The quality of fingerprint images (FIs) varies depending on the image taking method. The performance of automatic fingerprint recognition systems highly depends on the quality of the fingerprint images. In these images often are distorted areas, where the valleys and ridges are not clear and there are discontinuities at in the ridge flows. To overcome these unwanted defects fingerprint recognition systems employ enhancement algorithms to improve the quality of the images. Following, binarization, feature extraction, and matching algorithms are executed on the enhanced image. In real-time systems, the overall execution time has primary importance and therefore, especially the enhancement algorithm must be fast and effective. In this study, we propose a fast filtering method with a mask of parabolic coefficients based on the directions. The experiments executed on FVC2000 have proved that, the algorithm provides faster and better enhancements from those described in the literature. (C) 2007 Elsevier Ltd. All rights reserved.