均衡FCM算法在农作物遥感影像解译中的应用

Application of balanced FCM algorithm on the interpretation of crops remote sensing image

针对传统的模糊C-均值聚类算法(FCM算法)对大数据集收敛速度慢,聚类不均匀类别样本时出现大类吃小类现象以及对初始聚类中心点要求高等问题,提出了一种基于均衡样本集思想的模糊C-均值聚类算法(均衡FCM算法)。选取Landsat8、Sentinel2A遥感卫星采集获得的哈尔滨市宾县2018年遥感图像,验证方法的有效性。结果显示,提出的均衡FCM算法可以改善传统FCM算法存在的问题,验证了均衡FCM算法的有效性。

To solve the conventional fuzzy C-means clustering algorithm(FCM algorithm)problems including slow convergence speed for large data sets,the occurrence of neglect of smaller clustered groups when the clustering categories are uneven,and high requirement on the initial clustering center points,this paper proposed a fuzzy clustering algorithm model based on balanced data sets(BDS-FCM algorithm).To verify the effectiveness,the remote sensing images of Bin County,Harbin City collected by Landsat8 and Sentinel2A remote sensing satellites in 2018 was selected as experimental subjects.Results of the experiment show that the proposed BDS-FCM algorithm can improve the conventional FCM algorithm and verify the effectiveness of BDS-FCM.