矩匹配与卡尔曼滤波在红外图像非均匀性校正应用中的比较研究

Comparison Between Applications of the Moment Matching and the Kalman Filter Methods in Infrared Non-uniformity Correction

红外焦平面阵列的非均匀性校正是其定量应用之前必须解决的关键难题之一。将基于场景的矩匹配和卡尔曼滤波方法引入到红外图像非均匀性校正中,通过实验对比分析了这两种方法的校正效果。定性与定量分析结果表明:这两种方法各具优缺点,矩匹配方法在低值区会产生新的条带并改变了图像上同一探测器中不同地物的光谱分布差异,而卡尔曼滤波需经过多帧图像数据迭代后,才能取得更好的校正效果。

Non-uniformity of infrared focal plane arrays is a key problem that must be solved before its quantitative applications. The moment matching and the Kalman filter methods are adopted in correction of infrared non-uniformity, which are compared by the experimental image. The quantitative and the qualitative analysis results show that each method has its advantages and disadvantages respectively. The new stripe noises might occur and the image spectral variance may be changed after the moment matching correction, while the Kalman filter method could get a better correction result only after several frames iterations.