Mathematical Methods for Signal and Image Analysis and Representation

0374142 - ÚTIA 2012 RIV GB eng M - Monography Chapter
Haindl, Michal
Visual Data Recognition and Modeling Based on Local Markovian Models.
Mathematical Methods for Signal and Image Analysis and Representation. Vol. 14. London: Springer London, 2012 - (Florack, L.; Duits, R.; Jongbloed, G.; Lieshout, M.; Davies, L.), s. 241-259. Computational Imaging and Vision. ISBN 978-1-4471-2353-8
R&D Projects: GA MŠMT 1M0572; GA ČR GAP103/11/0335; GA ČR GA102/08/0593
Grant - others:CESNET(CZ) 387/2010
Institutional research plan: CEZ:AV0Z10750506
Keywords : Markov random fields * image modeling * image recognition
Subject RIV: BD - Theory of Information
http://library.utia.cas.cz/separaty/2012/RO/haindl-0374142.pdf

An exceptional 3D wide-sense Markov model which can be completely solved analytically and easily synthesised is presented. The model can be modified to faithfully represent complex local data by adaptive numerically robust recursive estimators of all its statistics. Illumination invariants can be derived from some of its recursive statistics and exploited in content based image retrieval, supervised or unsupervised image recognition. Its modelling efficiency is demonstrated on several analytical and modelling image applications, in particular on unsupervised image or range data segmentation, bidirectional texture function (BTF) synthesis and compression, dynamic texture synthesis and adaptive multispectral and multichannel image and video restoration.
Permanent Link: http://hdl.handle.net/11104/0207127