自适应空间信息MRF的FCM遥感图像聚类

FCM remote sensing image clustering based on adaptive spatial information MRF

针对传统的模糊C均值聚类算法在图像聚类时存在的不足,提出一种基于自适应空间信息MRF的FCM遥感图像聚类算法。利用空间引力模型和空间结构特征,构造一个自适应空间信息的马尔可夫随机场,将其引入到FCM模型中,使其能够自动平衡对噪声的不敏感并保留图像边缘细节的有效性,提高图像的聚类精度。为验证所提算法的有效性,利用不同的遥感图像进行实验,并与传统的聚类算法对比,实验结果表明,所提算法表现出更好的聚类效果。

Aiming at the shortcomings of traditional fuzzy C-means clustering algorithms in image clustering,an adaptive spatial information MRF based FCM remote sensing image clustering algorithm was proposed.Using the space gravitational model and spatial structure features,a Markov random field with adaptive spatial information was constructed and introduced into the FCM model,which enabled it to balance the noise insensitivity and the effectiveness of preserving the image edges automatically,in the meantime,the image clustering accuracy was improved.To verify the validity of the proposed algorithm,experiments were performed using different remote sensing images.Compared with the traditional clustering algorithm,experimental results show that the proposed algorithm shows better clustering effects.