摘要
本文根据高分辨率遥感影像城市道路与房屋等建筑物在空间域中光谱特征差异很小,而在频率域中区别很大的特点,提出一种基于频率域的道路提取方法。在图像频率域中利用Butterworth高通滤波器对图像进行锐化增强处理,突出道路的边缘信息,将道路与建筑物初步区分开来;再对增强后的图像二值化,通过形态变换等方法对图像中的建筑物进行归类合并,并去除道路上的行人、汽车、斑马线、树的阴影等噪声点;最后对图像进行细化和修剪,得到单像素宽的道路中心线信息。通过遥感图像实验验证,该方法可以快速准确提取复杂的城市道路信息。
This paper presened a high-pass filter based on the road extraction. The image frequency domain using Butterworth high-pass filter on the image were enhanced, highlighting the edges of the road information, road and buildings were separate. Then on the enhanced image binarization, by morphological transformation the buildings were classified, the pedestrian on the road, ears, crossing, tree shadows, noise points were remored. After thinning and pruning the image, finally it got a single pixel wide road centerline information. Verified by the actual remote sensing images, the method could accurately extract the complex urban road network.
出处
《测绘科学》
CSCD
北大核心
2011年第4期50-52,共3页
Science of Surveying and Mapping
基金
信阳师范学院青年自然基金(20100057)
