摘要
针对遥感图像超分辨率重建算法特征利用率低、重建速度慢等问题,提出一种基于多路径特征融合的遥感图像超分辨率重建算法。该算法模型包括浅层特征提取模块、特征融合模块和图像重建模块3个部分。首先,利用浅层特征提取模块提取浅层特征信息;然后,通过级联的方式将每个多路径特征融合模块输出的特征信息进行融合,提高了特征利用率;最后,通过图像重建模块重建高分辨率图像。由于减少了基本单元中卷积层的通道数,从而提高了重建效率。实验表明,该算法在客观评价指标上相对于比较算法均有提高,重建的遥感图像主观效果更好,能够恢复出更多的地物细节信息。
Aiming at the problems of low feature utilization rate and slow reconstruction speed of remote sensing image super-resolution reconstruction algorithm,a remote sensing image super-resolution reconstruction algorithm based on multi-path feature fusion is proposed.The model includes three parts:shallow feature extraction module,feature fusion module and image reconstruction module.Firstly,the shallow feature extraction module is used to extract the shallow feature information.Then,the feature information outputting from each multi-path feature fusion module is fused through the cascade method,which improves the feature utilization.Finally,the high-resolution image is reconstructed through the image reconstruction module.As the number of channels of the convolutional layer in the basic unit is reduced,the reconstruction efficiency is improved.Experimental results show that the objective evaluation index of the proposed algorithm is improved compared with the comparison algorithm,the subjective effect of the reconstructed remote sensing image is better,and more detailed information of the ground objects can be recovered.
作者
张艳
卢宣铭
刘国瑞
刘树东
孙叶美
ZHANG Yan;LU Xuanming;LIU Guorui;LIU Shudong;SUN Yemei(School of Computer and Information Engineering,Tianjin Chenjian University,Tianjin 300384,China)
出处
《遥感信息》
CSCD
北大核心
2021年第2期46-53,共8页
Remote Sensing Information
基金
天津市教委科研计划项目(2019KJ105)。
关键词
超分辨率重建
特征融合
多路径
卷积神经网络
深度学习
super-resolution reconstruction
feature fusion
multipath
convolutional neural network
deep learning