场景自适应的红外焦平面阵列非均匀性校正新方法

提出了一种基于神经网络的红外焦平面阵列 (IRFPA) 非均匀性自适应校正方法，重新设计了隐含层结构以获得更接近真实信号的期望信号，此外，还将变步长归一化LMS (VSS-NLMS)自适应滤波技术引入到校正参数的迭代估计过程中，以实现对校正速度和稳定性的提升．实验结果表明，该方法在校正精度、收敛速度和稳定性方面的性能均显著优于传统的神经网络校正算法．

Novel scene adaptive method for non-uniformity correction in infrared focal plane arrays

A neural network based scene-adaptive non-uniformity correction (NUC) algorithm for infrared focal plane arrays (IRFPA) is proposed. This method redesigns the structure of the hidden layer, which is helpful to estimate the desired target value more precisely. In addition, the variable step size normalized LMS (VSS-NLMS) based adaptive filtering technique is introduced to the correction parameter estimation process, which yields faster convergence and higher stability. Anyway, the proposed NUC method achieves high performance in correction precision, convergence rate and stability, which is validated by experimental results.