Abstract: The performance of the segmentation algorithm for fingerprint images affects the performance of the whole fingerprint identification system．It presents a segmentation algorithm for fingerprint images，based on the traditional segmentation algorithms and several mixed—technology segmentation algorithms．This algorithm can accurately segment the fingerprint images into the background area and the foreground area，and then segment the foreground area into the clear—cut area and the blurred area，and finally segment the blurred area into the retrievable area and the irretrievable area．The result shows that this segmentation algorithm provides a good segmentation result，and that it is robust against noise．Therefore，it has much higher practical value．

Key words: fingerprint image；segmentation；characteristic quantity；directional graph；gray variance；frequency of gray levels