The drawbacks of traditional temporal high pass filter were "ghost artifacts" and fixed pattern noise can′t be removed completely. A non-uniformity correction algorithm that combines weighted guided filter and improved temporal high pass filter was proposed. Firstly, weighted guided filter was used to separate spatial high frequent components from infrared images accurately. Then, the change amplitude of every pixel value was calculated. Finally, different time constants were applied to motion regions and static regions to conduct non-uniformity correction. Two real infrared sequences were adopted in experiments, and space low-pass and temporal high-pass(SLTH) as well as bilateral filter based temporal high-pass filter(BFTH) were used to compare with the proposed algorithm. The experimental results show that the proposed algorithm is superior to the other two algorithms in subjective visual effect and objective evaluation inder. The proposed algorithm can reduce non-uniformity without causing "ghost artifacts" and achieves a better effect of non-uniformity correction.