Research Of Fingerprint Image Preprocessing And Minutiae Extraction Algorithm

With the modern socio-economic and technological development and people's constant expansion of their areas of activity,identification technology is becoming more and more important in today's social life and economic activities. Fingerprint identification turned out to be the most popular, convenient and reliable personal authentication technologies because of the advantages of safety, easy operation, low price and so on. Up to now, identification technology has been developing for many years, but core technology is still confidential for the reason of commercial interests and others. Along with the development of society, the system should have a higher and better performance, so a more in-depth study in theory and practical aspects is necessarily needed.This paper focuses on the study of the fingerprint image's preprocessing and feature extraction under the automated fingerprint identification systems.In the pre-processing, since Teager energy operator can suppress the noises and wavelet can transform with the features of multiresolution, this paper improves Teager energy operator, takes full advantage of its non-linear high-pass characteristics and uses the Gabor function together in fingerprint enhancement based on the wavelet transformation. In the aspect of getting rid of noise and of edge enhancement in the low-quality fingerprints , this method achieved good results, significantly better than the traditional Gabor function-based fingerprint enhancement algorithm in wavelet domain. Next, we achieved binary image by using an iterative dynamic threshold seeking the best partition of the two values of the algorithm on the base of the enhanced fingerprint images, obtained thinning image by using an improved OPTA thinning algorithm, and achieved good results.In the aspect of feature extraction, this paper describes the refinement of traditional graph-based fingerprint feature point extraction. Due to the problem of too much pseudo-feature points and of time consuming brought by the origin method, we present a new feature point extraction algorithm in the binary of fingerprint image. The Feature point extraction templates were defined to extract feature point, and removal rules of pseudo-feature were defined for post-processing in the algorithm. In the end, simulation analysis of the algorithm has achieved good results, faster than the traditional feature point extraction algorithm.We implemented the algorithm mentioned in this paper with Visual C++. The experiment results show that the new algorithm proposed in this paper greatly improved the image processing effects, well extracted fingerprint feature points, Finally, we have an analysis of existing problems and indicate the direction of the future researching. With the continuous development of science and technology, we believe that fingerprint identification system will make common people's daily lives more convenient in the near future.