为了提高光谱遥感图像数据的利用率, 提出了一种基于光谱反射率特性的星载高光谱图像云检测方法。首先分析了各类地物在不同光谱下的反射率特征, 根据其特点从星载高光谱图像中选取特定5个波长的图像数据, 然后依照图像地物反射率及阈值判定法区分出高云区和中低云区, 最后对二者合并, 得到目标云区。试验结果表明, 该方法的云检测率可达90%以上, 虚警率在1%以内, 可以准确区分云与雪地区域、沙漠区域以及植被区域, 实现云检测。

To improve the effectiveness of remote-sensing image data, a cloud detection method based on the reflectivity of hyperspectral remote image is proposed. According to the characteristic of the reflectivity between different earth background, image data of five specific wavelengths are chosen, then high-cloud area and low-cloud are determined by judging the threshold of different spectrum. The object cloud area can be obtained by combining the high-cloud and low-cloud areas. The experiment result indicates that the accuracy of cloud detection is more than 90%, while false alarm ratio is less than 1%, which proves that the cloud area can be accurately identified from multiple types of earth background.