A comparison of deghosting techniques in adaptive nonuniformity correction for IR focal-plane array systems

Focal-plane array (FPA) IR systems are affected by fixed-pattern noise (FPN) which is caused by the nonuniformity of the responses of the detectors that compose the array. Due to the slow temporal drift of FPN, several scene-based nonuniformity correction (NUC) techniques have been developed that operate calibration during the acquisition only by means of the collected data. Unfortunately, such algorithms are affected by a collateral damaging problem: ghosting-like artifacts are generated by the edges in the scene and appear as a reverse image in the original position. In this paper, we compare the performance of representative methods for reducing ghosting. Such methods relate to the least mean square (LMS)-based NUC algorithm proposed by D.A. Scribner. In particular, attention is focused on a recently proposed technique which is based on the computation of the temporal statistics of the error signal in the aforementioned LMS-NUC algorithm. In this work, the performances of the deghosting techniques have been investigated by means of IR data corrupted with simulated nonuniformity noise over the detectors of the FPA. Finally, we have made some considerations on the computational aspect which is a challenging task for the employment of such techniques in real-time systems.