基于频谱空间的高分辨率遥感图像城市道路绿地轮廓特征检测  

Contour extraction of urban green spaces along roads using high resolution remote sensing imagery based on the spectrum space

作　　者：赵好好 ZHAO Haohao(School of Remote Sensing&Geomatics Engineering,Nanjing University of Information Science and Technology,Nanjing 210044)

机构地区：[1]南京信息工程大学遥感与测绘工程学院,南京210044

出　　处：《南京信息工程大学学报（自然科学版）》2020年第5期625-633,共9页Journal of Nanjing University of Information Science & Technology（Natural Science Edition）

基　　金：国家自然科学基金(41801239)。

摘　　要：通过高分辨率遥感图像频谱能量特征分析,提出了一种基于频谱空间的城市道路绿地轮廓特征检测方法.首先对遥感图像进行傅里叶变换,分析频谱能量分布曲线,确定道路绿地轮廓的方向和截止频率,设计Gabor滤波器提取绿地轮廓特征;然后利用低通滤波获取道路中心线,根据道路中心线的缓冲区对绿地轮廓进行定位,得到道路绿地轮廓特征.结果表明,基于图像频谱特征分析的方法,使滤波器参数的设计更加精确,可有效实现高分辨率遥感图像城市道路绿地轮廓特征提取.A contour extraction method for urban green spaces along roads was proposed based on the spectrum domain.Remotely sensed images were transformed into the Fourier domain,the spectrum energy was analyzed,and the direction and cut-off frequency corresponding to the contours of green spaces along roads were established.Gabor filters were specifically designed to extract the contours of these green spaces.The central lines on roads were detected by filtering at a low frequency,and the green spaces along the roads were extracted as a buffer of these central lines.The results showed that the filter parameters can be set accurately based on the frequency spectrum analysis of image features,and the contours of urban green spaces along roads can be effectively extracted.

关 键 词：高分辨率遥感图像 城市道路绿地 轮廓特征 频谱空间 GABOR滤波器 

分 类 号：P237[天文地球—摄影测量与遥感]