|Title:||Novel View Synthesis in Tensor Space
|Authors:||S. Avidan and A. Shashua
|Abstract:||This paper presents a new method for synthesizing novel views of a 3D scene from three reference images and their trilinear tensor. No use of a 3D model of the scene is made, nor any use of camera geometry (relative locations of the cameras) of the three views is required - only the computation of the trilinear tensor. The novel views are modeled by specifying rotation and translation of the virtual camera starting from one of the reference views - which is an ideal ``driving'' mode for generating a synthetic movie. The synthesis process is governed by a simple equation (relating the model tensor and the new tensor) in which the user-specified virtual camera is plugged in, followed by a warping step - hence, the only time consuming component in this process is image warping. The virtual camera can be anywhere in 3D and the quality of the synthesized images are limited only by the quality of the (dense) correspondence between two of the reference images and the amount of self-occlusions in the scene.|
|Copyright||The above paper is copyright by the Technion, Author(s), or others. Please contact the author(s) for more information|
Remark: Any link to this technical report should be to this page (http://www.cs.technion.ac.il/users/wwwb/cgi-bin/tr-info.cgi/1996/CIS/CIS9602), rather than to the URL of the PDF files directly. The latter URLs may change without notice.
To the list of the CIS technical reports of 1996
To the main CS technical reports page
Computer science department, Technion