Research in 3D computer vision is focused on the two following applications:
3D surveillance and identification technologies
3D site modelling and verification of plant design
3D Surveillance and Identification Technologies
There are situations where visual surveillance proves ineffective in detecting changes, which may be either too small to be reliably detected or generated by inconsistent illumination. Further, illumination blackouts, prevent cameras from their surveillance function. High security installations need reliable, independent and effective sensors capable of maintaining continuity of knowledge. In the past, the SIR sector introduced the concept of laser surveillance (both in 2D and 3D) to complement existing video surveillance systems. This concept was later extended for the identification of fuel elements and other nuclear containers and research is undergoing on the use of this technology for container authentication. There is increasing awareness from the inspectorates on the potential of 3D technologies for different applications, namely High Security Surveillance, and Monitoring Nuclear Containers (Fuel Elements, waste drums, …).
The research aim at:
continuing the developments in laser surveillance driven by IAEA requests,
extending the concept of laser surveillance to higher dimensions by including multi-sensors aiming at unattended fault tolerant monitoring systems,
continuing the work for ESO on the identification and authentication of nuclear containers.
3D Site Modelling and Verification of Plant Design
It is important in Safeguards to detect changes made in a given installation or track the progression of the construction work in a new plant. This problem is commonly known as Design Information Verification (DIV). The process of DIV can be divided into two main steps: i) acquire raw data from the plant; ii) compare the "as-built" data and the reference model. The objective of this work is to develop a prototype accepting multi-sensory, variable scale data as input. Scalability allows for different acquisition systems and algorithms according to the size of the objects/buildings to be modelled. Indeed, a building, large tanks or small pipes require different equipment and modelling algorithms. Semi-automated tools support the comparison between the acquired "as-built" and the approved design models. The proposed system is a scaled based approach combining different sensors and 3D reconstruction techniques depending on the size of the object to be modelled and accuracy of the final model. Both software and hardware are designed to cover a wide range of distances between 1mm to 200m. This includes 3D modelling of buildings, large objects, e.g., fuel containers, large pipes and very small pipes. A further objective of this project is to profit from the site modelling tools of large facilities and extend them to plan rescue or intervention strategies after severe accidents. Nuclear safety considerations could be easily embedded into the final plan.
Software packages:
Reconstructor, (developed by the JRC,
distribuited by Inn.TEC), software for 3D laser data processing: http://www.topotek.it/pagine/software_E.htm
Scansoft Capture Development System v11.0 (Scansoft Inc.), optical character recognition SDK: http://www.scansoft.com/capturesdk/
HALCON
(MVTech Software GmbH), software libraries for machine vision applications: http://www.mvtec.com/halcon/
Hardware:
Shapegrabber SG100/SG1000 (Shapegrabber Inc.), laser scanning head for small/medium distances: http://www.shapegrabber.com/ScanHeadsSG100.shtml
Shapegrabber Rotation/Linear Positioning Stage (Shapegrabber Inc.): http://www.shapegrabber.com/Products.shtml
Physik Instrumente Rotation/Linear Positioning Stage M-531.PD (Physik Instrumente GmbH):`http://www.physikinstrumente.de/products/prdetail.php?secid=7-40
Sick
LMS200, 2D laser Scanner: http://www.sick.de/de/products/categories/auto/lasermeasurementsystemsindoor/lms200indoor/en.html
Riegl LMS-Z210, 3D laser Scanner: http://www.riegl.co.at/lms-z210i/e_lms-z210i.htm
Z+F IMAGER 5003, LARA 53500: http://www.zf-laser.com/e_bildgebende.html
Vialux
z-Snapper 3D camera: http://www.vialux.de/snapper.pdf
1.
Combined
Video and 2D/3D Laser Surveillance
This project is part of the European Union Support Programme to IAEA. The objective is to develop a multi-sensory surveillance system aiming at unattended fault tolerant monitoring systems.
2.
3D
Identification of nuclear containers
This
project is a task commissioned to SIR sector by DG TREN I - Euratom Safeguards.
The aim is to use laser-scanning technologies alone or combined with a camera to
provide a robust and efficient way for reading and automated recognition of
identification numbers.
One of the developed systems is installed in a nuclear fuel fabrication plant. It is integrated in an Unattended Monitoring Station (UMS), where it scans and reads an identification string engraved in the head of the fuel element.
This project is a Collaboration project between Oak Ridge National Laboratory (USA) and Joint Research Centre. The aim is to Design, implement and testing of a Laser Scanning System to provide immediate and continuous monitoring of the integrity of items. The security of an item is verified by determining that the target is still present and that its position has not changed.
4.
Semi-automated DIV System accepting multi-sensory, multi-scale data
This project is part of the European Union Support Programme to IAEA. The aim is to develop equipment, techniques and procedures to quickly and efficiently scan areas that are under Safeguards to record significant constructional features. The system (including hardware, software and associated procedures) is required to record those features and to indicate conclusively if changes have taken place between DIVs.
5.
Automatic scene change analysis of large areas
The objective of this exploratory research is the development of an automatic scene reconstruction tool for damage assessment of large areas. This research proposes a method to detect the change specifically associated with structural and major damage caused by natural disasters such as a strong earthquake, floods and accidental or terrorist explosion. To achieve this it utilizes pre- and post-disaster (bi-temporal) multi sensory and multi scale imagery.
The
implementation of the Additional Protocol to the Non-Proliferation Treaty
requires modelling capabilities from outdoor facilities in order to support IAEA
inspectors while doing random, short notice or regular inspections. The results
of this exploratory research are expected to contribute towards the automatic 3D
site modelling and verification of large environments.
JRC
3D Reconstructor®, 3D Reconstruction and Verification software.
Analysis of techniques for managing and processing huge 3D data models
3D Reconstruction and trajectory planning for an 8 DoF Mobile-Robot Platform
Safeguards Simulation and Training through Augmented Reality
Vitor
Sequeira (
),
Axel Busboom, Gunnar Boström, Guillaume Boutin, Giuseppe Donato, Marco Fiocco, Thomas Hollands,
David Puig, Martino Salvato and Elena Stringa