Lu, MinZheng, BoTakamatsu, JunNishino, KoIkeuchi, KatsushiFranco Niccolucci and Matteo Dellepiane and Sebastian Pena Serna and Holly Rushmeier and Luc Van Gool2013-10-312013-10-312011978-3-905674-34-71811-864Xhttps://doi.org/10.2312/VAST/VAST11/161-168The Bayon temple is known for its numerous massive stone faces with serene smiles, often referred to as the 'Khmer Smile.' Many of these sculptures are, however, only partially preserved, making it difficult to see the original appearance of these faces. To restore the Bayon faces, we propose a novel method that builds upon the matrix recovery theory. The method achieves accurate restoration by adopting a two-step shape recovery strategy. Rough restoration and clustering processes are first carried out using the entire database to group similar samples together. Then refined restoration using high resolution data is executed in each cluster to restore higher details while retaining the characteristics of each face. Experimental results demonstrate the effectiveness of our proposed method.Categories and Subject Descriptors (according to ACM CCS): I.3.5 [Computer Graphics]: Computational Geometry and Object Modeling-Curve, surface, solid, and object representations