6 - Estimation et structuration d'indices spatio-temporels pour l'analyse du mouvement dans une séquence d'images

Show simple item record

dc.contributor.author BOUTHEMY (P.) en_US
dc.date.accessioned 2005-07-20T12:50:43Z
dc.date.available 2005-07-20T12:50:43Z
dc.date.issued 1987 en_US
dc.identifier.citation Traitement du Signal [Trait. Signal], 1987, Vol. 4, N° 3-NS, p. 239-257 en_US
dc.identifier.issn 0765-0019 en_US
dc.identifier.uri http://hdl.handle.net/2042/1632
dc.description This paper describes a motion estimation scheme front image sequences, which consists of three main stages : local processing, intermediate-level structuring, optic flow field estimation . The first stage is concerned with either the determination of spatiotemporal edges along with their local spatial direction in the image plane and the component of their associated velocity vector perpendicular to this direction, or the segmentation into regions according to motion-based hierarchically performed criteria which take into account an explicit partial motion information . Both designed algorithms utilize principles of local modeling and likely hypothesis testing. Concerning the first one, the defined likelihood ratio test is implemented according to some appropriate mask convolution, the complexity order of which is similar to conventional spatial gradient computation . The purpose of the second stage is to obtain a structured partition of the image, resulting front edge linking and/or region segmentation . Thfienall stage deals with the velocity field estimation, that-is-to-say the reconstruction of the second component of displacement vectors by combining local observations. First, a recursive stochastic gradient, used to achieve the minimization of some simple functional, enables to estimate optic flow along contours . Then, the estimation within delineated domains is considered . Our approach is in particular distinguished by treating beforehand potential discontinuities of the velocity field in the image . Moreover, it provides with a set of intermediate-level spatio-temporal primitives .
dc.description.abstract Schéma d'estimation du mouvement dans des séquences d'images s'articulant en 3 étapes fondamentales: traitement local, structuration intermédiaire, estimation du champ des vitesses en_US
dc.format.extent 51964 bytes
dc.format.mimetype application/pdf
dc.language.iso en_US
dc.publisher GRETSI, Saint Martin d'Hères, France en_US
dc.relation.ispartofseries Traitement du Signal
dc.rights http://irevues.inist.fr/utilisation en_US
dc.source Traitement du Signal [Trait. Signal], ISSN 0765-0019, 1987, Vol. 4, N° 3-NS, p. 239-257 en_US
dc.subject.cnrs Traitement image en_US
dc.subject.cnrs Mouvement en_US
dc.subject.cnrs Estimation en_US
dc.subject.cnrs Analyse image en_US
dc.subject.cnrs Séquence image en_US
dc.subject.cnrs Vision ordinateur en_US
dc.subject.cnrs Segmentation en_US
dc.subject.cnrs Traitement numérique en_US
dc.subject.cnrs Scène primitive en_US
dc.title 6 - Estimation et structuration d'indices spatio-temporels pour l'analyse du mouvement dans une séquence d'images en_US
dc.title.alternative Motion segmentation and estimation through the determination of spatio-temporal primitives in an image sequence en_US
dc.type Article en_US
dc.contributor.affiliation IRISA/cent. INRIA, Rennes 35042 en_US

Files in this item

PDF 006.PDF TEXTE.pdf 2.218Mb

This item appears in the following Collection(s)

Show simple item record

Advanced Search