Paper Info
Title
Robust camera self-calibration from monocular images of Manhattan worlds
Abstract
We focus on the detection of orthogonal vanishing points using line segments extracted from a single view, and using these for camera self-calibration. Recent methods view this problem as a two-stage process. Vanishing points are extracted through line segment clustering and subsequently likely orthogonal candidates are selected for calibration. Unfortunately, such an approach is easily distracted by the presence of clutter. Furthermore, geometric constraints imposed by the camera and scene orthogonality are not enforced during detection, leading to inaccurate results which are often inadmissible for calibration. To overcome these limitations, we present a RANSAC-based approach using a minimal solution for estimating three orthogonal vanishing points and focal length from a set of four lines, aligned with either two or three orthogonal directions. In addition, we propose to refine the estimates using an efficient and robust Maximum Likelihood Estimator. Extensive experiments on standard datasets show that our contributions result in significant improvements over the state-of-the-art.
Year
DOI
Venue
2012
10.1109/CVPR.2012.6248008
CVPR
Keywords
Field
DocType
feature extraction,geometry,maximum likelihood estimation,pattern clustering,Manhattan worlds,RANSAC-based approach,geometric constraints,line segment clustering,line segment extraction,monocular images,orthogonal vanishing points,robust camera self-calibration,robust maximum likelihood estimator,two-stage process
Line segment,Computer vision,Pattern recognition,RANSAC,Clutter,Computer science,Orthogonality,Image segmentation,Robustness (computer science),Artificial intelligence,Cluster analysis,Vanishing point
Conference
Volume
Issue
ISSN
2012
1
1063-6919
Citations 
PageRank 
References 
13
0.59
10
Authors
2
Name
Order
Citations
PageRank
Horst Wildenauer112612.81
Allan Hanbury21756144.05