GF-1 PMS multi-spectral data reconstruction based on multivariable linear regression model
王伟强 1李永康 1盛雅丽 1刘佳乐 1李新伟 1刘吉凯 1马强1
扫码查看
点击上方二维码区域，可以放大扫码查看
作者信息
1. 安徽科技学院资源与环境学院,安徽凤阳 233100
折叠
摘要
目的:多元线性回归模型在保持输入自变量光谱信息和空间特征的同时,通过线性变换获取自变量和因变量的光谱拟合关系,对原输入自变量的光谱信息进行优化,从而获得高空间分辨率和丰富光谱信息的重构数据.方法:利用同期获取的OLI(Operational Land Imager)和 PMS(Panchromatic and Multi-spectral Scanner)多光谱遥感影像,根据最小二乘法构建多元线性回归模型,重构生成具有丰富光谱特征和空间特征的遥感影像,从主客观两个方面评价重构影像的质量.结果:在目视解译(主观)方面,重构影像在一定程度上保留了原OLI影像的光谱特性,提升了原PMS影像的清晰度和分辨性;在量化角度(客观)方面,重构影像的信息量和平均梯度比原OLI对应波段影像的信息量(在部分波段上)和平均梯度要低,但比原PMS影像的信息量和平均梯度要高,可见重构影像的质量介于原PMS影像和OLI影像的质量之间.结论:以青海省门源回族自治县的耕地内不同作物为实例对象,利用最大似然法获取门源县青稞和油菜的空间分布,研究区实测数据验证表明,重构影像对耕地内部青稞与油菜的提取精度高于原PMS和OLI多光谱影像的提取精度.
Abstract
Objective:High spatial resolution and spectral information data are obtained by a multiple linear regression model which can express fitting relationship between the spectrum of dependent variables and independent variables.The multivariable linear regression model keeps the input variables of spectral information and spatial characteristics and optimizes the spectral information of the original input variables.Methods:The paper used the OLI and PMS multi-spectral remote sensing images acquired at the sameperiod to construct the multivariable linear regression model based on the least squares method to generate thereconstructed remote sensing image with abundant spectral information and high spatial resolution.Then the quality of reconstructed image was evaluated from subjective and objective aspects.Results:In terms of the visual interpretation(subjective),the reconstructed image retained the original OLI spectral characteristics on a certain extent and improved the clarity and resolution of original PMS image.In terms of quantitative indicators(objective),the information and the average gradient of reconstruction image were lower than the original OLI corresponding band image information(on the part of bands)and average gradient,but were higher than that of the original PMS image,and the quality of reconstructed image was between the original PMS image and OLI image.Conclusion:The paper took the different crops in the cultivated field in Menyuan County of Qinghai Province as the study object,and took highland barley and canola's classification accuracy as research objectives,and used the maximum likelihood method to obtain the spatial distribution of highland barley and canola in Menyuan county.The validated result by the measured GPS data in the study area on August 10-11,2014 illustrated that the reconstruction image's extraction accuracy of highland barley and canola was higher than the original PMS and OLI multi-spectral images.
关键词
多元线性回归模型/OLI/PMS/重构影像/质量评价
Key words
Multivariable linear regression model/OLI/PMS/Reconstructed image/Eoaluaction
维普
万方数据