巢湖流域水体信息提取方法研究

以Landsat-5 TM影像为数据源,在对研究区典型地物光谱特征值分析的基础上,利用各水体指数模型之间的关系,探讨了二值化提取水体中阈值确定的其他方法;通过对地物光谱值的分析,发现水体由于泥沙等含量的不同以及深浅的不同,其灰度值变化范围较大,与阴影的重叠较多,而采用谱间关系法,不能确定合适的阈值。提出了加权各波段灰度值并设定阈值去除阴影的方法--加权灰度阈值法,可以有效地去除阴影。研究表明结合水体指数模型法和加权灰度阈值法可以准确地提取水体信息。

The Research on Extraction Method of Water Body Information in Chaohu Lake Basin

Using Landsat-5 TM image,this paper tries to find another way to determine the threshold when extract water body through image binarization based on spectral analysis and water indices model,and improves the defects of single threshold by studying the spectrum value of the typical objects and the principle of some different water indices model.Analyzing the spectrum value find that the spectrum value of shadow stay within a stable range,while the spectrum value of water change in a big range for different depth and the content of the sediment and contaminant.Due to the high overlap of the gray value in remote sensing data,it is difficult to define shadow and water with the formula:(TM2+TM3)-(TM4+TM5)>T,in which T is difficult to be determined.To solve this problem,the paper presents the weighted gray threshold method,which give each band a weight and then sum the product of gray value and its weight,to find the threshold.The shadow and water can be separated effectively through this method.The study conclude that the water area can be obtained more accurately with the method of both water indices model and weighted gray threshold method combined.