A Novel Segmentation Algorithm for Fingerprint Image Based on Region Merging

The segmentation of fingerprint image is to extract the fingerprint region from the image. The traditional methods usually divide the fingerprint image into small blocks and then segment them into the foreground blocks and the background ones ignoring the relationship of the neighbors. In this paper, a novel segmentation approach is proposed based on the region merging with full benefit of the local similarity. The quality of the fingerprint regions can also be evaluated. What is more, the robustness of the algorithm is guaranteed by using an improved energy ratio and an effective threshold learning method. The testing results on FVC database are given to verify the feasibility and effectiveness.