遥感图像空间分辨率改进值计算方法  

A Quantification Method of Spatial Resolution Improved Value on Remote Sensing Image

作　　者：郭从洲[1] 赖广陵 童晓冲[1] 丁璐[1] 

机构地区：[1]信息工程大学,河南郑州450001

出　　处：《测绘科学技术学报》2017年第1期38-43,共6页Journal of Geomatics Science and Technology

基　　金：国家863计划项目(2012AA7032031D);国家自然科学基金项目(11373043)

摘　　要：为了克服超分辨重建后的遥感图像空间分辨率的界定还采用人工对比判读存在误差和结论不统一的缺点,利用调制传递函数、奈奎斯特采样定理和人眼极限频率,建立了一种新的空间分辨率客观评价方法。利用该方法计算超分辨率重建后的遥感图像相比原始遥感图像空间分辨率的提高倍数,从而推断出重建后遥感图像的空间分辨率的大小。在数值测试中,利用不同的超分辨率方法对分级变频矩形光栅图像进行重建,采用提出的空间分辨率评价方法,与归一化均方误差、峰值信噪比、信息熵、灰度平均梯度进行客观评价的结果一致。此方法为遥感图像空间分辨率改进值的计算提供了一种可行的量化模型。In order to overcome the weakness that the spatial resolution of remote sensing image after super-resolution reconstruction is still defined by artificial interpretation contrast method which leads to a disunity between the error and the conclusion,a new method of spatial resolution objective evaluation is put forward in this paper,based on the modulation transfer function,the Nyquist sampling theorem and the human eye limit frequency.This method can be used to calculate the multiplier of spatial resolution which are from the original remote sensing images to the super-resolution reconstructed ones,so that the value of reconstructed remote sensing image spatial resolution can be deduced.In the numerical test,various super-resolution methods are used to reconstruct the classifying frequency rectangular raster image.The result of spatial resolution evaluation methodology presented in this paper is consistent with outcomes evaluated objectively by the normalized mean square error,the peak signal to noise ratio,the information entropy or the gray average gradient.A new feasible method for the improved value calculation about spatial resolution of remote sensing image is provided by this model.

关 键 词：空间分辨率 调制传递函数 截止频率 人眼极限频率 超分辨率重建 

分 类 号：P237[天文地球—摄影测量与遥感] TP391.9[天文地球—测绘科学与技术]