基于MPMAP序列红外图像高分辨力重建和非均匀性校正  

High-Resolution Reconstruction and Non-Uniformity Correction from Images Sequences Based on Poisson-Markov Model MAP

作　　者：刘秀[1] 金伟其[1] 徐超[1] 

机构地区：[1]北京理工大学光电学院光电成像技术与系统教育部重点实验室,北京100081

出　　处：《电子学报》2011年第9期2103-2107,共5页Acta Electronica Sinica

基　　金：国家自然科学基金(No.60877060);十一五总装预研项目(No.404050301)

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.

关 键 词：红外图像 序列图像 微位移运动估计 非均匀性校正 图像复原 最大后验概率 

分 类 号：TN911.73[电子电信—通信与信息系统]