Hybrid geodesic region-based active contours for image segmentation |
| |
Authors: | Haiyong Xu Tingting Liu Guotao Wang |
| |
Affiliation: | 1. College of Science & Technology, Ningbo University, Ningbo 315211, PR China;2. Department of Mathematics, Shanxi Normal University, Shanxi 041004, PR China |
| |
Abstract: | In this paper, we propose novel hybrid edge and region based active contour models. First, we consider geodesic curve and region-based model, and evolve contours based on global information to segment images with intensity homogeneity. Second, we extend the global model to the local intensity fitting energy for segmenting the images with intensity inhomogeneity. Moreover, the level set regularization term is added to the energy functional to ensure accurate computation and avoid expensive re-initialization of the evolving level set function. Experimental results indicate the proposed method has advantage over the geodesic active contour (GAC) model, the Chan–Vese (C–V) model, the Lankton’s method and the local binary fitting (LBF) model in terms of efficiency and robustness. |
| |
Keywords: | |
本文献已被 ScienceDirect 等数据库收录! |
|