Aiming at the big computational complexity and memory requirement of infrared focal plane non-uniformity correction algorithm based on Kalman filtering, it cannot realize real-time correction shortage, an improved Kalman filtering non-uniformity correction algorithm is given. The proposed algorithm modified the observation model by applying linear recursion filter that the observation was instead of the mean value of every block image, then the matrix will be simplified. The simulation of Matlab proves that the computational complexity and memory requirements are significantly reduced, while the correction result is similar.