一种面向高分五号高光谱影像的云检测算法.

Cloud cover affects the quality of hyperspectral images (HSIs) and reduces the application value of HSIs. Therefore, cloud detection has become a main issue that need to be solved in HSIs applications. According to the characteristics of GF5 HSI, a cloud detection algorithm that combines cloud visual features and physical spectrum features was proposed. Based on the original digital number(DN) image, the cloud distribution feature in the HSV color space was used, combining with the spectral feature difference between the cloud and the typical ground objects, a cloud probability image was constructed. Then, the cloud detection threshold was dynamically set to realize the cloud detection of GF5 HSI. The overall cloud recognition accuracy of the algorithm is more than 87%, and the algorithm is less time-consuming and highly automated. It has certain adaptability to GF5 HSIs in different regions and at different times, and is convenient for engineering applications. [ABSTRACT FROM AUTHOR]