改进的混合像元分解法用于快速水污染遥感评价研究  

An Improved Method of Spectral Unmixing of Remote Sensing Image and Its Application in Water Pollution Monitoring and Assessing

作　　者：钱少猛[1] 蔺启忠[1] 陈雪[2] 

机构地区：[1]中国科学院遥感应用研究所,北京100101 [2]北京师范大学资源与环境科学系遥感与地理信息系统研究中心,北京100875

出　　处：《地理与地理信息科学》2003年第2期36-38,共3页Geography and Geo-Information Science

基　　金：十五"国家科技攻关项目"滇池蓝藻水华污染控制技术研究"子课题"滇池水华污染遥感监测与水域生态环境管理信息系统

摘　　要：该文介绍了一种改进的混合像元分解方法。利用该方法对滇池的各类污染(叶绿素,营养盐等)进行快速评价,可以直接得出滇池各个区域所受各类污染的程度,从而给出滇池全区各类污染的情况分布。The spectral signal of a pixel in remotely sensed image in most cases is the result of the reflected spectral properties of mixed land cover types constituting the pixel. However, most remotely sensed image classification algorithms aim at sorting a pixel according to its spectral statistic feature. Spectral unmixing can not only give the abundance images of surface cover types constituting the pixel, but also get the classification image. This paper introduced an improved method of spectral unmixing of remote sensing image and its application in water pollution monitoring and assessing. A TM image of the Dianchi Lake received on October 4,2001 was processed and analyzed as the following: (1) Atmospheric calibration of the image data, (2) Extracting the water area from other land cover types, (3)Definiting the unpolluted water area and calculating the average spectral value of all selected pixels,(4) Subtracting the average value from all pixels, (5) Selection of the endmembers representing different water pollution types, (6) Spectral unmixing of the image data by the linear model, (7) Getting the abundance image of every endmember constituting the pixel, and the classification image. By the classification image,the distribution of each type of water pollution in the Dianchi Lake was showed. Moreover, the factors affecting the classification product were discussed and some research aspects for the future were put forward.

关 键 词：遥感 混合像元分解 水污染 滇池 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置] X524[自动化与计算机技术—控制科学与工程]