摘要
由于遥感影像的复杂性和相关先验知识的缺失,传统的聚类方法在遥感影像聚类任务中往往表现不佳。文中提出一种多目标聚类算法同时优化广义模糊C均值(FGFCM)的目标函数和其相应的XB指数。与FCM相比,FGFCM对噪声更鲁棒,而且通过引入局部空间信息和灰度级信息显著提高了其聚类表现。在模拟和真实数据集上进行的实验证实了该算法的有效性和优越性。
Due to the complexity of remote sensing images and the lack of relevant prior knowledge,the traditional clustering methods cannot work well in remote sensing images clustering tasks. In this paper,a multi-objective clustering algorithm is developed,which simultaneously optimizes both the energy function from the fast generalized fuzzy c-means( FGFCM) and its corresponding XB index. Compared with FCM,FGFCM enhances its insensitiveness to noise and outliers and can achieves better clustering performance by incorporating local spatial and gray level information together. Experiments are carried out on both the simulated and real remote sensing images,and the experimental results demonstrate the effectiveness and superiority of the proposed algorithm.
作者
高博
GAO Bo(The 20th Research Institute, China Electronics Technology Group Corporation, Xi'an 710068, China)
出处
《电子科技》
2018年第6期1-4,共4页
Electronic Science and Technology
基金
国家自然科学基金(61372136)