基于频谱分析的高分辨率遥感图像边缘特征检测研究  

Edge Feature Detection of High Resolution Remote Sensing Images Based on Frequency Spectrum Analysis

作　　者：赵好好[1,2] 肖鹏峰[1] 冯学智[1] 

机构地区：[1]南京大学地理信息科学系,南京210093 [2]南京信息工程大学遥感学院,南京210044

出　　处：《遥感信息》2013年第6期96-101,115,共7页Remote Sensing Information

基　　金：全球变化研究国家重大科学研究计划课题(2011CB952001)

摘　　要：针对高分辨率遥感图像的边缘特征,提出了一种基于频谱分析的图像边缘检测方法。首先对南京市主城区典型地物QuickBird图像进行傅里叶变换,分析频谱能量的角向分布特征,对最大角向能量分布曲线求一阶导数,将其峰值作为边缘检测滤波器的中心频率。通过设计方向Gabor滤波器进行频域滤波,达到图像边缘特征检测的目的。结果表明,进行频谱分析能够确定图像边缘特征的最佳截止频率,从而为高分辨率遥感图像边缘特征检测提供一种新方法。A method of image edge detection based on spectrum analysis is proposed aiming at the features of high resolution remote sensing images.Firstly the QuickBird images in typical area of Nanjing City were transformed to frequency domain by Fourier transform,and the angular energy distribution was analyzed.The first derivative of the maximum angle energy distribution curves was calculated,and the frequencies at peak value were considered to be the center frequency of the filters.Frequency domain Gabor filters were designed with the purpose of the edge detection of the images.The results showed that the optimum cut-off frequency can be determined by spectral energy analysis,and the proposed algorithm provides a new method for the edge feature detection of high resolution remote sensing images.

关 键 词：高分辨率遥感图像 频谱分析 边缘特征 边缘检测 GABOR滤波器 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]