An image interpretation system is described that incorporates scene knowledge in order to provide a reliable interpretation of the objects that appear in the image. Scene knowledge consists of models of objects expected in the scene and their spatial and temporal relationships. Extracted image features represented symbolically, and a network of scene object models described using frames, drive the interpretation in both a bottom-up and a top-down fashion. The interpretation process is object-oriented. The system has been exercised on a number of forward-looking infrared images. The symbolic representation of the image features and the network representation of the object models are briefly discussed. The processes of model instantiation and evidence gathering and evaluation are then described. The spatial and temporal knowledge which captures the relationship between roads, vehicles, collections of vehicles, and object persistence from frame to frame are discussed. Lastly, the system is demonstrated with an example.
Infrared image interpretation using spatial and temporal knowledge
Interpretation eines Infrarotbildes durch raeumliche und zeitlicher Kenntnisse
1987
4 Seiten, 16 Quellen
Conference paper
English
Knowledge-based approach to image interpretation
British Library Online Contents | 1993
|Qualitative spatial relationships for image interpretation by using a conceptual graph
British Library Online Contents | 2009
|Taylor & Francis Verlag | 2019
|Fuzzy spatial relationships for image processing and interpretation: a review
British Library Online Contents | 2005
|