【摘要】 文章提出一种小波神经网络的遥感图像融合算法,在特征级对遥感信息进行融合。并针对小波网络的“维数灾”问题,通过采用改进的遗传算法对时频相位点进行优化选择,从而确定小波网络的隐层节点数目乃至整个小波网络。仿真结果表明利用该方法设计的小波网络具有网络结构小、收敛速度快的优点,证明了该方法在处理遥感信息上优于经典算法。更多还原

【Abstract】 The method of remote information fusion that is based on wavelet neural network is brought forward,which fuse remote sensing information in characteristic class.Therefore using the improved GA optimized the time frequency that is aim at the problem of "dimension disaster" ,thereby it make sure the conceal layer node of whole wavelet network.The result of emulator indicate that the make sure of this wavelet neural network have excellent on simple network structure and quickly speed on convergence ,which testify the method is better than sutra arithmetic.更多还原