Incremental recognition of traffic situations from video image sequences

Our image evaluation system XTRACK tracks multiple-vehicle-configurations i n image sequences. The resulting geometric state descriptions are associate d with fuzzy attributes and relations and thereby form the basis for increm ental characterization of traffic situations from the point of view of sele cted road users or observers. Knowledge representation and inference is per formed by means of fuzzy metric temporal logic in order to provide an in-de pth analyzable transition from raw video data to conceptual descriptions of traffic situations. (C) 2000 Elsevier Science B.V. All rights reserved.