首页|基于深度学习的多光谱与全色遥感图像融合方法综述

基于深度学习的多光谱与全色遥感图像融合方法综述

扫码查看
多光谱(multispectral,MS)遥感图像具有丰富的光谱信息,但其空间分辨率相对较低;全色(panchromatic,PAN)遥感图像具有较高的空间分辨率,但缺乏光谱信息.通过图像融合技术将多光谱和全色遥感图像进行信息集成,生成一幅光谱信息丰富、空间分辨率高的融合图像,以便将二者的优势互补,从而有利于后继视觉任务的完成.近年来,随着深度学习的兴起及其在计算机视觉领域的广泛应用,研究者们提出了许多面向图像融合的深度学习方法.鉴于国内鲜有MS和PAN遥感图像融合方面的研究综述,故对基于深度学习的多光谱和全色遥感图像融合方法进行归纳、分析和总结,并对基于深度学习的多光谱和全色遥感图像融合的发展方向进行展望.
Deep Learning-based Multispectral and Panchromatic Remote Sensing Image Fusion Methods
Multispectral(MS)remote sensing image has rich spectral information,but its spatial resolution is relatively low.By contrast,panchromatic(PAN)remote sensing image has high spatial resolution,but lacks spectral information.In practice,a fused image can be obtained by integrating the information of MS with that of PAN via image fusion technology,and the fused image contains both rich spectral information and high spatial resolution.In other words,the fused image has the complementary information coming from the MS and PAN remote sensing images,and is more suitable for down-stream vision tasks.With the rise of deep learning and its wide applications in the field of computer vision,researchers have developed numerous deep learning-based methods for image fusion tasks in the past few years.However,the survey works on MS and PAN image fusion are rarely reported in Chinese journals.To this end,according to the different learning manner of network models,this paper classifies,analyzes,and summarizes the deep learning-based methods for MS and PAN remote sensing image fusion.In addition,this paper puts forward some possible future research directions related to deep learning-based MS and PAN remote sensing image fusion.

image fusionremote sensing image fusiondeep learningconvolutional neural net-workgenerative adversarial networkTransformer

康家银、姬云翔、马寒雁、章洋洋、张文慧、王怀友

展开 >

江苏海洋大学 电子工程学院,江苏 连云港 222005

图像融合 遥感图像融合 深度学习 卷积神经网络 生成对抗网络 Transformer

2024

江苏海洋大学学报(自然科学版)
淮海工学院

江苏海洋大学学报(自然科学版)

影响因子:0.433
ISSN:1672-6685
年,卷(期):2024.33(4)