For reducing the nonuniformity effect of readout circuit and detectors,a new improved recursive least square (RLS)nonuniformity correction algorithm is presented to improve correction accuracy and convergent speed.Firstly,the nonuniformity of focal plane array detectors is modeled and simulated,and the nonuniformity from readout circuit is corrected by using the constant statistics method.Then,the images are preprocessed by using RAMF algorithm to provide lower noise images for RLS algorithm.The nonuniformity correction is achieved with RLS algorithm.The simulation results indicate that the proposed algorithm can effectively suppress the influence of readout circuit on the calibration accuracy,and eliminate the nonuniformity of the image.The pretreatment using RAMF algorithm can accelerate convergent speed of RLS algorithm,improve the signal-to-noise ratio,and acquire good NUC effect.