A Layered Segmentation Algorithm of Fingerprint

Fingerprint image segmentation heavily influences the performance of fingerprint verification systems. The task of generic segmentation algorithm is only to find the background and the foreground. But the layered segmentation method in this paper is not only to distinguish the background and the foreground but also to be helpful for the matching of fingerprint by endowing different weight to the minutiae in different area. The traditional methods that usually use only one feature to segment diversified images are broadly introduced. In this paper, we introduce a new approach to segment the fingerprint image that uses both the information of gray level of fingerprint image and the direction information from fingerprint image. This approach can segment the fingerprint image to three kinds of area according to the definition of every block of image. The segmentation method described above has been tested on the fingerprint images from the FVC2004 database. The fingerprint is segmented to the background, the clear area of the foreground and the dim area which can be renewed in the foreground using different criteria of estimation in different steps. The effect of this segmentation achieves the anticipative aim that we want to get.