Cloud cover estimation by analysing SPOT monochromatic images

Cloud detection is a very important part of Earth-atmosphere system study. The results of these studies lead to a lot of implications for climatology knowledge. There are also satellite image users who want to work with cloudless images. An automatic sorting technique is needed to separate a cloudless image from cloudy one by affecting it with a certain cloudiness index. The authors describe a technique called the 'spatial coherence method', which is based on the hypothesis that the cloudy structures are 'irregularly shaped and structures highly structured objects against a relatively uniform background'. They build a monodimensional histogram of small local standard deviations of radiances.<
>