New Fingerprint Image Thinning Algorithm

Abstract

Minutiae-based fingerprint recognition systems rely heavily on efficient and fast image enhancement algorithms. An image thinning is a very important stage of the image enhancement. A good thinning algorithm preserves the structure of the original fingerprint image, reduces the amount of data needed to process and helps to improve the feature extraction accuracy and efficiency. In this paper we describe and compare some of the most used fingerprint thinning algorithms. The results show that the faster algorithms have difficulty in preserving connectivity. Zhang and Suen’s algorithm gives the least processing time, while Guo and Hall’s algorithm produces the best skeleton quality. In this paper we propose a modified Zhang and Suen’s algorithm that is efficient and fast. Some test results show that the proposed modification better preserves the structure and connectivity of the original fingerprint image.