基于CV模型的CT图像分割研究

Study of CT Image Segmentation Based on CV Model

由于医学图像边缘模糊、不均匀性等特点,使用传统的Chan-Vese(CV)模型方法难以达到分割要求,同时该方法存在计算量大、分割速度慢的问题。本文提出了一种基于CV模型改进的分割算法,在水平集演化迭代过程中,根据当前主动轮廓线的位置,引入图像局部灰度信息,提高了水平集能量项的有效性和该模型的收敛速度,并提出了一种关于图像序列的分割方法。实验结果表明,运用本文提出的方法能够快速、准确地提取图像中感兴趣目标,是一种较为理想的医学图像分割方法。

The medical image has the characteristics of blurred edges and heterogeneity, it is difficult to achieve the goal of segmentation using the traditional Chan-Vese model method, at the same time, the method is large amount of calculation and the speed is slow. Therefore, this paper presents an improved segmentation algorithm based on the Chan-Vese model, the convergence speed of the level set and the effectiveness of energy item are enhanced according to the current active contour and the local information of image during the iterative process, and a new segmentation method for image sequence is proposed. The experiments show that the improved Chan-Vese model can extract the object interested in the image quickly and exactly, it is an ideal method for medical image segmentation.