Photometric Stereo

Photometric stereo is a computer vision technique that reconstructs 3D surface shape from multiple images of an object taken under varying lighting conditions. Current research emphasizes improving accuracy and efficiency, particularly through the development of deep learning-based methods, including neural networks that leverage multi-view data, attention mechanisms, and physically-based rendering models to handle complex reflectance properties and challenging lighting scenarios. These advancements are significant for applications ranging from cultural heritage preservation (e.g., analyzing ancient artifacts) to robotics and autonomous navigation (e.g., 3D mapping of small celestial bodies), where accurate and efficient 3D shape recovery is crucial.