NL FR EN
www.belgium.be

Reliable Image Management Systems in support of urban services and disaster management (RIMS)

Research project SR/00/106 (Research action SR)

Persons :

  • Dr.  GAUTAMA Sidharta - Universiteit Gent (UGent)
    Coordinator of the project
    Financed belgian partner
    Duration: 1/12/2006-30/6/2008
  • Dr.  CAENEN Geert - Incubatiepunt voor Geoinformatie (INCGEO)
    Financed belgian partner
    Duration: 1/12/2006-30/6/2008
  • Dr.  BIESEMANS Jan - Vlaamse Instelling voor Technologisch Onderzoek (VITO)
    Financed belgian partner
    Duration: 1/12/2006-30/6/2008

Description :

Context

The High Altitude Long Endurance Unmanned Aerial Vehicle (HALE UAV) is a valuable missing link to the traditional platforms used in Remote Sensing (RS) with respect to (a) disaster monitoring, in response to the user requirement of near-real-time product delivery and (b) photogrammetic application in response to the user requirement of the regional availability of decimeter resolution imagery. Due to design limitations, the ground segment shall completely rely on computer vision techniques to register the imagery geometrically.

The aim of this project is to introduce a novel methodology for Image Information Management within the registration process of the processing and archiving facilities. Image registration is the process of matching overlaying images and/or vector information of the same scene taken at different times, from different viewpoints, and/or by different sensors. Several techniques exist and are described in the literature that allow automated registration. The main bottleneck with automated registration techniques however is quality assurance.


Objectives

1) Design and assembly of image information management (IIM) tool
2) Integration of the tool(s) in the Pegasus processing chain
3) Scenario testing: performance under different boundary conditions
4) Scientific publication and promotion


Methodology

Techniques for performance evaluation and characterization will be applied to solve the problem of quality assurance and to develop an image information management (IIM) system that allows the best processing chain in the PAF to be selected in near-real time based on the image scene content and external data at hand. By analysing the image information and predicting the expected quality of the registration result given this data, the best set of techniques can be selected for processing. In addition, depending on the requirements of the applications, the needed spatial quality and speed of processing can vary. To be able to manage this, full control over the registration algorithms and information processing is required in terms of performance prediction. In addition, the expected positional quality is documented as metadata and can be used by the operator to decide on additional intervention. This is essential information in the framework of operational UAV mission planning. The methodology however is generic and is applicable to other processing systems.


Results expected

The main outcome of the RIMS project is a methodology for Image Information Management within the process of image registration. This will allow the optimal automatic choice of registration algorithm based on the available image information content and the estimation of the expected local positional quality.

Deliverables:
D1.1: Blue print of IIM integration within PAF
D1.2: Operational scenarios for stress and validation tests

D2.1: Report on state-of-the-art in image registration
D2.2: Methodology for IIM in image registration

D3.1: C++ library for image registration
D3.2: C++ library for IIM in image registration
D3.3: MEDUSA image simulator
D3.4: PAF test bed

D4.1: Report on performance IIM registration for selected operational scenarios

D5.1: Project web site
D5.2: Participation in intl. conferences (IEEE IGARSS, ESA, ISPRS)
D5.3: Publication submitted to peer reviewed journal

Documentation :