Research On Fingerprint Matching And Ridge Distance Estimating Algorithm

Fingerprint recognition technology, belonging to biometric features recognition, is an integrated technology that use computer to recognize personal fingerprint. Fingerprints are used for personal identification long before, many researchers have done much work on fingerprint identification and put forward some effective algorithms. But, there are some incomplete resolved problems in fingerprint identification. With the applied area extending and the requirements increasing, the reliable and robust fingerprint matching algorithms are needed. In addition, because of the impact of fingerprint image quality, the fingerprint frequency estimating is a difficulty. Therefore, we research fingerprint matching and fingerprint frequency estimating deeply.The fingerprint match algorithm based on DT is a kind of minutiae-based algorithm. On the principle of Delaunay, triangulation uses a set of minutiae to form a DT net. In a triangle, length of edges and inclination between edges are also used in matching process. After analysis, we find that there are two serious shortages in minutiae orientation and edge orientation. When 360 degree is between the corresponding minutiae orientation, we name it minutiae orientation spanning 360 degree. When the starting point and ending point of the corresponding edge are reverse, we name it edge orientation spanning 90 degree. Aiming at the two problems, we provide effective improvement method. Besides, we introduce the concept of the trend consistency of minutiae orientation and edge orientation change. Experiment results show that the identification performance has an extent improvement.Fingerprint frequency is an essential attribute of fingerprint, also is an important parameter of the fingerprint enhancement algorithm based on Gabor filter which is now a popular fingerprint enhancement method. Fingerprint enhancement has an important function in fingerprint identification, which can emphasize the key feature information and impair the impact of noises. But the essence of fingerprint frequency determines that it is incident to be impacted by noises, therefore, it is difficult to estimate fingerprint frequency accurately. To achieve this goal, we attempt to estimate ridge distance in frequency domain, provide two ridge distance estimating algorithms which are both based on spectrum analysis: , and then give the comparison between two algorithms in terms of enhancement and extracting minutiae. Experiment results show that the ridge distance is more precise. However, in the identification process of poor quality fingerprint, the identification performance relies on the improvement of enhancement algorithms.