基于BP神经网络的焦平面阵列的非均匀性校正

Nonuniformity correction of focal array system based on BP Neural network

传统的基于BP神经网络的非均匀性校正算法由于采用了四邻域均值代替期望值,使得图像呈现低通的特性.本文针对红外焦平面阵列成像系统,对传统的神经网络算法进行了改进,将加权中值滤波处理后的结果作为期望输出,并在神经网络算法中权值修正时加入了动量项,加快了算法的收敛速度.通过仿真实验,与传统的神经网络相对比,校正效果得到明显的改善.

The traditional non-uniformity correction algorithm based on the neural network uses the average of four nearest neighborhood as the expected output,which made the imaging system low pass,and the correction is not very well.This paper analyzed the infrared focal plane array imaging system,and a new improved NUC algorithm based on neural network was presented,which made the weight-medfilter as the expected output,and additional momentum item was applied to the neural network in the modified weight,which can accelerate the convergence rate.Through the simulation experiment,it shows that the improved methods are better contrast to the traditional methods.