基于线性方向特征的高分辨率遥感图像道路提取研究  

Study on Road Extraction from High Resolution Remote Sensing Image Based on Linear and Directional Features

作　　者：高海峰 赵好好 Gao Haifeng;Zhao Haohao(School of Earth Sciences and Engineering,Hohai University,Nanjing 211100,China;School of Remote Sensing and Geomatics Engineering,Nanjing University of Information Science and Technology,Nanjing 210044,China)

机构地区：[1]河海大学地球科学与工程学院,江苏南京211100 [2]南京信息工程大学遥感与测绘工程学院,江苏南京210044

出　　处：《甘肃科学学报》2021年第3期23-28,共6页Journal of Gansu Sciences

基　　金：全球变化研究国家重大科学研究计划课题(2011CB952001)。

摘　　要：当前高分辨率遥感图像自动识别技术的关键就是图像的特征提取。针对传统道路提取自动化程度较低,提取精度不高等问题,提出一种基于线性方向特征的道路自动提取算法。首先,利用图像光谱特征进行图像预处理;其次,运用傅里叶变换在频率域上进行Gabor滤波,通过优化滤波器的尺度、频率和方向参数达到识别道路线性方向特征的目的;再次,通过图像二值化并采用形态学的图像后处理方法提纯道路。以南京主城区QuickBird遥感影像为实验数据,选取3块具有典型特征的区域完成道路提取。研究结果表明3块区域道路提取的平均精度达到95.67%,合理地验证了该算法对于高分辨率遥感图像复杂的城市道路提取具有一定的可靠性。At present,the key to the high resolution remote sensing image automatic recognition technology is features extraction.Aiming at the problems of low automation and accuracy of traditional road extraction,an automatic road extraction algorithm based on the linear and directional features was proposed.First,images preprocessing was performed by using image spectral features;second,Gabor filtering was performed on the frequency domain by using Fourier transform,and the scale,frequency and directional parameters of the filter were optimized to identify the linear and directional features of the road;third,image binarization and morphological image post-processing methods were used to purify the road.Taking the QuickBird remote sensing image of the main district of Nanjing as the experimental data,three areas with typical features were selected to finish the road extraction.The results show that the average accuracy of the road extraction in these three areas is up to 95.67%,which is reasonable to verify that this algorithm has certain reliability for complex urban road extraction from high resolution remote sensing image.

关 键 词：高分辨率遥感图像 道路提取 线性方向特征 GABOR滤波器 数学形态学 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]