Integrating Parallel Attention Mechanisms and Multi-Scale Features for Infrared and Visible Image Fusion

Abstract

Infrared and visible image fusion (IVIF) aims to synthesize images that capitalize on the strengths of both modalities. Addressing the common challenge in IVIF of preserving thermal radiation from infrared and textural details from visible images, we introduce AMFusionNet. AMFusionNet uniquely combines a multi-kernel convolution block (MKCBlock) with parallel spatial attention and channel attention modules (PSCNet), streamlining the feature extraction process. This integration enhances the model’s ability to simultaneously capture essential details from both image types. Additionally, we incorporate a multi-scale structural similarity (MS-SSIM) loss function in our comprehensive loss function to further refine the detail preservation in the fused images. Our experimental evaluations on the TNO and FLIR datasets demonstrate that AMFusionNet achieves superior performance in both objective and subjective assessments compared to recent methods.

Keywords

• Infrared and visible image fusion
• parallel attention mechanism
• multi-kernel convolution
• MS-SSIM