Show simple item record

dc.contributor.advisorStrickland, Robin N.en_US
dc.contributor.authorMao, Zuhua.*
dc.creatorMao, Zuhua.en_US
dc.date.accessioned2011-10-31T17:47:50Z
dc.date.available2011-10-31T17:47:50Z
dc.date.issued1992en_US
dc.identifier.urihttp://hdl.handle.net/10150/185769
dc.description.abstractAn application-motivated approach is proposed for rigid and nonrigid motion analysis. Optical flow from a sequence of 2-D images is estimated by computing the velocity field along a moving contour in the scene. This new approach is different from others in that it combines displacement computed by feature matching with a smoothness constraint on the second derivative of velocity. First, a new relaxation matching technique is used to find correspondences between contour features in adjacent image frames. Displacements for discrete points along the contour are interpolated from the magnitudes and directions of neighboring matching points. The displacements so computed are used as initial estimates for the velocity (magnitudes and direction) along the contour. The final estimated velocities are required to yield components which are close in a least-squares sense to these initial velocity magnitudes, when projected along the same directions. The second derivative of velocity is constrained to be minimum when integrated along the contour, leading to a unique solution for the motion of a straight line undergoing an affine transformation. The second derivative constraint gives better results for most second order flows. In cases where it does not, a combination of first and second derivative constraints can be used. A two component model (mean and residual) is investigated for vortex flow motion analysis. The Hough transform and phase portrait method can be used for vortex texture recognition in an orientation field. They are totally image independent; and not sensitive to noise. The phase portrait is also domain independent, and has great potential as a symbolic descriptor for flow analysis. The mean component for each vortex can be estimated by a vortex matching procedure. The final residuals are computed along the vortex streamlines. They are required to yield components which are close, in a least squares sense, to the initial normal velocity along the contours, when projected to the same direction.
dc.language.isoenen_US
dc.publisherThe University of Arizona.en_US
dc.rightsCopyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.en_US
dc.subjectDissertations, Academic.en_US
dc.subjectElectrical engineering.en_US
dc.titleComputing optical flow in rigid and nonrigid object motion.en_US
dc.typetexten_US
dc.typeDissertation-Reproduction (electronic)en_US
dc.identifier.oclc712179581en_US
thesis.degree.grantorUniversity of Arizonaen_US
thesis.degree.leveldoctoralen_US
dc.contributor.committeememberSchowengerdt, Robert A,en_US
dc.contributor.committeememberRodriguez, Jeffrey J.en_US
dc.identifier.proquest9220696en_US
thesis.degree.disciplineElectrical and Computer Engineeringen_US
thesis.degree.disciplineGraduate Collegeen_US
thesis.degree.namePh.D.en_US
refterms.dateFOA2018-06-04T13:31:21Z
html.description.abstractAn application-motivated approach is proposed for rigid and nonrigid motion analysis. Optical flow from a sequence of 2-D images is estimated by computing the velocity field along a moving contour in the scene. This new approach is different from others in that it combines displacement computed by feature matching with a smoothness constraint on the second derivative of velocity. First, a new relaxation matching technique is used to find correspondences between contour features in adjacent image frames. Displacements for discrete points along the contour are interpolated from the magnitudes and directions of neighboring matching points. The displacements so computed are used as initial estimates for the velocity (magnitudes and direction) along the contour. The final estimated velocities are required to yield components which are close in a least-squares sense to these initial velocity magnitudes, when projected along the same directions. The second derivative of velocity is constrained to be minimum when integrated along the contour, leading to a unique solution for the motion of a straight line undergoing an affine transformation. The second derivative constraint gives better results for most second order flows. In cases where it does not, a combination of first and second derivative constraints can be used. A two component model (mean and residual) is investigated for vortex flow motion analysis. The Hough transform and phase portrait method can be used for vortex texture recognition in an orientation field. They are totally image independent; and not sensitive to noise. The phase portrait is also domain independent, and has great potential as a symbolic descriptor for flow analysis. The mean component for each vortex can be estimated by a vortex matching procedure. The final residuals are computed along the vortex streamlines. They are required to yield components which are close, in a least squares sense, to the initial normal velocity along the contours, when projected to the same direction.


Files in this item

Thumbnail
Name:
azu_td_9220696_sip1_m.pdf
Size:
4.031Mb
Format:
PDF
Description:
azu_td_9220696_sip1_m.pdf

This item appears in the following Collection(s)

Show simple item record