由于单一水平集只能通过其符号表达目标和背景两个区域,因此采用单水平集的Chan和Vese（C-V）模型无法分割出目标内部的子目标.为此,提出了基于C-V模型的目标多层次算法.首先给出了目标多层次分割策略;然后,提出了实现本策略的关键技术--背景填充技术,并从其视觉原理、技术实现和理论证明3个方面详细进行了论述;最后,将该技术与C-V模型相结合,提出了目标多层次分割算法;实验结果表明,本文算法能够实现目标多层次分割,并对目标内部含有弱目标的图像特别有效.

The model proposed by Chan and Vese using one level set function is not able to obtain sub-objects in the object because one level set can only represent one object and one background via its sign.To solve the problem,an algorithm for multi-layer object segmentation based on Chan-Vese(C-V) model is proposed.Firstly,an idea for multi-layer object segmentation is proposed after the analysis of the C-V model.Secondly,a key technique,named as the technique of painting background,is developed and proved following the theory of the simultaneous brightness contrast.Thirdly,the proposed algorithm is presented using the proposed technique and the C-V model.Finally,experimental results show that the proposed algorithm is especially effective for the detection of the sub-objects with weak boundaries in the object.