Technologies of Vision

TAS: Traffic Analysis System

Progetto Pilota:
Improving Traffic Safety in the Urban Environment

Descrizione completa del Progetto Pilota sul sito del Comune di Trento (in italiano)

This project supports the ICTCT's motto: "Don't wait for accidents to happen". The goal is the identification and automatic extraction of visual characteristics of road traffic that are related to dangerous situations.

At the beginning of the year 2003, ITC-irst in collaboration with Comune di Trento starts a pilot project that aims at mitigate traffic accidents in Trento city. The project is partially sponsored by the Italian Ministero delle Infrastrutture e dei Trasporti (within Programma Nazionale della Sicurezza Stradale).

The emphasis of the project is the developement of new methods for the automatic collection of traffic parameters and features describing the road users behaviour in order to build an accident risk map. Objective of the study is the quantification of the accident risk mitigation by changing some characteristics in the road structure.

Project working steps for TeV:

• Sample roads detection: two steets in Trento have been selected as test bench.
• Establishing interesting visual events to be automatically extracted from the sequences.
• Choice of points to be monitored: for each street, 2 nodes in the road network have been judged as very relevant for the project. For each node, 2 or 3 views are particularly interesting (long and short views).
• Deliverable document on requirement and system architecture.
• Hardware specification: selection of cameras, lenses, frame grabbers and computers right for the application.
• Installation of poles, video cameras and computers in the 4 nodes.
• Acquisition and storing of video sequences (60 hours) from each camera.
• Image processing - Development of sophisticated algorithms of video analysis able to extract automatically the desired features from the stored sequences.
• Risk map - Identification of the features that are strongly related to the accident risk and population of the risk map.
• Variations of the infrastructure or the manner in which it is used (by Comune di Trento).
• Second acquisition of video sequences.
• Risk map re-computation with the new road situation and comparison.
• Deliverable document on methodology and practice for the accident risk evaluation starting from traffic microparameters.

The proposal of the project was presented at the 57th National Conference on Traffic, Riva del Garda, Trento, October 15-20, 2001.

Project end: May 2004.

Results.

• The Project results have been presented to Consulta Nazionale sulla Sicurezza Stradale in the "Sessione Plenaria su Individuazione e Diffusione delle Migliori Pratiche per la Sicurezza Stradale: Il Contributo dei Progetti Pilota", Roma, 21 luglio 2004.
• N. Segata, "Sistemi di monitoraggio video: il problema dello sfondo in presenza di variazioni globali di illuminazione", Tesi di Laurea in Informatica, Universita' degli Studi di Trento, 2004, ITC-irst P04-12-27. In the Appendix: Statistics Suite (in Italian).
• A conclusive publication is the following: S. Messelodi, C.M. Modena, "A Computer Vision System for Traffic Accident Risk Measurement: A Case Study", Advances in Transportation Studies, an International Journal, 7(B), pp. 51-66, November 2005.

Related projects.

• SCOCA: research and technology