Non-uniformity correction algorithm for IRFPA based on local scene statistics and improved neural network

In this paper a new non-uniformity correction algorithm is proposed which is based on local scene statistics and improved neural network. The new algorithm firstly uses local scene statistics to filter out low frequency noise over infrared image and reduce the ghosting artifacts of a previously developed scene-based non-uniformity correction method, then uses improved neural network algorithm to correct the non-uniformity of infrared focal plane array (IRFPA). Experiments show that the proposed algorithm can effectively filter out low frequency noise and improve corrected image quality.