<p>To address issues of insufficient brightness, noise interference, and detail loss in color polarization images under low-light conditions, we propose a low-light color polarization image enhancement method based on latent diffusion model. Specifically, we design an Intensity Image content decomposition network (ICDN) and a contour extraction module (CEM) to process color polarization intensity images. The ICDN performs Retinex decomposition in latent space, obtaining reflectance maps with rich details and illumination maps without content information. The CEM extracts contour features from low-light color polarization intensity images and generates conditional feature after latent space encoding. Subsequently, the reflectance maps of low-light color polarization intensity images are multiplied by the illumination maps of normal-light color polarization intensity images and used as the input of the diffusion model to perform the forward diffusion process. Finally, under the guidance of the conditional features, the reverse denoising process of the diffusion model is executed to achieve high-quality image enhancement. Additionally, we design a polarization information consistency loss function to ensure the effective preservation of polarization information during the enhancement process and constructs a dataset named POLI. Experimental results demonstrate that the proposed method outperforms mainstream algorithms in terms of quality improvement and the naturalness of enhancement effects in the task of low-light color polarization image enhancement.</p>