Scenario-based Non-uniformity Correction

Infrared Focal Plane Array device is the most advanced and most commonly used infrared imaging system detector, but its inherent non-uniformity has seriously affected the imaging quality, so the non-uniformity correction becomes one of the key technologies in the imaging processing. Because scene-based non-uniformity correction doesn't need calibration equipment, it makes the system the advantages of continuous research in this area and has become a hot spot. In this paper, scene-based non-uniformity correction algorithm was in-depth studied, and the ghost-artifact issues were discussed.First of all, this paper introduced an infrared focal plane array devices and the mechanism of the non-uniformity, explained the characteristics of the infrared image's non-uniformity, and described the detector response model and the based correction model. Then on the base of the analysing of the non-uniformity correction's meaning, briefly introduced the two-point nonuniformity correction method and the constant-statistics constraint method, and described the advantages of the scene-based non-uniformity correction algorithm.Then identified the scene-based algorithms by two main approaches, namely, statistical methods and registration-based methods. Focused on the temporal high-pass filter technique and neural-network-based NUC:analyzed the advantages and disadvantages of these algorithms. Some improved algorithms were presented here, which can get better convergence speed, correction effect and practicality. Simultaneously, subjective evaluation and objective evaluation were done about this algorithms. The results can illustrate they have better perfprmance.Finally, the generation of ghost-artifact and some de-ghost algorithms were analyzed.