期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2009
卷号：XXXVIII-5/W1
出版社：Copernicus Publications
摘要：In this paper we present a system for three-dimensional reconstruction of architectural scenes from uncalibrated videos. These videos might be recorded using hand-held cameras, downloaded from the internet or taken from archival sources. Because we do not require prior knowledge of the camera's internal parameters such as focal length, center of projection, and radial distortion we can deal with videos from uncontrolled sources. We present fast algorithms for 2D feature tracking on the GPU, real-time robust estimation and bundle adjustment. We also demonstrate a new type of feature, the viewpoint invariant patch (VIP), in its application to loop detection and closing
关键词：structure from motion; 3d reconstruction; self-calibration; bundle-adjustment; gpgpu
