从分析古长城的遥感信息机理及其光谱特征入手,发现陕北榆林地区古长城与沙地、道路的光谱特性极其相似,因此用单一的阈值法很难将古长城信息提取出来。在综合考虑古长城的空间几何特性(包括形状、大小等)及其在遥感影像上的光谱表现特征的基础上,提出了灰度坡度法对古长城信息进行提取,即利用古长城在IKONOS图像上仅有1—2个像元宽,且与背景地物的灰度差较大的特征,把它与沙地和道路分离出来。运用这一方法并结合光谱阈值分析,从图像上进行了明代长城的提取,取得了较满意的效果,并为进一步的长城考古研究和遥感图像信息提取提供了参考。

The famed Great Wall of China stretches some 3,000 kilometers (1,850 miles). High-resolution satellite remote sensing provides cheap and quick data resources for delineating the Great Wall. In this paper, the remote sensing mechanism and spectral characteristics of the Great Wall segment in remote sensing imagery were analyzed. Due to the similar spectral response feature of Great Wall to road and sand, threshold for bands is not an effective method for the extraction of Great Wall. Thus, A gray-slope algorithm is introduced, which can extract Great Wall information effectively and easily. Due to the extensive destroy to the Great Wall, it is only 1 or 2 pixels wide in IKONOS imagery. And the gray level between Great Wall and other objects around it is very different, that is, the change of gray level along Great Wall is greater than that of road and sand. Based on these two characteristics and bands threshold, the gray-slope method makes road and sand separated from the Great Wall easily. Yulin County in Shan'xi Province is selected as a case study area, and IKONOS imagery is used as a data source. The result is satisfied when this method is applied to identify the Ming Great Wall submerged in the Maowusu desert. However, the universality of this algorithm needs to be tested further in other areas as the spatial characteristics and spectral response of Great Wall may vary.