HDCTfusion: Hybrid Dual-Branch Network Based on CNN and Transformer for Infrared and Visible Image Fusion DOI Creative Commons
Wenqing Wang, Lingzhou Li,

Yifei Yang

и другие.

Sensors, Год журнала: 2024, Номер 24(23), С. 7729 - 7729

Опубликована: Дек. 3, 2024

The purpose of infrared and visible image fusion is to combine the advantages both generate a fused that contains target information has rich details contrast. However, existing algorithms often overlook importance incorporating local global feature extraction, leading missing key in image. To address these challenges, this paper proposes dual-branch network combining convolutional neural (CNN) Transformer, which enhances extraction capability motivates contain more information. Firstly, module with CNN as core constructed. Specifically, residual gradient used enhance ability extract texture Also, jump links coordinate attention are order relate shallow features deeper ones. In addition, based on Transformer Through powerful context can be captured fully extracted. effectiveness proposed method verified different experimental datasets, it better than most current advanced algorithms.

Язык: Английский

HDCTfusion: Hybrid Dual-Branch Network Based on CNN and Transformer for Infrared and Visible Image Fusion DOI Creative Commons
Wenqing Wang, Lingzhou Li,

Yifei Yang

и другие.

Sensors, Год журнала: 2024, Номер 24(23), С. 7729 - 7729

Опубликована: Дек. 3, 2024

The purpose of infrared and visible image fusion is to combine the advantages both generate a fused that contains target information has rich details contrast. However, existing algorithms often overlook importance incorporating local global feature extraction, leading missing key in image. To address these challenges, this paper proposes dual-branch network combining convolutional neural (CNN) Transformer, which enhances extraction capability motivates contain more information. Firstly, module with CNN as core constructed. Specifically, residual gradient used enhance ability extract texture Also, jump links coordinate attention are order relate shallow features deeper ones. In addition, based on Transformer Through powerful context can be captured fully extracted. effectiveness proposed method verified different experimental datasets, it better than most current advanced algorithms.

Язык: Английский

Процитировано

1