Dim target detection in cloud clutter image based on variogram function and rough set

A new method is provided for detecting dim targets in cloud clutter image sequences. The method is based on the temporal domain, where there are clear distinctions between targets and cloud clutter, and much similitude between cloud clutters. Variogram function can estimate the randomicity of the region variation and the structure of special distribution. It is good for value the statistical characters of adjacent points in image. The different texture should have different value of single step variogram function, so it is used for detecting the target from the cloud clutter in this paper. On the other hand, if the gray values of the adjacent pixels are higher than a threshold, they always belong to the same objects. So a detection model of weak targets in cloud clutter background is established based on indiscernibility relation of rough sets theory, according to variogram functions and gray values. Application to real IR image sequences shows that the method is very feasible and effective to discriminate the man-made objects from the nature cloud. The total signal-to-clutter-ratio (SCR) can lower than 2, but the local signal-to-clutter-ratio ( LSCR ) of 32times32 domain requires ges 2 for detection probability exceeding to 90%.