One of the necessary techniques for constructing a virtual museum is to estimate the surface normal and the albedo of the artwork which has high specularity. In this paper, we propose a novel photometric stereo method which is robust to the specular reflection of the object surface. Our method can also digitize the artwork arranged inside a glass or acrylic display case without bringing the artwork out of the display case. Our method treats the specular reflection at the object surface or at the display case as an outlier, and finds a good surface normal evading the influence of the outliers. We judiciously design the cost function so that the outlier will be automatically removed under the assumption that the object's shape and color are smooth. At the end of this paper, we also show some archived 3D data of Segonko Tumulus and objects in the University Museum at The University of Tokyo that were generated by applying the proposed method.
All Science Journal Classification (ASJC) codes
- Computer Vision and Pattern Recognition
- Artificial Intelligence