融合空间信息的改进FCM图像分割算法

Improved FCM Image Segmentation Algorithm Based on spatial information fusion

摘要:

针对FCM(Fuzzy C-Means)算法在图像分割时存在选取初始聚类中心不佳与算法抗噪性差的问题,提出一种融合空间信息的改进FCM图像分割算法；首先采用了直方图算法和LOF(Local Outlier Factor)算法自适应地选取初始聚类中心,之后使用马尔科夫随机场得到先验概率改进目标函数,使用修正隶属度矩阵的方法改进算法流程,最后使用改进算法进行图像分割；为验证该算法性能,使用Berkeley图像数据集作为实验数据,选取Dice系数、JS系数、SA系数、PSNR指数、运行时间及迭代次数作为评价标准；实验结果表明,该算法能够获取更优初始聚类中心,在处理不同噪声图像上有更好的鲁棒性。

Abstract:

Aiming at the problems of poor selection of initial clustering center and poor noise resistance of Fuzzy C-Means (FCM) algorithm in image segmentation, an improved FCM Image segmentation algorithm integrating spatial information is proposed; Firstly, the histogram algorithm and Local Outlier Factor (LOF) algorithm are used to adaptively select the initial clustering center, then the Markov random field is used to obtain the a priori probability to improve the objective function, the method of modifying the membership matrix is used to improve the algorithm flow, and finally the improved algorithm is used for image segmentation; In order to verify the performance of the algorithm, the Berkeley image data set is used as the experimental data, and Dice coefficient, JS coefficient, SA coefficient, PSNR index, running time and iteration times are selected as the evaluation criteria; Experimental results show that the algorithm can obtain better initial clustering centers and has better robustness in dealing with different noisy images.