A palm vein enhancement method based on adaptive fusion and Gabor filter is proposed to overcome the degradation of recognition performance due to low image contrast, unclear details, and effects of noise, palm print, and false veins. First, the image is enhanced on the basis of the difference of Gaussian (DOG) algorithm and the partial overlapped sub-block histogram equalization (POSHE) algorithm, respectively. Second, the two kinds of enhanced images are fused adaptively according to the second-order statistic of each gray-level co-occurrence matrix. Lastly, a multidirectional Gabor filter is used to obtain the final enhanced image. This method not only retains the advantages of the DOG algorithm in separating the background from the veins, indicating an enhanced vein edge details, but also preserves the benefits of the POSHE algorithm in enhancing image contrast and local details. Simultaneously, the multidirectional Gabor filter is used to enhance the direction information, eliminate the interferences of noise, palm print, and false veins, and obtain a clear enhanced image of the palm vein. Experiment results carried out on two public databases and a self-built database yield the equal error rates of 0.0004, 0.0451, and 0.0395 and correct recognition rates of 99.96%, 94.50%, and 94.89%, respectively. The results indicate that the proposed method has better abilities of reducing equal error rate and improving recognition accuracy.