resolve the problem of adding macular noise for fingerprint. In addition, a new scrapping fingerprint background is introduced in this paper, which greatly enriched the differences and diversity of the fingerprint image. Finally, this paper through a series of experiments described the steps of SFINGE generation method,generation results and the performance of the improved methods.