FLIR mainly treats with the objects such as bridges,airdromes and big buildings etc,which are based on complicated background on earth. The general method for tracking these kinds of targets is correlation matching by standard templates.The key technique of matching in target tracking is how to renew and displace the templates rightly. In this paper, the method of correlation matching is adopted to search targets, Kalman filtering was used to renew every pixels in the templates to be matched and to correct the resulting position of target tracking. To minimize the computing time and insure the precision of tracking at the same time, the templates are renewed and the tracking position is corrected once every several frames.Simulation results show that tracking effect is good.