基于改进C-V模型的图像分割算法

Image Segmentation Algorithm Based on Improved C-V Model

提出一种新的模型——Chan-Vese模型,该模型是基于曲线演化、水平集方法、局部的统计信息,新模型包括两个方面:局部核心函数和惩罚项。引入局部统计信息后的新模型可以对非同质图像进行有效的分割。另外,核心函数中加入惩罚项,可以有效避免水平集函数初始化,缩短模型演化时间。通过实验的仿真结果发现,新模型在对非同质图像进行分割时得到了良好的结果。

In this paper, a new mode called Chan-Vese model is proposed, which is based on curve evolution, the level set method and local statistics. The new model includes two aspects ： local core functions and penalty term. Due to the introduction of local statistical information can make segmentation on heterogeneous image effectively. In addition, intruducing penalty term in the core functions can be useful to avoid the level set functions initialization and decrease time to model evolution. The simation of the experiment shows that, the good results are obtained in making segmetation on heterogeneous image.