En-tête de navigationNavigation principaleSuiviFiche


Unité de recherche
PCRD EU
Numéro de projet
99.0677-1
Titre du projet
CROSSES: Crowd simulation system for emergency situations
Titre du projet anglais
CROSSES: Crowd simulation system for emergency situations

Textes relatifs à ce projet

 AllemandFrançaisItalienAnglais
Mots-clé
-
-
-
Anzeigen
Autre Numéro de projet
-
-
-
Anzeigen
Programme de recherche
-
-
-
Anzeigen
Description succincte
-
-
-
Anzeigen
Partenaires et organisations internationales
-
-
-
Anzeigen
Résumé des résultats (Abstract)
-
-
-
Anzeigen
Références bases de données
-
-
-
Anzeigen

Textes saisis


CatégorieTexte
Mots-clé
(Anglais)
Emergency situation; crowd modeling and simulation; VR system for training.
Autre Numéro de projet
(Anglais)
EU project number: IST-1999-10510
Programme de recherche
(Anglais)
EU-programme: 5. Frame Research Programme - 1.2.4 Essential technologies and infrastructures
Description succincte
(Anglais)
See abstract
Partenaires et organisations internationales
(Anglais)
Coordinator: Matra Systemes & Information (F)
Résumé des résultats (Abstract)
(Anglais)
Our achievements this year lies mainly on the completion of the population generation systems whose requirements definition and architectural design have been established and continuously fine-tuned since last year.
· Population face generation module
Based on previously constructed database of approximately 500 photo-cloned face models, population face generation module was developed to transform population control parameters into queries to the underlying database. Our population face generation module is responsible for loading records that meet the requests, proper positioning in the viewer, and the skin colour calculation through averaging pixel values of the cheek region that will be used when integrating the body to get a coherent avatar model.
· Population body generation module
Despite our initial efforts to develop software to generate photo-cloned seamless body models, it has been decided among partners early this year that we use rigidly segmented body models for the crowd simulation due to the expected computational overload to the system. Faces are considered to be relatively more important in identifying an individual and therefore we sacrifice the quality on the body part when a compromise is necessary. Accordingly, a texture coordinate generation method was newly developed for the rigidly segmented body models. We then could produce minimum required professions - fireman, policeman and workers - for the scenario realisation on the CROSSES system through appropriate texture coordinate generation from input photo images we collected at CSP.
In the meantime, we carried out surveys in anthropometry to develop ways of deforming and resizing generic body models to produce a population. A population body generation module was developed that contains statistical information and dynamically distribute the measurement values on selected parts of human body in the target population.
· Integration of components and user interface
A user interface has been implemented aiming at a population generation system that provides an intuitive and versatile way of controlling the target population. Through the integration of population face and population body modules respectively with the user interface, we completed and delivered the first version of the population generation system together with the population models generated. This includes the development of an H-Anim exporter to allow the final description of avatar models in a standard way. After fine-tuning and internal testing of the system, yet another version has been released and installed at Matra and CSP together with a user manual.
· System enhancement
Population face generation system was then continuously extended, focusing on the following issues:
(1) Enhancement of the database, particularly focusing at obtaining aged persons. Apart from direct acquisition through reconstruction of real people, a morphing technique was developed for indirect acquisition of models through geometric morphing and texture blending of existing models.
(2) Generation of different LoD(Levels of Detail) models. We studied and developed mesh simplification methods to produce different LoD face models. Further exploration of LoD techniques applicable for human body models was partially accomplished.
(3) Deeper study of statistics on anthropometry for various parts of the human body.
Références bases de données
(Anglais)
Swiss Database: Euro-DB of the
State Secretariat for Education and Research
Hallwylstrasse 4
CH-3003 Berne, Switzerland
Tel. +41 31 322 74 82
Swiss Project-Number: 99.0677-1