A technique for ghosting artifacts removal in scene-based methods for non-uniformity correction in IR systems

In this work we analyze the problem of the ghosting artifacts coming out from non-uniformity correction (NUC) in infrared focal-plane array (IRFPA) imaging systems. We have employed a well-established least mean square (LMS)-based NUC technique which was first introduced by D.A. Scribner. Slow global motion and edges in the scene are the main responsible of the generated ghosting artifacts that can be very damaging especially in target detection and tracking applications. To mitigate the effects of ghosting we propose to replace the linear spatial filter of the analyzed NUC scheme with a non-linear one, known in the literature as bilateral filter, which is able to preserve edges. The proposed technique has been evaluated over an infrared (IR) image sequence with simulated fixed-pattern noise (FPN). A detailed analysis of the results has shown the advantages of the novel deghosting method in terms of accuracy of the calibration and quality of the corrected frames.