基于稳态卡尔曼滤波的红外焦平面阵列非均匀校正算法  

Nonuniformity Correction Algorithm of Infrared Focal Plane Arrays Based on Steady-State Kalman Filtering

作　　者：刘永进[1] 朱红[1] 赵亦工[1] 

机构地区：[1]西安电子科技大学模式识别与智能控制研究所,陕西西安710071

出　　处：《光学学报》2008年第9期1697-1702,共6页Acta Optica Sinica

基　　金：国家自然科学基金(60572151)资助课题

摘　　要：传统的卡尔曼滤波算法利用批处理实现非均匀校正,这样不仅会带来较大的计算量和存储量,而且更重要的是该方法不能实时处理。为此,提出了一种基于稳态卡尔曼滤波的红外焦平面阵列非均匀校正算法,它可以根据固定图案噪声的特点离线计算出滤波器的增益矩阵,并且非均匀校正过程也采用了逐帧迭代方式,因此大大降低了计算复杂度和存储占用量。详细阐述了该算法的基本原理,并用仿真数据和真实红外数据对算法性能进行验证,实验结果表明:本方法平均校正一帧图像所需CPU时间和内存占用量分别为1.7188 s和131.25 KB,完全可以满足实时处理的要求。The traditional Kalman-filtering algorithm uses batch processing to realize nonuniformity correction, and causes high computational complexity and memory requirement, and it is more important is that it cannot realize real- time processing. For this reason, a novel nonuniformity correction algorithm based on steady-state Kalman filtering is developed, which can compute the gain matrix of the filter off line according to characteristics of fixed-pattern noise and correct images by using a frame-by-frame iteration. Therefore, this algorithm has lower computational complexity and memory requirements. The fundamental of the proposed technique is described in detail and the performance is demonstrated with both simulated and real infrared imagery. Experimental results indicate that the CPU time and memory requirements that the proposed algorithm needs to correct a frame of image are 1. 7188 s and 131.25 KB respectively, which are very suitable for real-time processing.

关 键 词：红外焦平面阵列 非均匀校正 固定图案噪声 卡尔曼滤波 

分 类 号：TN215[电子电信—物理电子学]