基于Landsat影像自身特征的薄云自动探测与去除

Automatic detection and removal of thin haze based on own features of Landsat image

为有效地去除Landsat云的影响,恢复云区地物信息,基于单景Landsat影像的自身特征,通过利用影像中无云区地物的band1和band3高度相关特性确定晴空线,使用修改的最优化薄云变换算法计算受云影响的像元相对于晴空线的偏离距离(HOT),依据HOT的大小实现云的自动探测.对HOT图像进行阈值分割,将云从薄到厚进行分级;然后利用近红外和短波红外波段对云区和非云区的地物进行自动聚类,并根据云的等级和地物类型将云区可见光影像和对应地物的无云区影像进行匹配,实现薄云的去除.试验结果表明,云的探测快速准确,薄云的去除效果较好.

To remove the haze from Landsat images and restore the information on objects in haze region, the Landsat image features were used to ascertain the clear line according to the high correlation between band 1 and band 3 of Landsat image in the clear region, then the deviation distance from the clear line was calculated by the modified algorithm of haze optimized transformation （HOT）, and the haze was detected according to HOT value. The haze images were classified by the HOT value to represent the haze thickness, then the Landsat image in haze region and clear region was automatically classified using just one near-infrared band and two shortwave-infrared bands. The image in the haze region of visible bands was matched to the image in the clear region according to the haze class and object classification to remove the effect of the haze. Experimental results show that the haze detection is rapid and accurate, and that the thin haze is removed effectively.