Practical texture mapping on free-form surfaces

Concerns the problem of mapping photographic images on to the surface of a 3D object whose geometric data are already known. Although the minimal view selection algorithm of K. Matsushita and T. Kaneko (1999) enables the selection of a minimal number of camera positions, it often results in untextured patches due to camera misalignment: a deviation of actual camera position from the specified position. This paper proposes two new practical algorithms (double and product viewing) to ensure that all the polygons are textured even if there are unavoidable human errors in camera alignment. The performance of the algorithms is evaluated by simulation experiments.