Fingerprint Image Feature Extraction And Matching

Because of uniqueness and invariability, the fingerprint identification is becoming one of the most popular personal authentication technologies, and many fingerprint products have been offered nowadays. However, most of the key methods wouldn't be publicized, while better performance is required. So more reliable and accurate automated fingerprint identification systems (AFIS) have been of great interest to researchers.A new AFIS is established in this paper.(1) After many algorithms are compared, several algorithms including segmentation, directional image, binarization and thinning are selected and simulated. The fingerprint image preprocessing is effective.(2) In the aspect of feature extraction and post-processing, the 8-neighbour coding ridge tracing algorithm is proposed firstly. It imports 8-neighbour coding and lookup-table method into ridge tracing to make it fast. Then an elimination algorithm of false features for fingerprints based on local structural information is developed. Making use of the ridge tracing results, the algorithm can obtain some attributes of the feature points. Combining the attributes and the local structural information, it can identify and eliminate the false feature structures, so as to eliminate the false features.(3) In the matching section, a reference point orientation algorithm based on block direction and its sine component is given at first. The method finds the singular points according to their neighbor area characters, and then orients the reference point. After that, a staged fingerprint-matching algorithm based on structural model is proposed. The method constructs a local feature vector, which imports the ridge tendencies and the local structural relationship among feature points. The reference point is imported to sort the feature vectors by distance. And the staged matching approach can desert the unlike fingerprints at the first stage. All the techniques above can improve the robustness and speed of the matching.Simulation experiments have been made. The results show that the performance of all algorithms given in this paper is satisfying, and the whole system can identify fingerprints fast, accurately and reliably.