双重注意力下的多尺度残差遥感图像去雾网络.docxVIP

双重注意力下的多尺度残差遥感图像去雾网络

目录

1.内容简述................................................3

1.1研究背景.............................................4

1.2现有方法概述.........................................4

1.3本文贡献.............................................6

1.4文档结构.............................................6

2.相关工作................................................8

2.1遥感图像去雾.........................................9

2.2注意力机制..........................................10

2.3残差学习............................................11

2.4多尺度特征处理......................................12

3.去雾问题背景...........................................13

3.1大气现象............................................15

3.2遥感图像去雾的挑战..................................16

3.3去雾方法分类........................................17

4.双重注意力机制.........................................18

4.1注意力机制原理......................................20

4.2双重注意力机制设计..................................21

4.3注意力机制在遥感去雾的应用..........................22

5.多尺度残差网络.........................................23

5.1残差学习概述........................................24

5.2多尺度特征提取......................................26

5.3残差网络的结构设计..................................27

6.网络架构设计...........................................28

6.1网络整体结构........................................29

6.2卷积层设计..........................................31

6.3全连接层与损失函数..................................32

6.4网络训练与优化......................................33

7.实验验证...............................................34

7.1实验环境与数据集....................................35

7.2实验设置............................................37

7.3性能指标............................................37

7.4结果分析............................................39

8.结果分析与讨论.........................................40

8.1网络性能评估........................................42

8.2对比实验分析........................................42

8.3多尺度残差与注意力机制的综合效果....................44

9.结论与展望..