EUROGRAPHICS 2006 / D. W. Fellner and C. Hansen
Short Papers
Rapid Interactive Modelling from Video with Graph Cuts
Anton van den Hengel1, Anthony Dick1, Thorsten Thormählen1, Ben Ward1, and Philip H. S. Torr2
1School of Computer Science, University of Adelaide, Australia
2Department of Computing, Oxford Brookes University, UK
Abstract
We present a method for generating a parameterised model of a scene from a set of images. The method is novel
in that it uses information from several sources—video, sparse 3D points and user input—to fit models to a scene.
The user drives the process by providing selected high-level scene information, for instance selecting an object
in the scene, or specifying the relationship between a pair of objects. The system combines this information with
image and 3D data to dynamically update its model of the scene. In doing so it avoids common pitfalls of both
automatic structure and motion algorithms, and image-based modelling packages.
Categories and Subject Descriptors (according to ACM CCS): I.4.8 [Image Processing and Computer Vision]: Scene
Analysis
1. Introduction
Building 3D models of scenes from image data has been the
subject of significant research effort and commercial inter-
est. Structure-and-motion techniques for recovering point-
based scene reconstructions and camera parameters are be-
coming well understood, to the point where commercial ap-
plications offer the technology (such as boujou from 2d3,
amongst others). These techniques automatically generate a
3D point cloud, and an understanding of the relationship be-
tween the camera and the scene. The camera information is
particularly useful for a number of video manipulation pro-
cesses, including the insertion of computer generated ele-
ments into real video. The 3D point cloud, however, is a
sparse reconstruction of the scene structure which can be dif-
ficult to interpret and use for modelling purposes.
There are also a number of systems which facilitate the
modelling of scenes from image data; these include Fa-
ca