基于频域特征的遥感图像城市道路绿地覆盖轮廓提取

Contour Extraction of Green Cover along Urban Roads from Remote Sensing Imagery Based on Frequency Domain Features

提出了一种基于频域特征的高分辨率遥感图像城市道路绿地覆盖轮廓提取的方法。对遥感图像进行二维离散傅里叶变换,利用最大能量方向的频谱曲线特征,确定对应道路绿地轮廓的截止频率。建立频域Gabor滤波器,提取图像轮廓信息。根据道路特征,进行低频方向滤波,提取道路中心线。利用道路绿地覆盖与道路中心线的几何关系,实现道路绿地覆盖轮廓的信息提取。结果表明,本文方法可有效实现高分辨率遥感图像城市道路绿地覆盖轮廓的信息提取。

A new method for extracting green cover along urban roads of high resolution remotely sensed imagery is proposed,based on features in frequency domain.Cutoff frequency corresponded to contours of green cover along roads is founded through the feature analysis of the frequency curve in the direction of max energy after discrete Fourier transformation of the remote sensing image.Contour information was extracted by Gabor filter in frequency domain.The road center line was extracted by low frequency filtering based features of the road.Contour information of green cover along road was extracted using the geometrical relationship of roads and green cover along roads.Results show the efficiency of the proposed algorithm.