Fusing infrared polarization images for road detection via denoising diffusion probabilistic models

Recent advancements in road detection using infrared polarization imaging have shown promising results. However, existing methods focus on refined network structures without effectively exploiting infrared polarization imaging mechanisms for enhanced detection. The scarcity of datasets also limits the performance of these methods. In this Letter, we present a denoising diffusion model aimed at improving the performance of road detection in infrared polarization images. This model achieves effective integration of infrared intensity and polarization information through forward and reverse diffusion processes. Furthermore, we propose what we believe to be a novel method to augment polarized images from different orientations based on the angle of polarization. The augmented polarized image serves as the guiding condition, enhancing the robustness of the diffusion model. Our experimental results validate the effectiveness of the proposed method, demonstrating competitive performance compared to state-of-the-art methods, even with fewer training samples.