常用大气校正模型对图像清晰度提升的对比分析

发布时间：2018-05-26 12:29

  本文选题：大气校正 + 大气传输模型　； 参考：《航天返回与遥感》2017年05期

【摘要】：大气散射、吸收及临近效应等降低了大气调制传递函数而影响遥感图像清晰度,去除大气影响对提高图像清晰度具有重要意义。文章采用典型遥感卫星Landsat-8多光谱数据进行大气校正对图像清晰度提升的研究,基于6S模型、FLAASH模型和黑暗像元法(DOS)模型进行大气校正,得到各谱段地物反射率图像。采用常用的基于图像特征参数(灰度梯度、边缘、熵及频谱)和多光谱图像色彩保真度的清晰度评价方法对校正前后图像清晰度进行评价。结果表明:采用FLAASH、6S和DOS三种模型,大气校正后的清晰度特征参数(以熵为例)较原图平均提升程度分别为27%、10%、1.3%。而色彩保真度方面,各谱段反射率与实际反射率差(以草地为例)的平均值分别为0.018、0.028、0.038。因此,基于辐射传输模型的方法具有更高的大气校正精度,其中FLAASH模型对图像清晰度的提升最明显。
[Abstract]:Atmospheric scattering, absorption and proximity effects reduce the atmospheric modulation transfer function and affect the clarity of remote sensing images. Removing the atmospheric effects is of great significance to improve the image clarity. In this paper, the atmospheric correction of typical remote sensing satellite Landsat-8 multispectral data is used to improve the image sharpness. Based on the 6S model FLAASH model and the dark pixel method (DOS) model, the atmospheric correction is carried out, and the reflectivity images of each spectral region are obtained. The definition evaluation methods based on image feature parameters (grayscale gradient, edge, entropy and spectrum) and color fidelity of multi-spectral images were used to evaluate the image clarity before and after correction. The results show that by using the three models of FLAASHZ6S and DOS, the sharpness parameters (entropy as an example) after atmospheric correction are 27 1010 / 1.3, respectively, compared with the original map. In terms of color fidelity, the average value of reflectivity difference between each spectral segment and the actual reflectivity (taking grassland as an example) is 0.018 / 0.028 / 0.038, respectively. Therefore, the method based on radiative transfer model has higher atmospheric correction accuracy, and the FLAASH model can improve the image sharpness most obviously.
【作者单位】： 北京空间机电研究所;
【分类号】：TP751
，

本文编号：1937237