Key Algorithms Of Automatic Fingerprint Identification Technology

In recent years, with the development of biometric identification, especially the gradually maturation of fingerprint identification technology, constructing with safety, reliability, and practical application value of the automatic fingerprint identification system becomes the focus of the researchers at home and abroad, and it has broad application prospect. Research of fingerprint identification algorithm mainly around the improvement of recognition speed and accuracy, and fingerprint feature extraction and feature matching have the closest relationship with them. This paper dose research around the feature extraction and feature matching algorithm of automatic fingerprint identification system and the main results are as follows:1. On the stage of fingerprint image preprocessing, one is to adopt the feature fusion algorithm combined with the feature of image gradient and periodicity. This method uses the rotation-invariant sobel operator to calculates the fingerprint image gradient.Calculate the fingerprint image periodicity via using Fourier transformation on the spectrum of fingerprint image periodicity, and then can achieve the the fingerprint image segmentation.The other one is to adopt a ridge frequency extraction algorithm based on statistical window and baseline. This method calculates the ridge orientation and ridge frequency first, and then uses the result as the parameters of Gabor function to achieve the fingerprint image enhancement.2. As for the feature extraction, propose a fast fingerprint global feature singular point extraction algorithm. It uses Max-Min operator on edge detection, and then deletes reasonably according to the gradient information of the edge pixels, finally calculates the orientation consistency according to a few edge pixels to decide the type of the global feature. Compared with the current popular methods, this method not only improves the anti-noise performance obviously, but also speeds up the single point location greatly.3. In the fingerprint matching, propose a multi-level fingerprint mixed matching algorithm fused with fingerprint texture feature and minutiae feature. This method fuses the fingerprint texture feature and minutiae feature, extracts the fingerprint WFMT feature by using some transformation such as Fourier-Mellin, and then compares the Euclidean distance of fingerprint feature between two images to complete the matching and re-matching. Finally match the minutiaes which have matched again after screening the multiple reference points. Compared with the current popular methods, this method has higher accuracy, and it is simple. It also improves the speed of fingerprint match.