面向对象的遥感影像地物信息提取方法研究  

Research on Object-Oriented Extraction Method of Ground Object Information from Remote Sensing Image

作　　者：庄胜兵 ZHUANG Shengbing(Guangzhou Urban Renewal Planning Institute.,Guang Dong Guang Zhou 510000)

机构地区：[1]广州市城市更新规划设计研究院有限公司,广东广州510000

出　　处：《长江信息通信》2021年第11期39-41,共3页Changjiang Information & Communications

摘　　要：随着遥感技术的不断发展,卫星拍摄的影像清晰度不断升高,纹理几何信息不断丰富,出现了高分辨率遥感影像。如何对高分辨率遥感影像中的地物信息进行高精度的提取,是当前研究的热点问题。面向对象的方法主要是利用描述的一类相似的影像像素的集合,将像素值在某一范围的像素分在某个区域范围内进行描述。文章使用面向对象的方法,利用遥感影像中丰富的几何、纹理、光谱等特征信息,通过SVM算法对高分辨率遥感影像中的地物信息进行提取,实验证明该方法的有效性。With the continuous development of remote sensing technology, the definition of the images captured by satellites is increasing, and the texture geometric information is constantly enriched, resulting in the emergence of high-resolution remote sensing images. How to extract the ground object information from high-resolution remote sensing images with high precision is a hot issue in current research. Object-oriented method mainly uses a set of similar image pixels to describe the pixels in a certain range in a certain area. In this paper, the object-oriented method is used to extract the ground object information from the high-resolution remote sensing image by using the abundant feature information such as geometry, texture and spectrum in the remote sensing image. Experiments show that the method is effective.

关 键 词：面向对象 特征信息 SVM 地物提取 

分 类 号：P237[天文地球—摄影测量与遥感]