6533b86efe1ef96bd12cb2be

RESEARCH PRODUCT

Evaluation of a Computer Vision Based Automatic Incident and Congestion Detection System in an Urban Context

subject

description

Abstract The objective of the document is to describe an Automatic Incident and Congestions Detection System based on computer vision technology, in an urban environment. The document outlines the main technical features of the system and presents the first evaluation results obtained in a field trial currently carried out in protect INVAII) II. within the Transport Telematics Programme of the EC. CCTV cameras are installed in urban street intersections lor surveillance purposes. Based on image processing techniques. the Video Analyzers or Local Sensor Modules (LSM) connected to a number of selected cameras, obtain the traffic status measurements within the field of view of each one of them. The data are further processed at central level, where the system calculates the traffic status of the full network, based on spatial and temporal relations. The system displays this information to the operator and triggers alarms if any abnormal situation is detected. The system developed by an international consortium. was undertaken to a full scale test in a field trial carried out in Valencia. Spain. The installed pilot system consists in 8 image analyzers installed at four intersections, while data processing equipment is at the Traffic Control Center.