基于主成分分析法的城市建筑物变化检测  

Urban Buildings Change Detection Based on PCA-CVA

作　　者：高扬[1] 李艳[1] 黄小赛[1] 吴剑亮 马佩坤 

机构地区：[1]南京大学国际地球系统科学研究所,江苏南京210023

出　　处：《地理空间信息》2018年第4期105-106,112,共3页Geospatial Information

基　　金：国家自然科学基金资助项目(41371331)

摘　　要：提出一个集合主成分分析法(PCA)和变换向量法(CVA)的新方法,用来探测城市环境下建筑物屋顶的几何特征和颜色变化。在主成分分析法中,原始的RGB三通道被转换成3个成分,前2个成分被用来进行下一步的变换向量法。实验证明,该方法能有效地检测出微弱的屋顶颜色变化,同时能避免阴面屋顶的漏检误检。We proposed a method combining principal component analysis(PCA) with change vector analysis(CVA) to detect changed roofs in urban environment in this paper. In the PCA, the original red, green and blue bands were transformed into three principal components(PCs). The first two PCs were used for CVA. The result shows that this method can detect the slight changes of roof color, and avoid the false detection of big shaded roofs.

关 键 词：变化检测 遥感数据 建筑物 主成分分析法 变化向量法 

分 类 号：P237[天文地球—摄影测量与遥感]