Seam Elimination in Free-Viewpoint Video

Abstract

Free-viewpoint video (FVV) is a term for any system that lets users freely adjust the viewpoint in a video to novel positions, synthesized from source data in the video stream. Recent advances in the area by K ampe et al. suggest encoding the geometric information separately as dynamic 3D voxel data, together with RGB-video streams from a set of standard RGB cameras capturing the scene from individual view points. During play back, the 3D voxel data is rendered and the color information from the RGB cameras are back-projected onto this geometry, to provide the nal result per frame. In the setting of a voxelized geometry and a number of RGB camera streams, some graphical artifacts may occur if the images of the cameras are projected onto the geometry with equal weights. We refer to this as the naive method. To alleviate these issues, three methods are proposed, all building on calculating weights for use in a weighted averaging when calculating the final color of a piece of geometry. The most promising of the proposed methods is implemented and tested against an implementation of the naive method. The results show that our method reduces the number of graphical artifacts while keeping the processing time low.