<p>Image feature matching is a key technique in computer vision to identify corresponding points between two images. The LightGlue-based method performs well in certain conditions but struggles with subtle differences in objects with large size and shape variations, making accurate matching challenging in complex environments.To address the above issues, we propose an adaptive multi-awareness network for feature matching. First, we design a scale-aware module to adaptively enhance the feature representation by aggregating features from different scales based on the concept of self-attention mechanism. Next, we devise an adaptive differential cross-attention module. This module comprehensively considers the similarities and differences between images, which enhances the model’s ability to distinguish key features and thus improves the accuracy of matching. Furthermore, we exploit a non-parametric attention mechanism to strengthen crucial features and adaptively optimize feature integration. Experiments on different datasets show that our method achieves good results and has advantages in feature matching.</p>