File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Conference Paper: Motion recovery from image sequences: Discrete viewpoint vs. differential viewpoint

TitleMotion recovery from image sequences: Discrete viewpoint vs. differential viewpoint
Authors
KeywordsEpipolar constraint
Motion estimation
Optical flow
Issue Date1998
Citation
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 1998, v. 1407, p. 337-353 How to Cite?
AbstractThe aim of this paper is to explore intrinsic geometric methods of recovering the three dimensional motion of a moving camera from a sequence of images. Generic similarities between the discrete approach and the differential approach are revealed through a parallel development of their analogous motion estimation theories. We begin with a brief review of the (discrete) essential matrix approach, showing how to recover the 3D displacement from image correspondences. The space of normalized essential matrices is characterized geometrically: the unit tangent bundle of the rotation group is a double covering of the space of normalized essential matrices. This characterization naturally explains the geometry of the possible number of 3D displacements which can be obtained from the essential matrix. Second, a differential version of the essential matrix constraint previously explored by is presented. We then present the precise characterization of the space of differential essential matrices, which gives rise to a novel eigenvector-decomposition-based 3D velocity estimation algorithm from the optical flow measurements. This algorithm gives a unique solution to the motion estimation problem and serves as a differential counterpart of the SVD-based 3D displacement estimation algorithm from the discrete case. Finally, simulation results are presented evaluating the performance of our algorithm in terms of bias and sensitivity of the estimates with respect to the noise in optical flow measurements.
Persistent Identifierhttp://hdl.handle.net/10722/327084
ISSN
2023 SCImago Journal Rankings: 0.606

 

DC FieldValueLanguage
dc.contributor.authorMa, Yi-
dc.contributor.authorKošecká, Jana-
dc.contributor.authorSastry, Shankar-
dc.date.accessioned2023-03-31T05:28:41Z-
dc.date.available2023-03-31T05:28:41Z-
dc.date.issued1998-
dc.identifier.citationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 1998, v. 1407, p. 337-353-
dc.identifier.issn0302-9743-
dc.identifier.urihttp://hdl.handle.net/10722/327084-
dc.description.abstractThe aim of this paper is to explore intrinsic geometric methods of recovering the three dimensional motion of a moving camera from a sequence of images. Generic similarities between the discrete approach and the differential approach are revealed through a parallel development of their analogous motion estimation theories. We begin with a brief review of the (discrete) essential matrix approach, showing how to recover the 3D displacement from image correspondences. The space of normalized essential matrices is characterized geometrically: the unit tangent bundle of the rotation group is a double covering of the space of normalized essential matrices. This characterization naturally explains the geometry of the possible number of 3D displacements which can be obtained from the essential matrix. Second, a differential version of the essential matrix constraint previously explored by is presented. We then present the precise characterization of the space of differential essential matrices, which gives rise to a novel eigenvector-decomposition-based 3D velocity estimation algorithm from the optical flow measurements. This algorithm gives a unique solution to the motion estimation problem and serves as a differential counterpart of the SVD-based 3D displacement estimation algorithm from the discrete case. Finally, simulation results are presented evaluating the performance of our algorithm in terms of bias and sensitivity of the estimates with respect to the noise in optical flow measurements.-
dc.languageeng-
dc.relation.ispartofLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)-
dc.subjectEpipolar constraint-
dc.subjectMotion estimation-
dc.subjectOptical flow-
dc.titleMotion recovery from image sequences: Discrete viewpoint vs. differential viewpoint-
dc.typeConference_Paper-
dc.description.naturelink_to_subscribed_fulltext-
dc.identifier.doi10.1007/BFb0054751-
dc.identifier.scopuseid_2-s2.0-84957797530-
dc.identifier.volume1407-
dc.identifier.spage337-
dc.identifier.epage353-
dc.identifier.eissn1611-3349-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats