Imaging all Visible Surfaces Inproceedings

Abstract:

Today many systems exist to generate geometric models of existing scenes and objects. However, very few systems store accurate data about surface appearance such as colors and textures. One way to capture surface texture data is to record a series of images that, collectively, captures all visible surfaces of the object. Finding good viewpoints for this task is not easy.
This paper presents a new heuristic method to find a good set of viewpoints for a given geometric model. Taking images from the computed viewpoints will show every visible part of every surface at least once. The approach uses a hierarchical visibility algorithm to preprocess the scene. A good set of viewing regions is identified with simulated annealing and then good viewpoints are derived. Results and visualizations of the computed solutions are presented.