摘要
高分辨率遥感影像中水系的有效提取,对于实现快速、高效、范围广的整体水系监测是必不可少的。针对山西浑源县的遥感影像,选择最合适的分割尺度和最佳的光谱因子与形状因子(紧致度和光滑性)利用易康软件进行了多尺度分割和面向对象的分类,选取对象的光谱属性信息中的Brightness和对象的几何特征中的length/width两个特征,分别采用最邻近分类法和隶属度函数法进行分类,将遥感影像分为水系和其他地类两种情形。结果表明,面向对象的分类结果要比单纯依靠光谱信息的基于像素的分类方法的结果好,形成的分类结果图更符合人的思维方式。与第二次全国土地调查成果图对比结果和本次分类结果得出的最佳分类结果概率,均表明了最邻近分类法在水系分类中更能准确的进行分类。
Effective extraction of the information of water system from the high-resolution remote sensing image is essential for the rapid, high-efficiency, and wide-rartge monitoring of the overall water system. In this paper, the remote sensing images were taken in Hunyuan County of Shanxi Province. The best segmental scale, spectral factor, and shape factor （compactness and smoothness） were selected, and the eCognition software was used to perform multi-scale segmentation and object-oriented clas- sification for the remote sensing images. The Brightness in the spectral properties of the object and length/width in the geome- try properties of the object were selected for the classification. The remote sensing images were classified into two types,water system and other type of land,using the nearest neighbor classification and the membership function classification. The results showed that the object-oriented classification provides better results than the pixel-based classification relying solely on the spectral information,and the classification results were more consistent with the way of human thinking. Through the compari- son of the land-use classification maps generated by the second Land Inventory Work and the best classification results obtained in this study, the nearest neighbor classification offers more accurate classification for the water system.
出处
《南水北调与水利科技》
CAS
CSCD
北大核心
2013年第2期157-161,共5页
South-to-North Water Transfers and Water Science & Technology
关键词
水系分类
面向对象
多尺度分割
高分辨率遥感影像
water-system classification
object-oriented
multi-scale segmentation
high-resolution remote sensing image
