基于神经网络的船舶航拍图像融合研究.pdf

基于神经网络的船舶航拍图像融合研究

魏峰

(新乡职业技术学院，河南新乡453006)

摘 要: 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。

Abstract:Aerialimagefusionisanimportantresearchdirectioninthefieldofshipimageresearch.Aimingatthede-

fects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com-

ponents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ob-

tain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.

2019年10月 SHIPSCIENCEANDTECHNOLOGY Oct.,2019