An infrared and visible image fusion method based on latent low rank representation and Karhunen-Loeve transform

Abstract:
Infrared and visible light image fusion can make up for the shortcomings of a single imaging method, and generate a fusion image that can not only retain the infrared thermal radiation information but also contains the detailed information of visible light. Due to the difference between the two imaging methods, extracting more detailed information from visible light images and more thermal radiation information from infrared images is the key to the fusion technology. This paper proposes a TDLatLRR image feature extraction method, and then proposes a new infrared and visible light fusion framework based on TDLatLRR. First, the infrared and visible light images are decomposed by TDLatLRR to obtain the detailed part and the basic part. Then the Karhunen-Loeve Transform strategy is used for the infrared and visible light details, and the basic part weighted average strategy is fused to obtain the fused detailed image layer and basic image layer. Finally, the fused detail image layer and the basic image layer is linearly combined to generate the final fused image. The experimental results show that this method is better in detail preservation and edge texture information processing, and the objective evaluation index is also improved.