基于证据理论的脑部图像分割算法

针对在马尔可夫随机场和模糊聚类二维直方图方法中存在的像素点分割结果不一致的现象，提出一种基于Dempster-Shafer(D-S)证据理论的图像融合分割算法。利用模拟退火算法恢复出脑部轮廓，根据该轮廓从原图中提取出脑核部分，采用马尔可夫随机场和模糊聚类二维直方图方法分别对脑核部分进行分割，通过D-S证据理论将分类不确定的争议像素划归到似真区间，并进行融合分割。实验结果表明，该算法能解决争议点的归属问题，有效滤除噪声，稳健性较好。

Brain Image Segmentation Algorithm Based on Evidence Theory

This paper proposes an image segmentation algorithm,which combines Markov Random Field（MRF）,two-dimensional histogram of fuzzy clustering and Dempster-Shafer theory.The algorithm uses simulated annealing to restore the contour of the brain,and the nuclear part of the brain is extracted from the original image according to the profile,then the brain nuclear image is divided by MRF and two-dimensional histogram of fuzzy clustering.However,two ways above lead to different classification results while the controversial pixels are classified,so it makes use of the Dempster-Shafer evidence theory to assign the controversial points to the plausibility interval and divide them.Experimental results show that the proposed algorithm is robust,which can solve the problem that the class attribution of the controversial points and suppress the noise effectively