Electerical and Computer Engineering, Tarbiat Modarres University
Fast and accurate ridge detection in fingerprints is essential to each AFIS (Automatic Fingerprint Identification System). Smudged furrows and cut ridges in the image of a finger print are major problems in any AFIS. This paper investigates a new online ridge detection method that reduces the complexity and costs associated with the fingerprint identification procedure. The noise in fingerprint is highly correlated and the statistics of such noises are unknown. In this case, image enhancement techniques based on probabilistic approach may not be suitable. In view of imprecise knowledge about the fingerprint noise, a fuzzy set theoretical approach would be more effective. A new structural algorithm for ridge restoration which is based on unsupervised fuzzy classification technique is described. The accuracy and speed of the proposed method are tested for a large number of fingerprint images with different initial qualities, and are found to be superior to the conventional methods.