ABSTRACT
The non-uniformity of the response between the detectors of the infrared detector is large, which seriously affects the imaging quality. And the detector response parameters will slowly drift, making it impossible to completely solve the non-uniformity problem through a single calibration method. Traditional adaptive correction algorithms require sufficient scene motion, otherwise there will be some problems such as degradation or non-convergence. This paper proposes a single-frame infrared image adaptive correction algorithm based on residual network. This method uses residual network to obtain the non-uniformity responsive model and brings in batch normalization to solve the problem of gradient disappearance. Compared with other several scene-based algorithms, this algorithm can achieve single-frame infrared image non-uniformity correction, solves the ghosting problem, and has a high degree of detail retention.
