Abstract: In order to recognize one's identity accurately, through using the extension direction of vein and location relations among contexts, hidden Markov model (HMM) was applied in vein recognition system and a vein recognition algorithm based on establishing HMM adaptively through using GA (genetic algorithm) was proposed. The skeleton information of vein is acquired after image pretreatment, and the thinned vein image was processed by Radon transformation, then each vein object can be expressed as an HMM; for identified train sample library, the parameters of the HMM were adjusted through using GA such that the observation probability values of all the tested images provided by true matching models are much larger than those provided by false models, thus the degree of the differentiation among vein objects was significantly increased. Experiments showed that the proposed algorithm has higher correct recognition rate and better real-time performances.