A Comparative Study of Fingerprint Thinning algorithms

Thinning plays a very important role in the preprocessing phase of automatic fingerprint recognition/identification systems. The performance of minutiae extraction relies heavily on the quality of skeletons used. A good fingerprint thinning algorithm can depress image noise and promote the robustness of the minutiae extraction algorithm which helps improve the overall performance of the system. Many thinning algorithms have been devised and applied to a wide range of applications including, Optical Character Recognition (OCR), biological cell structures and fingerprint patterns. With so many thinning algorithms available, deciding which one is appropriate for a particular application has become very difficult. In an effort to assist fingerprint biometrics developers choose an appropriate thinning algorithm, a study was taken to compare performance of four different thinning algorithms. These four algorithms are implemented and their performance evaluated and compared. The algorithms are compared in terms of the quality of the skeletons they produce (i.e. connectivity and spurious branches) as well as the time complexity associated with each algorithm. Results show that faster algorithms have difficulty preserving connectivity. Zhang and Suen’s algorithm gives the least processing time, while Guo and Hall’s algorithm produces the best skeleton quality.