Infrared small target detection based on the block prediction of background

A new approach is presented for the detection of infrared small target in the single-frame cloud image, based on the block prediction of background. The new approach estimates the background of image with block prediction algorithm. Both the rule for reference window selecting and the method for weight matrix updating adopted by the new technique can improve the accuracy of background prediction, which can effectively realize the separation of target and complex background. After the separation, the target can be detected from the noise and the residual components of background using CFAR algorithm. The results presented show that the new method behaves superior performance for the detection of infrared small target in the single-frame image with complex cloud background.