Image quality is related to the performance of biometric identification system closely, and poor image quality often leads to bad recognition performance in biometric systems. Dorsal hand vein image is no exception. The most critical affect factor on image robustness and recognition rate for dorsal hand vein image is the image quality issues. This paper selects five parameters as the reference standard for image quality assessment, including effective region size, contrast, definition, geometric deviation and rotation angle. Based on the average slope of relation curve displayed about image quality versus recognition rate, this paper proposes a novel method to determine the optical weights for these five parameters. Then, the final score of image quality would be obtained based on the average weight. Experimental results prove that the image quality assessment score based on proposed method is accordance with human subjective assessment as well as recognition rate.