结合DEM的面向对象高分三号SAR影像高原山区水系提取

An Object-Oriented Water Extraction for Plateau Mountain Areas in GF-3 SAR Image Based on DEM

采用星载SAR对地观测是一项非常重要的工作,但受SAR的成像方式影响,高原山区山体阴影会对水系的提取造成很大干扰.为解决该问题,提出一种结合DEM和面向对象的高分三号SAR影像高原山区水系提取方法.首先对高分三号数据进行预处理获取辐射强度影像;然后采用分形网络演化算法对辐射强度影像进行分割获取影像对象,计算对象的后向散射特征,构建后向散射特征图;最后采用阈值计算方法获取最佳水系提取阈值得到最终结果.选取三景云南高原山区局部区域的高分三号SAR影像作为实验数据,实验结果表明采用提出的方法可有效解决山体阴影的影响,水系提取的精度分别达到99. 15%、97. 58%和97. 92%,是一种有效的高原山区水系提取方法.

Space borne SAR images can be employed to monitor the water,which is an important task in earth observation. However,it is found that water extraction is interfered by the mountain shadow in plateau mountain area because of the imaging methods of the synthetic aperture radar( SAR) images. In order to solve this problem,this paper presents an Object-Oriented water extraction scheme for plateau mountain areas in GF-3 SAR data based on DEM data. It firstly obtains radiation intensity image by the process of pretreatment technology from the data of GF-3. Then,the Fractal Net Evolution Approach( FNEA) is used for radiation intensity image segmentation to acquire the image object,and the back scattering characteristics of objects are calculated to construct the backscattering feature map. Finally,an optimal threshold calculation is established to realization the optimal water extraction threshold and the result is obtained. By using three GF-3 SAR images in local areas of the mountainous plateau in Yunnan Province,the experimental results show that the proposed method can a-chieved remove the interference of mountain shadows. In addition,the accuracy of water extraction reached99. 15%,97. 58% and 97. 92% respectively,which proves that it is an effective extraction method for plateau mountain regions.