The performance of an infrared imaging system is affected by spatial fixed-pattern noise

which is mainly due to the different photoresponse of each detector in a focal plane array for the same irradiance. In addition

the nonuniformity of the readout

which is involved in multichannel image data processing

causes the further noise. A novel nonuniformity correction algorithm based on the infrared focal plane array readout architecture is developed

which corrects the aggregate nonuniformity by two separate steps. First

the nonuniformity from the readout amplifiers is corrected by using the constant statistics method. Second

the noise resulting from the detectors is eliminated by exploiting the improved neural network method. The strength of the proposed algorithm lies in its high convergent speed and low computational complexity

and the performance is demonstrated with both simulated and real infrared imagery.