改进CV模型的医学图像分割

基于水平集的图像分割方法能有效处理拓扑结构较复杂、有分支的目标，分割结果对目标初始轮廓的位置不敏感，对图像中对比度低的边界的识别效果不佳。广义模糊算子能有效提高图像边界区域与非边界区域的对比度，图像细节分明、失真度小。运用广义模糊算子来改进水平集分割方法中的Chan-Vese模型（简称CV模型）的速度函数；并扩大传统CV模型的边界检测范围以减少迭代次数，加快收敛速度；最后消除误分割区域以进一步提高分割的准确性。对模拟和真实医学图像分割的实验结果表明：改进后的模型能较大提高分割的准确性及效率。

Improved CV model with its application in medical image segmentation

Level set based image segmentation methods can effectively handle complex topological structure and object with branches.It＇s insensitive to the position of initial contour.And it is inefficient in identifying image edges with low contrast.Generalized fuzzy operator can improve contrast between edge and nonedge pixels with clear details and low distortion.This paper uses generalized fuzzy operator to improve the evolution function of Chan-Vese mode（lCV model） in level set method;enlarges the detection region of traditional CV model to decrease iterations and improve convergent speed;at last,removes the mis-segmented area to farther improve accuracy of segmentation.Segmentation results of simulated and real medical images show that,the improved method has comparably significant improvements on precision and efficiency of segmentation.