Image Segmentation Algorithm Research Based on SVM

Abstract

Fingerprint image segmentation is an important part of fingerprint image preprocessing. It located in the front-end of preprocessing, aimed at separating the image area which is poor quality in fingerprint image and hard to recover in the subsequent processing, so that subsequent processing can focus on the effective area. Segmentation processing can not only improve the accuracy of feature extraction, and can greatly reduce fingerprint preprocessing time. Therefore, the quality of fingerprint image segmentation algorithm has a very important influence on fingerprint recognition. This paper presents a fingerprint image segmentation method basing on support vector which can block the fingerprint image, and according to the characteristics of image block contrast early split, in order to remove the relatively small change gray white background, using the RBF kernel SVM classifiers to classify, obtain a good segmentation effectiveness.