Chan-Vese模型下的复合多相水平集图像分割

Chan-Vese模型(C-V模型)能够实现图像的二区域分割，但在多区域分割上存在局限. 目前解决C-V模型多区域分割问题有两种方案：一是采用多水平集同时收敛的并行多相分割；另一种是采用多水平集依次收敛的串行多相分割. 文中将两种方案结合起来，利用并行多相算法表示区域量大和串行多相算法分割效率高的特点，提出基于C-V模型的复合多相水平集分割算法，增加了串行结构下的分割区域量，也提高了并行结构下各水平集的实际分割效率. 实验结果表明，该方法可实现多区域分割，并能检测由弱边缘构成的子目标.

Image Segmentation Based on Chan-Vese Model Using Composite Multiphase Level Set

The Chan-Vese model is used to partition an image into two regions. However it has limitation for multi-region segmentation. To deal with the problem, two models can be used: parallel multiphase algorithm using multiple level sets simultaneously, and serial multiphase algorithm using level sets hierarchically. In this paper, a composite multiphase level sets algorithm for image segmentation based on the C-V model is proposed. The two kinds of multiphase algorithms are combined, allowing detection of more regions with a serial structure and having high efficiency for segmentation with a parallel structure. Experimental results show that the proposed algorithm can partition a given image into multiple regions and detect sub-objects with weak boundaries.