基于改进FCM算法的遥感影像滑坡变化检测方法

A Remote Sensing Landslide Change Detection Method Based on Improved Fuzzy C-Means Clustering

针对传统基于像元的变化检测方法无法充分利用遥感多时相影像变化特征信息和检测精度低的问题,提出一种改进的模糊C均值聚类算法用于变化检测;首先利用对数法、差值法、比值法从不同代数角度获取影像的变化信息,然后将变化信息组成一幅3通道的变化影像,再利用主成分提取方法提取变化影像的主要特征,将变化强度图转换为向量样本集并映射到对应特征空间得到样本空间;最后在样本空间中进行聚类,利用粒子群算法的全局搜索特性解决传统模糊C均值聚类算法易受初始化中心影响陷入局部最优的问题,并以Davies Bouldin指标作为粒子群优化的目标值。实验结果表明,与传统的变化检测方法相比,改进算法能有效提高遥感多时相影像变化检测的精度。

In view of the problem that the traditional change detection method based on pixel can not make full use of the change information of remote sensing multi-phase image and the low detection accuracy,an improved fuzzy c-means clustering algorithm is proposed for change detection.At first,we used the logarithm method,difference method,and ratio method to get the change information of image from different algebraic angles,then we made up a 3-channel change image with this change information,and finally,we used the principal component extraction method to extract the main features of the changed image to form the sample space.Finally,clustering was performed in the sample space,and the global search characteristics of particle swarm optimization algorithm were used to solve the problem that traditional fuzzy c-means clustering algorithm is easily affected by the initialization center and falls into local optimum.The Davies Bouldin index was used as the objective value of particle swarm optimization.Experimental results show that compared with the traditional change detection method,the algorithm can effectively improve the accuracy of change detection of remote sensing multi-phase images.