Multi-scale spatial-spectral fusion based on multi-input fusion calculation and coordinate attention for hyperspectral image classification
作者：
Highlights：
• Multi-scale spectral features and spatial features are acquired and fused.
• The three-branch and the concatenation module are used to obtain multi-scale features.
• Use coordinate attention mechanism to enhance distinguishability characteristics.
• Combine multiple input patches according to the classification effect.
• Consider accuracy and precision to fuse the output results of multiple patches.
摘要
•Multi-scale spectral features and spatial features are acquired and fused.•The three-branch and the concatenation module are used to obtain multi-scale features.•Use coordinate attention mechanism to enhance distinguishability characteristics.•Combine multiple input patches according to the classification effect.•Consider accuracy and precision to fuse the output results of multiple patches.
论文关键词：Hyperspectral image(HSI),Multi-scale fusion,Fusion calculation,Coordinate attention,Image patch,3D convolution
论文评审过程：Received 11 June 2021, Revised 12 August 2021, Accepted 21 September 2021, Available online 5 October 2021, Version of Record 10 October 2021.
论文官网地址：https://doi.org/10.1016/j.patcog.2021.108348
