基于形态重建的高分辨率遥感影像城市道路提取

高分辨率遥感影像中,城市道路表现为局部灰度近似、宽度变化缓慢的狭长区域。根据城市道路这些形态特性,提出基于形态重建的道路提取方法。先对原始遥感影像作增强处理,突出影像边缘信息,用Otsu自动阈值分割法对增强处理后的图像进行初步分割;再根据图像中各要素的形态特征构建不同的标记图像,分别对原图像进行形态重构,将道路和建筑物等分别生成新的图像模块;对形态重构生成的道路模块进行轮廓提取和细化,提取出道路轮廓和中心线。实验结果表明,该方法能有效地从高分辨率遥感影像中提取出城市道路网。

Urban Road Extraction of High Resolution Remote Sensing Based on Shape Reconstruction

In the high resolution remote sensing images, urban roads are the narrow areas that have the performance of local gray approximation and the width with slow change. The method of road extraction is proposed based on morphological reconstruction because of these morphological characteristics. Firstly, enhancing the original remote sensing image to highlight the image edge information, and making the enhanced image initial segmentation by the Otsu threshold segmentation method. Secondly, the original image is morphological reconstructed respectively by the different constructed markers according to the morphological characteristics of image elements, and the roads and buildings become the new image module. Lastly, the reconstructed road modules are contour extracted and refined for the road profile and centerline. Experimental results show that the method can effectively extract the urban road network from the high resolution remote sensing images.