Technique for view synthesis

摘要

A technique for computing an image of a virtual view based on a plurality of camera views is presented. One or more cameras provide the plurality of camera views. As to a method aspect of the technique, two or three camera views that at least partially overlap and that at least partially overlap with the virtual view are selected among the plurality of camera views. The image of the virtual view is computed based on objects in the selected camera views using a multilinear relation that relates the selected camera views and the virtual view.

主权项

1. A method, performed by a device, for computing an image of a virtual view, the method comprising:
receiving a plurality of camera views from one or more cameras; selecting from among the plurality of camera views two or three camera views that at least partially overlap and that at least partially overlap with the virtual view; and computing the image of the virtual view based on objects in the selected camera views using a multilinear relation that relates the selected camera views and the virtual view, wherein the multilinear relation is represented by a trifocal tensor (T), if two camera views are selected, and wherein the multilinear relation is represented by a quadrifocal tensor (Q), if three camera views are selected, wherein the virtual view is specified by parameters of a projection, the parameters of the projection specifying the virtual view include a zoom level of the virtual view, wherein a mesh is represented in memory, the mesh including a vertex for each of the camera views and an edge for each pair of overlapping camera views among the plurality of camera views, and the selection of the camera views from among the plurality of camera views depends on the zoom level of the virtual view and one or more other parameters of the projection specifying the virtual view.