面向对象的高分辨率影像水体信息提取研究  

Study on Water Body Information Extraction from High Resolution Remote Sensing Image Based on Object-Oriented Method

作　　者：陈晨[1] 陈静欣[2] 李向前[1] 张琴[3] 郑淑倩[3] 

机构地区：[1]江苏省地质调查研究院,江苏南京210018 [2]江苏省测绘产品质量监督检验站,江苏南京210013 [3]河海大学地球科学与工程学院,江苏南京210098

出　　处：《人民黄河》2013年第9期68-70,73,共4页Yellow River

摘　　要：以昆山南部的WorldView-2影像为研究数据(全色波段分辨率为0.5 m,多光谱波段分辨率为1.8 m),基于eCognition软件平台,通过选择适宜的尺度,并结合纹理、形状、光谱等特征构建合理有效的规则集,实现面向对象的水体信息提取。利用实测结果对遥感解译结果进行形状一致性和面积一致性检验,精度分别达到89.14%和92.06%,表明利用面向对象技术从高分辨率遥感影像提取水体信息具有一定的可靠性和实用性。Based on the software of eCognition and using the WorldView-2 image data of the southern region in Kunshan, this paper realized water body information extraction with object-oriented technology through choosing suitable scale and combining the features of image texture, shape and spectrum to construct rule set. Using the measured results to check the consistency of shape and area, the degree of accuracy can reach 89.14% and 92.06% respectively, which indicates using object-oriented classification technology to extract water body information from high resolution re- mote sensing image is reliable and practical.

关 键 词：面向对象 水体信息提取 WorldView-2影像 特征集构建 

分 类 号：P331[天文地球—水文科学] P337[水利工程—水文学及水资源]