Towards a free viewpoint and 3D intensity adjustment on multi-view display

Abstract

As often declared by customers, wearing glasses is a clear limiting factor for 3D adoption in the home. Autostereoscopic systems bring an interesting answer to this issue. These systems are evolving very fast providing improved picture quality. The system we describe here is able to generate multi-view content for auto-stereoscopic displays adapted to some user requirements. The viewpoint and the 3D intensity are adjusted at the rendering side taking advantage of available texture and disparity maps provided by the heterogeneous network . A GPU-based solution is proposed that ensures real-time performances. Dedicated work has been performed exploiting network capabilities to satisfy the user request (viewpoint, 3D intensity) but still with an optimization of the bandwidth. Finally, proposals for evaluation methodology have been given.Index Terms-auto-stereoscopic, multi-view, depth based image rendering, sub pixel shuffling

Sparse and Noisy to Dense Depth Map Upsampling Based on Mesh and Colour Consistency

“…formation is scaled to have the same resolution as the original one by enforcing upsampling processes, for which various sparse-to-dense depth map upsampling techniques have been applied [21,22]. The reconstructed dense depth maps can be used to convert single-view videos into stereoscopic or multi-view videos for a 3D display [24,25,26]. In addition, in the area of object recognition and tracking, sparse-to-dense depth map upsampling techniques have been applied to obtain a high-quality dense depth map from low-resolution and noisy depth information acquired from depth sensors [4,27].…”