In a fingerprint recognition system, the process of extraction of feature points is affected significantly by noise. It happens frequently that important characteristic information is omitted due to an inappropriate noise reduction. In this paper, we present an approach to fingerprint image enhancement which uses a filter mask to increase the contrast between the ridges and valleys and to reduce noise. The proposed approach, different from the methods in the literature, determines mask coefficients according to the aspects of the fingerprint images. The experiments shows the proposed preprocessing for fingerprint images produces good results for most of the fingerprint images.