The traditional neural network correction does not fit scanning-type infrared imaging system.So an improved algorithm based on the traditional neural network has been proposed,which makes the median of nearest neighbors as the hidden layer outputs.New algorithm corrects values of coefficient row-by-row and has an increased convergence speed.Theoretical analysis and the experiment with real infrared data show the algorithm can eliminate the striping fixed pattern noise effectively.
参考文献
相似文献
引证文献
引用本文
郑逢勋,吉书鹏.一种改进的神经网络非均匀性校正算法[J].激光与红外,2008,38(9):937~938 ZHENG Feng-xun, JI Shu-peng. An Improved Nonuniformity Correction Algorithm for IRFPA Based on Neural Network[J]. LASER & INFRARED,2008,38(9):937~938