提出了一种基于神经网络的红外焦平面阵列(Infrared Focal Plane Array, IRFPA)非均匀性自适应校正算法。首先，利用归一化思想对图像进行 处理以利于选取迭代步长; 其次，优化隐含层结构以获得更接近于真实信号的期望信号。实验结果表明，该方法在校正精度、收敛速 度和稳定性方面均优于传统的神经网络校正算法。

A non-uniformity correction algorithm based on neural network for infrared focal plane arrays (IRFPA) is proposed. Firstly, a normalized idea is used to process an image so as to choose a proper interactive step. Then, the structure of the hidden layer is optimized to obtain the desired signal more close to the real signal. The experimental result shows that the proposed method is better than the traditional neural network correction algorithm in correction precision, convergence rate and stability.