Popham, T.J., Bhalerao, A.H. and Wilson, R.G. (2010) Multi-Frame Scene-Flow Estimation Using a Patch Model and Smooth Motion Prior. In: British Machine Vision Conference Workshop, London, UK.

Preview

PDF - Published Version Download (1756Kb) | Preview

Abstract

This paper addresses the problem of estimating the dense 3D motion of a scene over several frames using a set of calibrated cameras. Most current 3D motion estimation techniques are limited to estimating the motion over a single frame, unless a strong prior model of the scene (such as a skeleton) is introduced. Estimating the 3D motion of a general scene is difficult due to untextured surfaces, complex movements and occlusions. In this paper, we show that it is possible to track the surfaces of a scene over several frames, by introducing an effective prior on the scene motion. Experimental results show that the proposed method estimates the dense scene-flow over multiple frames, without the need for multiple-view reconstructions at every frame. Furthermore, the accuracy of the proposed method is demonstrated by comparing the estimated motion against a ground truth.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	pcav hpsg multimedia vision image signal processing video
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science Q Science > QA Mathematics > QA76.73 Computer algorithms. Data structures.
Divisions:	Faculty of Science > Computer Science
Depositing User:	Simon Hammond
Date Deposited:	06 Nov 2010 15:07
Last Modified:	23 Feb 2012 09:08
URI:	http://eprints.dcs.warwick.ac.uk/id/eprint/290

Actions (login required)

View Item