Researching the theory and algorithm of infrared dim and small target detection isof profound significance for the battle now and in the future, and it is beneficial for thefurther development of precision-guided weapon. In this paper, we study the temporalcharacters of pixels in infrared images, and the target, the background and the clutterpixels are analyzed and modeled respectively. Based on all above, according to thedifferences of their temporal characters, a new detecting algorithm is proposed in thispaper.The detecting algorithm in this paper is divided into three parts. Firstly, accordingto the differences of temporal variance for different kinds of pixels, a filtration based ontemporal variance is employed to preprocess input image, thus smooth background isremoved; secondly, for the remained pixels, we exploit the connecting line of thebottom stagnation points of the temporal profile as its baseline, and the residualtemporal profile is obtained, as a result, the large fluctuations of clutter pixel in itstemporal profile is eliminated and the pulse signal of target is remained simultaneously;finally, we calculate the cross product of residual temporal profile, and the problem oftarget detection is transformed into the detection of the cross product, consequently themoving trace of targets is acquired and the clutter pixels are eliminated.Experimental results show that the algorithm presented in this paper performs wellfor various background, and has practical value.... |