基于神经网络的船舶航拍图像融合研究

Research on fusion of ship aerial image based on neural network

航拍图像融合是船舶图像研究领域中的一个重要研究方向,针对当前船舶航拍图像融合过程复杂,计算量大,融合精度低等缺陷,提出了基于神经网络的船舶航拍图像融合方法。首先分析船舶航拍图像融合原理,并对船舶航拍图像进行分解,得到低频和高频的船舶航拍图像分量,然后采用神经网络对船舶航拍图像的高频分量进行处理,并用加权平均法对船舶航拍图像的低频分量进行处理,最后通过逆变换对船舶航拍图像融合,并进行了仿真实验。实验结果表明,神经网络可以获得高精度的船舶航拍图像融合结果,船舶航拍图像融合时间短,相对于其他船舶航拍图像融合方法,神经网络的船舶航拍图像融合结果更胜一筹。

Aerial image fusion is an important research direction in the field of ship image research.Aiming at the defects of complex fusion,large computation and low fusion accuracy,a method of ship aerial image fusion based on neural network is proposed.Firstly,the principle of ship aerial photography image fusion is analyzed,and ship aerial photography image is decomposed to get low-frequency and high-frequency components of ship aerial photography image.Then,the high-frequency components of ship aerial photography image are processed by neural network,and the low-frequency components of ship aerial photography image are processed.Finally,ship navigation is processed by inverse transformation.The image fusion was taken and the simulation experiment of ship aerial image fusion was carried out.Neural network can obtain high-precision ship aerial image fusion results.The time of ship aerial image fusion is short.Compared with other ship aerial image fusion methods,the result of ship aerial image fusion based on neural network is better.