Mosaic construction from a sparse set of views

This paper describes a flexible approach for mosaic construction from a sparse set of uncalibrated views. The observation, that in architectural environments the majority of lines is aligned with the principal orthogonal directions of the world coordinate frame, is exploited in different stages of the mosaic construction pipeline. In the first stage the vanishing points are automatically detected and used for partial calibration of the camera and estimation of camera's relative orientation with respect to the scene. This single view analysis enables efficient feature matching and alignment of multiple views. In the final mosaic construction stage the internal camera parameters are refined simultaneously using all available views. We point out some practical issues related to the conditioning of this self-calibration technique. While the approach described here can be presented in the context of rotational mosaics, the alignment and matching techniques are applicable for general displacements, where the constraints of man-made environments are present and the displacement between the views is large.

[1]  Richard I. Hartley Self-Calibration from Multiple Views with a Rotating Camera , 1994, ECCV.

[2]  Richard I. Hartley,et al.  Linear self-calibration of a rotating and zooming camera , 1999, Proceedings. 1999 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Cat. No PR00149).

[3]  Richard Szeliski,et al.  Creating full view panoramic image mosaics and environment maps , 1997, SIGGRAPH.

[4]  Roberto Cipolla,et al.  Combining Single View Recognition and Multiple View Stereo for Architectural Scenes , 2001, ICCV.

[5]  Kenichi Kanatani,et al.  Geometric computation for machine vision , 1993 .

[6]  S. Shankar Sastry,et al.  Euclidean Reconstruction and Reprojection Up to Subgroups , 2004, International Journal of Computer Vision.

[7]  Roberto Cipolla,et al.  PhotoBuilder-3D models of architectural scenes from uncalibrated images , 1999, Proceedings IEEE International Conference on Multimedia Computing and Systems.

[8]  Bernhard P. Wrobel,et al.  Multiple View Geometry in Computer Vision , 2001 .

[9]  Camillo J. Taylor,et al.  Reconstruction of Linearly Parameterized Models from Single Images with a Camera of Unknown Focal Length , 2001, IEEE Trans. Pattern Anal. Mach. Intell..

[10]  Andrew Zisserman,et al.  Viewpoint invariant texture matching and wide baseline stereo , 2001, Proceedings Eighth IEEE International Conference on Computer Vision. ICCV 2001.

[11]  Robert T. Collins,et al.  Vanishing point calculation as a statistical inference on the unit sphere , 1990, [1990] Proceedings Third International Conference on Computer Vision.

[12]  Roberto Cipolla,et al.  Combining single view recognition and multiple view stereo for architectural scenes , 2001, Proceedings Eighth IEEE International Conference on Computer Vision. ICCV 2001.

[13]  Jitendra Malik,et al.  Modeling and Rendering Architecture from Photographs: A hybrid geometry- and image-based approach , 1996, SIGGRAPH.

[14]  Reinhard Koch,et al.  Multi Viewpoint Stereo from Uncalibrated Video Sequences , 1998, ECCV.

[15]  Harpreet S. Sawhney,et al.  Robust Video Mosaicing through Topology Inference and Local to Global Alignment , 1998, ECCV.

[16]  Richard Szeliski,et al.  Video mosaics for virtual environments , 1996, IEEE Computer Graphics and Applications.

[17]  Seth J. Teller,et al.  Automatic recovery of relative camera rotations for urban scenes , 2000, Proceedings IEEE Conference on Computer Vision and Pattern Recognition. CVPR 2000 (Cat. No.PR00662).

[18]  Wei Zhang,et al.  Video Compass , 2002, ECCV.

[19]  Andrew Zisserman,et al.  Combining scene and auto-calibration constraints , 1999, Proceedings of the Seventh IEEE International Conference on Computer Vision.

[20]  Leonard McMillan,et al.  Plenoptic Modeling: An Image-Based Rendering System , 2023 .