Scene-Based Nonuniformity Correction with Multiframe Registration

A scene-based nonuniformity correction method for infrared focal plane arrays based on multiframe registration has been undertaken in this paper. It is based on the estimation of global translation between several adjacent frames, and the resulting mean square error function is optimized using a least mean square (LMS) algorithm. The proposed method make use of the relation of adjacent frames sufficiently, and it provides enhanced results when applied to infrared image sequences with simulated nonuniformity and infrared imagery with real nonuniformity. The experimental results have shown that it can estimate the nonunifomity parameters accurately of each detector with fast convergence speed and almost no ghosting artifacts.