Algorithm Research For Low Resolution Fingerprint Image Recognition

Because of the universality, uniqueness and invariance of the fingerprint, the fingerprint identification technology has high feasibility and practicability, which makes it become one of the most popular and reliable personal identification technology. But along with the urgent need of the modern society, and the lack of fingerprint authentication in the practical use, how to deal with the low resolution of the fingerprint image, speed up the recognition rate, improve the recognition rate of the fingerprint image, has become a hot spot for many research institutions.In this paper, we mainly deal with three aspects: fingerprint image preprocessing,feature extraction and feature matching.Image preprocessing includes four steps:image segmentation, image filtering, image binarization, image thinning.This algorithm mainly includes:(1)In the fingerprint image segmentation and enhancement, a method is proposed to combine the morphological and the variance method for fingerprint image segmentation. The shape of the object region can not be contained by morphological opening operation and closing operation.And the algorithm based on the enhanced image of the Gaussian-laplacian operator is proposed for fingerprint image enhancement, using the Gaussian is linear smoothing filter removal effect of Gaussian noise well and in most cases, the noise of other types also have very good effect and the Laplacian has the ability to sharpening.in order to achieve the goal of highlighting the details of the target and highlight the details of the target, the need for sharpening the image processing, the combination of two kinds of filters, to ensure that the details of the image lines and has a good filtering results.(2)After fingerprint image segmentation and enhancement processing, we have the image binarization processing, this is very important in image preprocessing step, this paper puts forward a kind of based on combination of adaptive threshold and partial threshold binarization method, the algorithm is very good to improve the fingerprint image contrast of the ridge line and line and at the same time remove the image of a large number of adhesion.(3)After binarization,fingerprint ridge width are still relatively crude.Therefore, we need to deal with only one pixel wide ridge.In view of this situation,this paper analyzes some defects in the improved OPTA thinning algorithm, and the improvement of OPTA for two times improvement, through the MATLAB simulation results, the improved thinning algorithm has good refinement effect, and improveing the fingerprint of smooth and continuous.(4)In view of the results after preprocessing, the image of the endpoints and branch point as a fingerprint to extract feature points.In the subsequent fingerprint matching,using the fingerprint similarity measure, similar to the triangle edge matching and point pattern matching algorithm, simplify the computational complexity of the algorithm, using three matching the comprehensive consideration of the two fingerprints image is from a finger.