Context and Hierarchy in a Probabilistic Image Model Export

@proceedings{citeulike:2404441, abstract = {It is widely conjectured that the excellent ROC performance of biological vision systems is due in large part to the exploitation of context at each of many levels in a part/whole hierarchy. We propose a mathematical framework (a "composition machine") for constructing probabilistic hierarchical image models, designed to accommodate arbitrary contextual relationships, and we build a demonstration system for reading Massachusetts license plates in an image set collected at Logan Airport. The demonstration system detects and correctly reads more than 98\% of the plates, with a negligible rate of false detection. Unlike a formal grammar, the architecture of a composition machine does not exclude the sharing of sub-parts among multiple entities, and does not limit interpretations to single trees (e.g. a scene can have multiple license plates, or no plates at all). In this sense, the architecture is more like a general Bayesian network than a formal grammar. On the other hand, unlike a Bayesian network, the distribution is non-Markovian, and therefore more like a probabilistic context-sensitive grammar. The conceptualization and construction of a composition machine is facilitated by its formulation as the result of a series of non-Markovian perturbations of a "Markov backbone."}, author = {Jin, Ya and Geman, S.}, booktitle = {Computer Vision and Pattern Recognition, 2006 IEEE Computer Society Conference on}, citeulike-article-id = {2404441}, citeulike-linkout-0 = {http://dx.doi.org/10.1109/CVPR.2006.86}, citeulike-linkout-1 = {http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1641016}, doi = {10.1109/CVPR.2006.86}, journal = {Computer Vision and Pattern Recognition, 2006 IEEE Computer Society Conference on}, keywords = {computer\_vision, hierarchical\_model, theory}, pages = {2145--2152}, posted-at = {2008-02-20 19:36:39}, priority = {0}, title = {Context and Hierarchy in a Probabilistic Image Model}, url = {http://dx.doi.org/10.1109/CVPR.2006.86}, volume = {2}, year = {2006} }

TY - CONF ID - citeulike:2404441 L3 - citeulike-article-id:2404441 N2 - It is widely conjectured that the excellent ROC performance of biological vision systems is due in large part to the exploitation of context at each of many levels in a part/whole hierarchy. We propose a mathematical framework (a "composition machine") for constructing probabilistic hierarchical image models, designed to accommodate arbitrary contextual relationships, and we build a demonstration system for reading Massachusetts license plates in an image set collected at Logan Airport. The demonstration system detects and correctly reads more than 98% of the plates, with a negligible rate of false detection. Unlike a formal grammar, the architecture of a composition machine does not exclude the sharing of sub-parts among multiple entities, and does not limit interpretations to single trees (e.g. a scene can have multiple license plates, or no plates at all). In this sense, the architecture is more like a general Bayesian network than a formal grammar. On the other hand, unlike a Bayesian network, the distribution is non-Markovian, and therefore more like a probabilistic context-sensitive grammar. The conceptualization and construction of a composition machine is facilitated by its formulation as the result of a series of non-Markovian perturbations of a "Markov backbone." JF - Computer Vision and Pattern Recognition, 2006 IEEE Computer Society Conference on EP - 2152 TI - Context and Hierarchy in a Probabilistic Image Model VL - 2 T2 - Computer Vision and Pattern Recognition, 2006 IEEE Computer Society Conference on SP - 2145 KW - computer_vision KW - hierarchical_model KW - theory AU - Jin, Ya AU - Geman, S PY - 2006/// UR - http://dx.doi.org/10.1109/CVPR.2006.86 ER -