Accurate Non-effective Pixel Detection and Replacement Based on Multi-Temperature Matching

摘要

The number and distribution of non-effective pixels is an important quality figure that defines a given infrared focal panel array (IRFPA). An accurate non-effective pixel detection algorithm is present in this paper. Through theoretical analysis of IRFPA responsivity suitable definition of non-effective pixel is given. Based on this definition an adaptive threshold is proposed to discriminate various non-effective pixels and normal pixels. Meanwhile, multi-temperature matching approach helps us to pick out the hided non-effective pixel under a certain temperature range. Finally, neighboring pixel interpolation is performed to substitute non-effective pixels according to spatial correlation. The benefit of this method is reducing misjudgment of non-effective pixels, which will degrade infrared image quality. This approach for detecting and replacing non-effective pixels is successfully applied to a set of frames obtained from an IRFPA imaging. Results show that the detection and replacement accuracy of non-effective pixels is greatly improved by this approach. Furthermore, the proposed algorithm is adaptable to a variety of non-effective pixel types.