基于Chan-Vese算法的自适应分等级分割方法

针对多目标物体图像的分割问题,在Chan-Vese多相分割模型的基础上,结合分等级分割的概念,提出自适应分等级分割方法,在每一阶段分割之前能够先根据图像中的物体数量判断出所需要的Level Set函数的个数,再进行分割工作.实验结果表明,自适应分等级分割方法不仅消除了多相分割模型对初始化曲线位置敏感的不足,而且能够充分利用每一个Level Set函数,减少分割步骤,并且能提高弱边界的提取精度,是一种有效且稳定的方法,能够产生光滑、准确的分割结果.

Adaptive Multiscale Image Segmentation Method Based on Chan-Vese Algorithm

Aiming at the problem of multi-object image segmentation,an adaptive multiscale segmentation model is proposed based on Chan-Vese model,which integrates the idea of multiscale segmentation and can calculate the number of Level Set functions according to the objects of images before each step of segmentation.The results of experiments indicate that the adaptive multiscale segmentation model can eliminate the sensitivity to the initial curve in multiphase segmentation model,make full use of every Level Set function,decrease the segmentation step and increase the segmentation precision.As a result,this method is effective and reliable enough to produce smooth and accurate segmentation results.