基于MPMAP序列红外图像高分辨力重建和非均匀性校正

摘要

红外焦平面阵列(IRFPA)的非均匀性校正是获得高性能热成像的基本保证,非均匀性校正(NUC)算法是当前国内外研究的重要方向.鉴于序列图像的超分辨力复原方法和基于场景的NUC算法都需要存在微位移的多帧序列目标场景图像,本文在Poisson和Markov分布假设的基础上,将超分辨力复原与NUC结合,针对存在非均匀性的红外低分辨力图像序列,研究提出了基于Poisson- Markov分布的最大后验概率(MPMAP)的序列红外图像高分辨力重建和非均匀性校正一体化处理算法.对仿真和实际热图像的处理结果表明,算法不仅可有效抑制随机噪声,实现高分辨力图像的复原,而且可有效地消除非均匀性噪声.%Non-uniformity correction(NUC)of infrared focal plane array(IRFPA)is the development direction of the future and the basic guarantee to gain high quality thermal imaging system. As both super-resolution restoration and NUC algorithm based on scene need the micro-displacement that between image sequences, we propose super-resolution image restoration and non-uniformity correction algorithm based on Poisson and Markov model maximum a posteriori (MPMAP) focus on infrared low resolution image sequences with non-uniformity noise. The results of simulator image sequences and real infrared thermal image sequences show that the algorithm is presented not only has high super-resolution performance for the image degraded with random noise, but also eliminate the fixed pattern noise effectively.