基于相对模糊连接度的联合主动轮廓模型及其在医学图像分割中的应用

针对医学图像背景复杂、边界模糊、局部不均匀等特点,提出了一种基于相对模糊连接度的联合主动轮廓模型,并将其应用于医学图像分割。首先介绍主动轮廓模型的曲线演化方程和模糊连接度的相关理论,然后将相对模糊连接度作为曲线演化驱动力引入曲线演化方程,最后用实验证明该方法对多目标医学图像和复杂医学图像的有效性。由于模糊连接度方法综合了局部信息和全局信息,因此可以克服Li方法容易陷入局部最优的问题和Chan-Vese方法不能越过局部伪边界的问题,从而使联合主动轮廓模型的演化曲线最终收敛于全局最优边界。

Relative Fuzzy Connectedness-based United Active Contours Model and its Applications in Medical Image Segmentation

In order to solve the difficulties of complex background, fuzzy boundary, and uneven local part in the segmentation of medical images, an united active contours model based on relative fuzzy connectedness was proposed. First, the curve evolution equation of the active contours model and the related theories of the fuzzy connectedness were introduced in detail. Then, the relative fuzzy connectedness was introduced into the curve evolution equation as the driving force. Finally, comparative experiments showed the efficacies of the proposed method for multi-object medical images and complex medical images. Because the fuzzy connectedness combined the local information and global information, theoproposed method overcome the problems of Li method for falling into local optimum boundary and Chan-Vese method unable to cross the local pseudo-boundary, and then the curve of the united active contours converged to the global optimum boundary.