354x Filetype PDF File size 0.85 MB Source: vision.in.tum.de
Reconstruction from
Multiple Views
Chapter 6 Prof. Daniel Cremers
Reconstruction from Multiple Views
FromTwoViewsto
Multiple View Geometry Multiple Views
Preimage & Coimage
Summer2022 from Multiple Views
FromPreimagesto
RankConstraints
Geometric
Interpretation
TheMultiple-view
Matrix
Relation to Epipolar
Constraints
Multiple-View
Reconstruction
Algorithms
Multiple-View
Reconstruction of
Prof. Daniel Cremers Lines
Chair of Computer Vision and Artificial Intelligence
Departments of Informatics & Mathematics
Technical University of Munich
updatedApril23,2022 1/43
Overview Reconstruction from
Multiple Views
Prof. Daniel Cremers
1 FromTwoViewstoMultipleViews
2 Preimage&CoimagefromMultipleViews FromTwoViewsto
Multiple Views
Preimage & Coimage
3 FromPreimagestoRankConstraints from Multiple Views
FromPreimagesto
RankConstraints
4 GeometricInterpretation Geometric
Interpretation
TheMultiple-view
5 TheMultiple-view Matrix Matrix
Relation to Epipolar
Constraints
6 Relation to Epipolar Constraints Multiple-View
Reconstruction
Algorithms
7 Multiple-View Reconstruction Algorithms Multiple-View
Reconstruction of
Lines
8 Multiple-View Reconstruction of Lines
updatedApril23,2022 2/43
Multiple-View Geometry Reconstruction from
Multiple Views
In this section, we deal with the problem of 3D reconstruction Prof. Daniel Cremers
given multiple views of a static scene, either obtained
simultaneously, or sequentially from a moving camera.
Thekeyideaisthat the three-view scenario allows to obtain FromTwoViewsto
moremeasurementstoinfer the same number of 3D Multiple Views
coordinates. For example, given two views of a single 3D point, Preimage & Coimage
from Multiple Views
wehavefourmeasurements(x-andy-coordinate in each FromPreimagesto
view), while the three-view case provides 6 measurements per RankConstraints
Geometric
point correspondence. As a consequence, the estimation of Interpretation
motion and structure will generally be more constrained when TheMultiple-view
reverting to additional views. Matrix
Relation to Epipolar
Thethree-view case has traditionally been addressed by the Constraints
Multiple-View
so-called trifocal tensor [Hartley ’95, Vieville ’93] which Reconstruction
Algorithms
generalizes the fundamental matrix. This tensor – as the Multiple-View
fundamental matrix – does not depend on the scene structure Reconstruction of
Lines
but rather on the inter-frame camera motion. It captures a
trilinear relationship between three views of the same 3D point
or line [Liu, Huang ’86, Spetsakis, Aloimonos ’87].
updatedApril23,2022 3/43
Trifocal Tensor versus Multiview Matrices Reconstruction from
Multiple Views
Prof. Daniel Cremers
Traditionally the trilinear relations were captured by
generalizing the concept of the Fundamental Matrix to that of a
Trifocal Tensor. It was developed among others by [Liu and
Huang’86], [Spetsakis, Aloimonos ’87]. The use of tensors
waspromotedby[Vieville ’93] and [Hartley ’95]. Bilinear, FromTwoViewsto
Multiple Views
trilinear and quadrilinear constraints were formulated in [Triggs Preimage & Coimage
’95]. This line of work is summarized in the books: from Multiple Views
FromPreimagesto
Faugeras and Luong, “The Geometry of Multiple Views”, 2001 RankConstraints
Geometric
and Interpretation
TheMultiple-view
Hartley and Zisserman, “Multiple View Geometry”, 2001, 2003. Matrix
Relation to Epipolar
In the following, however, we stick with a matrix notation for the Constraints
Multiple-View
multiview scenario. This approach makes use of matrices and Reconstruction
rank constraints on these matrices to impose the constraints Algorithms
Multiple-View
from multiple views. Such rank constraints were used by many Reconstruction of
authors, among others in [Triggs ’95] and in [Heyden, Åström Lines
’97]. This line of work is summarized in the book
Ma, Soatto, Kosecka, Sastry, “An Invitation to 3D Vision”, 2004.
updatedApril23,2022 4/43
no reviews yet
Please Login to review.