Premium
P‐8: Depth‐Map‐Based Multi‐View Synthesis Using Joint Bilateral Upsampling on GPUs
Author(s) -
Wu MengSung,
Chuang YungYu,
Liu YenTzu,
Hu ChengChung
Publication year - 2010
Publication title -
sid symposium digest of technical papers
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.351
H-Index - 44
eISSN - 2168-0159
pISSN - 0097-966X
DOI - 10.1889/1.3499895
Subject(s) - upsampling , depth map , computer science , stereoscopy , rendering (computer graphics) , computer vision , artificial intelligence , computer graphics (images) , autostereoscopy , joint (building) , image based modeling and rendering , image (mathematics) , architectural engineering , engineering
Stereoscopic displays have become more and more common. A popular method for synthesizing 3D content is depth‐image‐based rendering, in which multi‐view images are synthesized from a color frame and an accompanying depth map. However, when the image resolution increases, it is expensive to use a full‐resolution depth map because its high storage cost and decoding overhead. In addition, since the depth map is often smooth, it is often not necessary to store the full‐resolution depth map. This paper uses a low‐resolution depth map and increases its resolution at run time using joint bilateral upsampling on GPUs. By doing so, we are able to display high‐resolution stereoscopic videos on an autostereoscopic display.