A novel algorithm for fingerprint orientation extraction and image enhancement based on Gabor filters

The traditional orientation extraction algorithm based on pixels and the common used Gabor filter are both meliorated by compressing original fingerprint image to 1/4 size and getting the low-frequency image using Gauss filter for former, and by changing the rectangular template into 1-D linear template with two orthogonal beelines for latter. Experiments on FVC2004 databases and live-scan database both shows that, compared with other method, the accuracy and the reliability of the orientation extraction are obviously improved and the pre-processing time is greatly reduced while the image is further more enhanced with the algorithm proposed in this paper.