基于高斯金字塔的遥感云图多尺度特征提取

The Extracting of Multiscale Features of Remote Sensing Cloud Images Based on Gaussian Pyramid

提出了一种针对可见光遥感图像云图的多尺度特征提取方法。该方法通过高斯金字塔将遥感云图分解到多尺度空间,以此为基础将图像的灰度特征进行多尺度延拓,从而得到图像的多尺度特征矢量。实验结果表明在相同的特征算法和分类器条件下,多尺度延拓能够提升分类精度,更加有效地实现云图和地物的分类。

A method used for extracting multiscale features is applied to the optical remote sensing cloud im-ages.They are decomposed to multiscale space and the features of images are expanded during the process,from which a multiscale feature vector can be acquired.Provided identical feature algorithm and classifier,it is illustrated by experiment that multiscale expanded features can help raise the precision of classification,from which better performance of classifying clouds and earth targets can be achieved.