基于GA-RBF神经网络的光学图像背景抑制方法

Clutter removal method in optical images based on GA-RBF neural network

根据图像背景的局部平稳性,提出了通过逼近背景局部相关特性来估计图像背景的方法,并且采用 RBF 神经网络来逼近图像背景的这种局部相关性.文中根据径向基函数神经网络的训练要求,用遗传算法来训练神经网络中的各个参数。实验结果表明,训练后的神经网络背景抑制算法能够很好地抑制图像中的背景,并同时增强了目标信号。

An new method for estimating background of infrared images is proposed based on the local relativity of background.A radial basis function(RBF)neural network is used to approximate the local relativity.The genetic algorithm(GA)is introduced to Wain the parameters of RBF neural network.GA-RBF is employed for estimating background and detecting dim target in infrared images,preferable result is gained