Image/Video Understanding Systems based on network-symbolic models and active vision

摘要

Vision is a part of information system that converts visual information into knowledge structures. These structures drive the vision process, resolving ambiguity and uncertainty via feedback, and provide image understanding, which is an interpretation of visual information in terms of these knowledge models. It is hard to split the entire system apart, and vision mechanisms cannot be completely understood separately from informational processes related to knowledge and intelligence. Brain reduces informational and computational complexities, using implicit symbolic coding of features, hierarchical compression, and selective processing of visual information. Vision is a component of situation awareness, motion and planning systems. Foveal vision provides semantic analysis, recognizing objects in the scene. Peripheral vision guides fovea to salient objects and provides scene context Biologically inspired Network-Symbolic representation, in which both systematic structural/logical methods and neural/statistical methods are parts of a single mechanism, converts visual information into relational Network-Symbolic structures, avoiding precise artificial computations of 3-D models. Network-Symbolic transformations derive more abstract structures that allows for invariant recognition of an object as exemplar of a class and for a reliable identification even if the object is occluded. Systems with such smart vision will be able to navigate in real environment and understand real-world situations.