<p>Despite advancements in image fusion methods, existing techniques often fail to effectively preserve high-resolution details, especially in complex images. In this paper, we propose a novel method based on Bayesian pre-fusion and multi-scale transformations to address this challenge and enhance detail preservation. The method first uses a Bayesian-based pre-fusion technique to create a pre-fused image, preserving the main structure and information, providing a solid foundation for subsequent steps. Next, infrared and visible images are decomposed with the multilayer decomposition latent low-rank representation (MDLatLRR) method to extract the base layers, which are fused using a multi-level optimal fusion algorithm. MDLatLRR aims to maximize key information retention by decomposing low-rank features and detail layers. Then, using the pre-fused image, a visible light factor derived from the Structural Similarity Index (SSIM) is introduced in an L2-norm optimization to generate the final detail fusion layer. SSIM ensures structural similarity, preserving critical details and minimizing fusion errors. Finally, inverse transformations are applied to the fused base and detail layers to obtain the final fused image. Experimental results validate the effectiveness of our method, demonstrating superior performance in high-resolution detail preservation and edge contour clarity compared to 11 state-of-the-art fusion methods across four public datasets. The source code of our developed method is available at:<a href="https://github.com/YangZhengrun/A-detail-preservation-fusion-framework-for-infrared-visible-images-via-Bayesian-and-MDLatLRR.git">https://github.com/YangZhengrun/A-detail-preservation-fusion-framework-for-infrared-visible-images-via-Bayesian-and-MDLatLRR.git</a></p>