摘要
遥感影像具有数据量大、数据结构复杂、连续、存在缺损与误差等特点,根据遥感影像的特点,提出一种基于多代表特征树的CAMFT算法。该算法通过多代表点特征树把海量空间数据进行压缩来提高效率,并且可以捕捉复杂形状聚类;算法CAMFT融入了采样思想,进一步增强了处理大型数据的能力。实验结果表明,该方法聚类精度优于K-Mean算法。
Remote sensing image segmentation have the features of largeness, complexity, continuity, spatial autocorrelation, missing data and error in spatial database. The algorithm CAMFT based on multi-representation feature tree is proposed. It can process huge data by compressing, and can detect clusters of complicated shapes. CAMFT algorithm adopts the technique of random sampling, so greatly enhances the ability to detect clusters in large databases. Experimental results show that the algorithm outperforms the K-Mean algorithm in precision.
出处
《江苏技术师范学院学报》
2007年第6期28-34,共7页
Journal of Jiangsu Teachers University of Technology
基金
福建省青年创新基金(2006F3045)
福建省自然科学基金(2007J0016)资助
关键词
遥感影像分割
聚类
多代表特征树
remote sensing image segmentation
clustering
multi-representation feature tree
作者简介
黄添强（1971-），男，福建莆田人，讲师，博士，研究方向为空间、时空数据挖掘与GIS。
