期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：1992
卷号：XXIX Part B3
页码：930-937
出版社：Copernicus Publications
摘要：A general paradigm to image understanding is proposed. As knowledgeabout the scene captured in a given image plays the central roleto understanding of this image, generic model-based approach aims atthe most compact body of visual knowledge. The dynamics of visioncan be structured in three operations of spatial structure of the scene:production (genesis) of scene instances from a generic model, inversion(parsing) of an actual scene instance back to a generic model,and learning (induction) of a generic model from a set of providedsample scenes. The plausibility of this general paradigm will be notonly partially proved by theoretical analysis, but also evidenced bybiological facts and psychological empirical discoveries, as well as supportedby research trends in computational vision. As an instance ofthis paradigm and an actual application of this developing theory, thestochastic attributed polygon map grammars as a generic model ofrurallanduse maps and remote sensing images are demonstrated.
