Intelligent retrieval and storage of outdoor events derived from video
Jim Cotter · 1997
The CELS project aims to provide an intelligent databasing solution to a restricted but highly useful domain: event logging in carpark environments. Events are temporally extended phenomena such as the arrival, parking and departure of vehicles, and passage of people. A loosely coupled system of asynchronous information sources are employed to annotate the raw visual data to enable sophisticated although retrospective queries. The primary data source described at some length in this correspondence is FREDA, the fast and robust event detection algorithm, which detects motion and tracks events as they traverse the scene. Introduction Automatically deriving information from the tremendous number of video cameras currently being installed in public spaces has become a central focus for vision research[1]. The enormous range of commercial and public safety applications includes intruder alarms, collation of shopping behaviours, automatic event logging for surveillance and carpark administr...