Scalable Inside-Out Image-Based Rendering

作     者：Hedman, Peter Ritschel, Tobias Drettakis, George Brostow, Gabriel 

作者机构：UCL London WC1E 6BT England Inria Rocquencourt France 

出 版 物：《ACM TRANSACTIONS ON GRAPHICS》 (ACM Trans Graphics)

年 卷 期：2016年第35卷第6期

页      面：1-11页

核心收录：

学科分类：08[工学] 0835[工学-软件工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：EPSRC [EP/K023578/1  EP/K02339X/1] Funding Source: UKRI 

主　　题：Image-based rendering RGB-D 3D reconstruction tiled rendering 

摘      要：Our aim is to give users real-time free-viewpoint rendering of real indoor scenes, captured with off-the-shelf equipment such as a high-quality color camera and a commodity depth sensor. Image-based Rendering (IBR) can provide the realistic imagery required at real-time speed. For indoor scenes however, two challenges are especially prominent. First, the reconstructed 3D geometry must be compact, but faithful enough to respect occlusion relationships when viewed up close. Second, man-made materials call for view-dependent texturing, but using too many input photographs reduces performance. We customize a typical RGB-D 3D surface reconstruction pipeline to produce a coarse global 3D surface, and local, per-view geometry for each input image. Our tiled IBR preserves quality by economizing on the expected contributions that entire groups of input pixels make to a final image. The two components are designed to work together, giving real-time performance, while hardly sacrificing quality. Testing on a variety of challenging scenes shows that our inside-out IBR scales favorably with the number of input images.