Fingerprint Segmentation Based on Improved Active Contour
Weixin Bian, Deqin Xu, Yiwei Zhao · 2009
Snake (active contour) model, introduced by Kass in 1987, is a dynamic curve model with energy-minimizing. Snake algorithm, which has advantages in extracting target object from a certain region, is an effective method in image segmentation. Based on the analysis of the snake model and the regional information of the edges of the fingerprint images, an improved active contour for the segmentation of fingerprints is presented in this paper. In this paper the limitations of the segmentation of fingerprint images using the snake as suggested are pointed out. The authors present a solution to the fingerprint segmentation by replacing the standard external energy in the snake energy balance equation with the difference between peaks in the directional histogram and gray variance, and a new external energy that is applied to control the snake outward expansion or inward contraction. This method has been tested by a large number of fingerprint images from different sources, and is found to be more accurate and robust.