双中心组合迭代抑制式模糊C-均值聚类图像分割算法

针对抑制式模糊C-均值聚类算法应用于灰度图像分割时出现收敛速度较慢和像素误判的问题,通过挖掘图像同质区域内像素间的相关性与分析像素位置对类别判定的影响,提出一种双中心组合迭代抑制式模糊C-均值聚类图像分割算法.首先在图像上经选点、扩展、提取等环节优选出较好的初始聚类中心;然后按该中心分别查找图像中灰度值与其相等的像素位置并遴选产生隐藏中心;其次采用负指数函数对像素位置与隐藏中心之间的欧氏距离进行归一化,得到位置特征;接着在对该特征赋权后直接修正模糊划分矩阵;最后结合抑制式思想进一步减少算法的迭代次数.与现有的多种相关算法进行对比,实验结果表明,所提出算法在获得致密且分离性较好聚类的同时,能够改善图像分割的准确率和执行效率.

Suppressed fuzzy C-means clustering image segmentation algorithm based on combined iteration with double centers

Aiming at the problems of slow convergence speed and pixel misjudgment when a suppressed fuzzy C-means clustering algorithm is applied to gray image segmentation, the suppressed fuzzy C-means clustering image segmentation algorithm based on combined iteration with double centers is proposed via excavating the correlation between pixels in the homogeneous region of an image and analyzing the effect of pixel position on the category judgement. Firstly, by the three steps, i.e., selecting, expanding and extracting, the initial clustering centers are chosen from the pixels in an image. Then, for every initial clustering center, the pixels whose gray values are equal to that of the clustering center are searched in the image, and the hidden centers can be captured by filtering. Then, position features are calculated by the normalization of the Euclidean distance between the pixel positions and hidden centers by using a negative exponential function. Moreover, the fuzzy partition matrix is directly modified after the position features are weighted. Finally, in order to reduce the number of iteration further, the idea of the suppressed fuzzy C-means clustering algorithm is added. Experimental results show that the proposed algorithm can obtain dense and well-separated clustering in comparison with several existing algorithms, which improves accuracy and effectiveness in image segmentation.