Infrared radiation and LiDAR image fusion method

Abstract

To address the limitations of single-modality optical imaging data and the computational complexity of traditional fusion methods, which rely heavily on deep learning models, this study investigated methods for fusing infrared radiation and LiDAR images. First, a feature-based adaptive weighted fusion algorithm was proposed. By extracting features from both image types to calculate weights and dynamically allocating weights based on evaluation metrics, the algorithm effectively enhances the clarity and completeness of the fused images. Subsequently, comparative experiments were conducted using various typical material targets to validate the algorithm’s effectiveness and robustness. The experimental results demonstrate that the algorithm does not require massive training samples and achieves excellent fusion results for individual small targets. These findings enrich the methodological framework for heterogeneous image fusion and provide a new lightweight option for target detection and recognition.