Title :
Map representations and coding-based priors for segmentation
Author_Institution :
AT&T Bell Labs., Holmdel, NJ, USA
Abstract :
The Bayesian segmentation model developed is motivated by consideration of the information needed for higher-level visual processing. A segmentation is regarded as a collection of parameters defining an image-valued stochastic process by separating topological (adjacency) and metric (shape) properties of the subdivision and intensity properties of each region. The prior selection is structured accordingly. The novel part of the representation, the subdivision topology, is assigned a prior by universal coding arguments, using the minimum description-length philosophy that the best segmentation allows the most efficient representation of visual data
Keywords :
Bayes methods; encoding; pattern recognition; stochastic processes; Bayesian segmentation model; adjacency; coding-based priors; higher-level visual processing; image-valued stochastic process; intensity properties; map representation; minimum description-length philosophy; shape; subdivision; universal coding; visual data representation; Bayesian methods; Energy capture; Image coding; Image segmentation; Layout; Military computing; Probability distribution; Shape; Stochastic processes; Topology;
Conference_Titel :
Computer Vision and Pattern Recognition, 1991. Proceedings CVPR '91., IEEE Computer Society Conference on
Conference_Location :
Maui, HI
Print_ISBN :
0-8186-2148-6
DOI :
10.1109/CVPR.1991.139727