本文提出了基于Kullback-Leibler(KL)距离加权和局部邻域信息的Chen-Vese(CV)模型.引入KL距离作为内外部局部区域能量的权值系数;计算曲线附近点的局部邻域能量之和作为模型的内部能量,从而提高对边缘的检测性能,并降低区域内灰度不均匀等因素对曲线进化的影响.验证实验采用大量实际临床数据,结果表明该算法能准确地分割医学图像,且能量函数有较好的收敛性.
We propose an improved model that changes those parameters by Kullback-Leibler(KL) distance and global region by local neighborhood near the curve.Validation is implemented by experiments on mass clinical images.Take manual segmentation by a medical expert as a standard,we show that the new model improves the segmentation results compared to traditional CV model.The function has better convergence speed and stability.