Finger-vein image segmentation based on KFCM and active contour model

Due to inferior contrast and intensity inhomogeneity existing in finger-vein images, current segmentation methods cannot distinguish the venous and non-venous areas effectively. To address this issue, we propose a robust segmentation approach which merges the kernel fuzzy C-means (KFCM) algorithm with an active contour model (ACM). Firstly, the KFCM algorithm is adopted to segment the venous area roughly, which is served as the initial evolution outline of ACMs. Secondly, we present a novel region-based ACM where an edge fitting item is brought in to achieve better segmentation performance. Finally, finger-vein image segmentation is achieved by minimizing the proposed region-based ACM which is served as an energy function, and the level set method is introduced to solve the minimization problem efficiently. The experimental results show that the proposed approach has significant performance in segmenting finger-vein images.