论文部分内容阅读
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。
In this paper, a new method of road extraction based on frequency domain is proposed according to the characteristics of high resolution remote sensing images such as urban roads and buildings in the spatial domain in which the difference of the spectral characteristics is very small and the difference in the frequency domain is very large. In the image frequency domain, the Butterworth high-pass filter is used to sharpen and intensify the image, highlighting the edge information of the road and preliminarily separating the road from the building. Secondly, the image is binarized through the morphological transformation The buildings in the image are classified and merged, and noise points such as pedestrians, cars, zebra crossing and tree shadows on the road are removed. Finally, the image is refined and trimmed to obtain single-pixel wide road center line information. Experimental results of remote sensing images show that this method can extract complex urban road information quickly and accurately.