Enhancing low-quality fingerprint image is the effective method for improving the accuracy of minutia extraction and performance of the automatic fingerprint identification system. Fingerprint image is one kind of regular texture image in nature. To design two-dimensional sinusoidal surface model which accorded with the ridge gray distribution rule and propose the fingerprint image enhancement algorithm based on the two-dimensional sinusoidal surface model after analyzing the basic character of the fingerprint image. The experimental results indicate that the fingerprint image enhancement algorithm has better connecting ability for the broken ridges than the Gabor fingerprint enhancement algorithm. The algorithm can improve effectively the fingerprint image enhancement result and the accuracy of the minutiae extraction.