Motion Recognition Using Nonparametric Image Motion Models Estimated from Temporal and Multiscale Cooccurrence Statistics
Issue No. 12 - December (2003 vol. 25)
<p><b>Abstract</b>—A new approach for motion characterization in image sequences is presented. It relies on the probabilistic modeling of temporal and scale cooccurrence distributions of local motion-related measurements directly computed over image sequences. Temporal multiscale Gibbs models allow us to handle both spatial and temporal aspects of image motion content within a unified statistical framework. Since this modeling mainly involves the scalar product between cooccurrence values and Gibbs potentials, we can formulate and address several fundamental issues: model estimation according to the ML criterion (hence, model training and learning) and motion classification. We have conducted motion recognition experiments over a large set of real image sequences comprising various motion types such as temporal texture samples, human motion examples, and rigid motion situations. </p>
Nonparametric motion analysis, motion recognition, multiscale analysis, Gibbs models, cooccurrences, ML criterion.
P. Bouthemy and R. Fablet, "Motion Recognition Using Nonparametric Image Motion Models Estimated from Temporal and Multiscale Cooccurrence Statistics," in IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 25, no. , pp. 1619-1624, 2003.