The restoration of museum heritage is an important task with significant cultural and historical value; however, traditional methods of restoration are frequently constrained by the extent of the damage to the heritage as well as the constraints of the restoration techniques. In recent years, a method of restoration known as the diffusion model, which is based on computer vision and machine learning, has been gradually applied to the field of museum relics restoration and has shown enormous potential in relic restoration. This method of restoration was developed in the 1980s and is still in use today. In the field of image restoration, research and application of diffusion models are reviewed, and this article provides a summary of the development history, methodology principles, and application cases associated with these topics. This article provides an introduction to the fundamental ideas and principles underlying diffuse models, as well as a summary of the current state of research and an outlook on potential future trends and prospects pertaining to image repair within the domain of image processing.