In the field of modern image processing and computer vision, multi-modal image fusion technology has emerged as a promising research direction. By leveraging the unique advantages of different modalities, it can produce more comprehensive and enriched visual information. However, existing fusion methods often fail to fully capture key features from both modalities, leading to information loss or distortion in the fused output. To address challenges in infrared and visible light image fusion, we propose an innovative solution. Our method combines multi-scale feature extraction with a dynamic weight adjustment strategy, adjusting the weights of modal features based on input characteristics. This approach achieves optimal fusion results across complex scenarios. Experimental results demonstrate that MSADFusion performs on par with or surpasses existing state-of-the-art methods across multiple evaluation metrics. It enhances the quality and information retention of fused images. Our study provides new insights into multi-modal image fusion and has significant implications for advancing related technologies.