In order to improve the performance of iterative least square algorithm based on image registration for nonuniformity correction of infrared focal plane arrays,a new algorithm was presented.Firstly,according to the noise regularity of actual images,the noise model of the algorithm was constructed with the knowledge of readout architecture for imaging system.Then the principle of the algorithm was provided in two parts to be analyzed.Finally,results were presented to demonstrate the efficiency of the algorithm.Compared with other correction algorithms,the algorithm has some advantages,such as fast convergence rate,low computational complexity and small memory resources.