Research of the Matlab application in the fingerprint identification system

Abstract

The quality of fingerprint pictures is the key to fingerprint identification, but in actual practice, the pictures we obtained often have all kinds of noise such as scars, perspiration, and stains, as well as some noise caused by uniform contact with fingerprint collecting devices. Based on work of predecessors, we propose a fingerprint identification pre-treatment algorithm in Matlab. Based on Matlab, this article provides an algorithm implementation, and an improved method. The results of each fingerprint picture processing module, mainly including image segmentation which could be separated, obtained a fingerprint image from a background area. Image filtering, removing burr, cavity management and binarization processing (with the thought of self-adapted local threshold binariztion) which make the fingerprint image clearer, eliminate unnecessary noises and are beneficial to further identification. To thin the image, we first, adopt the quick thinning algorithm to handle the preliminary thinning. The streakline after thinning has a certain width, and secondly, the advanced one-pass thinning algorithm (OPTA) is adopted for use the fingerprint image that after preliminary thinning; this makes all areas, except the bifurcation point, remain a single-pixel wide, correcting the streakline that has been thinned by advanced OPTA. Then we get a clear fingerprint picture, extract the fingerprint feature point (spurious minutiae) from this picture; this feature point contains a lot of false features that take a lot of time and influences the matching precision. In this paper, the authors adopt eliminating the false features by edge and distance, lessening the false feature points by approximately a third, and then next extract reliable information of the feature points and store in the bookbuilding template. When matching a fingerprint, we get clear fingerprint image using the same method, and build a contrast template; at last, we compare the contrast template with bookbuilding template and then get ideal results. Based on Matlab, with this method we are unable to do the simulation step-by-step with the fingerprint identification pre-treatment algorithm, but also see the result of image processing algorithm intuitively, which cooperates with the algorithm research effectively. Experimental results show that with this algorithm, which is on the basis of Matlab, the processing result is more ideal, and this method is not only simple and quick, but also has a high precision, and satisfy the identification applicability.