Fingerprint recognition as one of the most successful applications of biometric technology in recent years has been rapidly developed and become more and more popular. Fingerprint image segmentation technique is the basis of fingerprint processing feature point extraction, and is the key to the efficient identification of fingerprints, which directly affects the performance of the fingerprint identification system. This paper is based on the analysis of the fingerprint image segmentation methods, and develops a hybrid segmentation method by considering the advantages of the particle swarm algorithm and directional image algorithm. Computational results have shown that the hybrid method can obtain a good fingerprint image segmentation.