基于水平集的多相活动轮廓图像分割模型

提出了一种新型的多相活动轮廓模型，是无边活动轮廓模型的广义形式。该模型具有如下特点：（1）提出了背景填充技术，可以在检测目标内部弱边缘时去除阻碍检测的背景信息；（2）在单次二相水平集收敛的基础上，采用多次收敛方式实现了多相分割模型（n-1次收敛实现n相分割模型，n〉1）；（3）介绍了一种提升算法，进一步增强了模型的计算稳定性。实验结果表明，该模型对弱边缘检测特别有效。

MULTI-PHASE ACTIVE CONTOUR MODEL FOR IMAGE SEGMENTATION BASED ON LEVEL SETS

A new multi-phase active contour model is proposed for the image segmentation. It is a generalization of the C-V model with the following characteristics: (1) A key technique, called the technique of painting background (TPBG), is developed to remove the information of the background, which blocks the detection of weak boundaries in the object; (2) The two-phase level set is applied multiple times for getting the multi-phase segmentation model (n-1 times for the n-phase model, n>1); (3) A scaling-based method is introduced to improve the basic model. Experimental results show that the proposed model is effective for detecting weak boundaries.