资 源 简 介
在任意未知的光照下对移动目标的密集形状3D重建。
This paper addresses the problem of shape reconstruction from multiple views of a moving object. We work
directly with image intensities and produce a dense reconstruction. The illumination of the scene can be arbitrary
but the same for all views, and our method does not rely
on any information about light distribution. We do require
prior knowledge of object motion (which is our primary
source of information), along with either epipolar geometry
(for matching) or camera calibration (for shape recovery).
The object is assumed to be (mainly) matte and piecewise
smooth, and the camera(s) to be orthographic.