结合梯度与区域信息的变分水平集图像分割方法

Image Segmentation Based on Variational Level Set with Grad and Region Information

在研究分析基于梯度的图像分割方法和基于区域的Chan-Vese(C-V)分割方法的基础上,提出了一种改进的分割方法,该方法将图像的边缘信息和区域信息相结合,既保留了基于图像边缘信息的分割方法对图像边缘逼近效果好、分割曲线光滑,又保留了基于图像区域信息的分割方法演化速度快、对弱边界和内部有空洞的情况处理效果较好两方面的优点.实验结果表明,该方法效果良好,鲁棒性较强.

This paper proposes an improved level set method for image segmentation after analysed methods based on image grad and image region information. The new method combines image grad with image region information, which not only retains the edge-based segmentation method on the effect of edge approximation, and segmentation smooth curve, but also retains the region-based on the evolution of the segmentation method is fast, weak borders and there are empty better situation to deal with the merits of both sides. The experimental results show that the method works well and with strong robustness.