Fingerprint classification is an important index mechanism. The performance of fingerprint recognition algorithm can be affected as a result of shift and rotation between a template fingerprint image and a compared fingerprint image or low quality fingerprint image. A novel fingerprint classification algorithm has been proposed in this paper

which is based on Singularity and HMM. Firstly

the belief function assignment of Singularity classification and HMM classification is determined

respectively

and then

the final classification result was obtained by using D-S evidence theory. The experimental results based on standard fingerprint datasets have verified the accuracy rate of classification was achieved to 94.5%. Moreover

the proposed algorithm had a good robustness to those fingerprint images of shift and rotation and low quality fingerprint image

thus this method has some values in practice.