ESA > Join & Share > Technology Projects > PIMS-DLR Project

PIMS-DLR Project

Project Title   Partner Information Mining Services System-DLR
Project Acronym   PIMS-DLR
Contractor(s)   DLR,  ACS, Werum


Image Context
          How it works


Context   Top

In recent years our ability to store large quantities of data has greatly surpassed our ability to access and meaningfully extract information from it. Emerging needs from big applications (e.g.: change detection, global monitoring, disaster management support, etc.) and the continuous increase in archives' size and Earth Observation (EO) sensors' variety require new methodologies and tools for information mining and management, supported by shared knowledge. The manual process performed by experts to mine information from images is currently too complex and expensive to be applied systematically on even a small subset of the acquired scenes. This limits the full exploitation of the petabytes of archived or new data. The issue might become even more challenging in future since more missions - including constellations - are being planned, with broader sensor variety, higher data rates and increasing complexity. This raises the need for knowledge supported automatic or semi-automatic feature recognition and classification techniques to replace the manual interpretation activity. Preliminary results from recent R&D activities are showing the possibility to use Image Information Mining (IIM) techniques to ease the access to the imagery also through their information content.

IIM techniques include:

  Image Probabilistic Information Mining methodologies
  Image Feature Extraction Algorithms

These techniques, already successfully explored in other ESA TRP projects, needs to be put together in a common environment, in order to build a (pre)operational systems at partners’ facilities for the provision of services related to satellite image analysis and understanding. 

Objectives   Top

The PIMS-DLR project aims at simplifying access to and therefore at possibly expanding the use of multi-mission EO data. A relevant contribution to this achievement can come from the use of emerging technologies for Probabilistic Information Mining (PIM), which permit selection, retrieval and understanding  of images through their information content. The system implemented through the PIMS-DLR project shall permit to apply Probabilistic Information Mining and Feature Extraction Algorithms to a selected set of EO data available at the German Aerospace Center (DLR), possibly allowing an easy provision of processing services, e.g. through the ESA Service Support Environment (SSE) infrastructure or through the KEO prototype.

The main objectives of the PIMS-DLR project are listed below:

  Image Identify in cooperation with DLR:
    - Missions and data suitable for applying IIM techniques (with particular interest to very-high resolution images)
    - Relevant related  services which can be made available through the SSE or KEO environments
  Image Implement at DLR a pre-operational system permitting to ingest data, apply on them IIM techniques and create related services
  Image Integrate the services into the SSE and KEO environments
  Image Validate and demonstrate the system

The system is developed and configured in a way to make maximum profit from the outcomes of related studies and ongoing ESA projects in the field:

  Image Knowledge based Information Mining (KIM)
  Image KIM Validation for EO archived data exploitation support (KIMV)
  Image EO domain specific Knowledge Enabled Services (KES)
  Image Knowledge-centred Earth Observation (KEO)
  Image Survey and classification of advanced feature manipulation techniques and tools for Earth Observation applications (SURF)

Architecture   Top


Figure 1: Architectural overview

The PIMS-DLR system processing functionalities (e.g. data ingestion, collection management, image analysis) is essentially based on the KIM and KEO system architectures (see related projects for details). The system is also linked to the Data Information and Management System (DIMS) of the German Aerospace Center (DLR). Figure 1 gives an overview of the overall architecture of PIMS-DLR system, and describes the modules developed for connecting the KIM-KEO part to DIMS:

  Image The User Authentication Service, which provides a method for an authorised access (according to DLR access rules)
  Image The DIMS Catalogue Access Services, which provides operations to access the DIMS catalogue
  Image The New Product Notification Service, which triggers the ingestion of new DIMS products into PIMS-DLR
  Image The Ingestion Management Service, which performs the ingestion of products from DIMS to PIMS-DLR
  Image The Feature Enrichment Service (or DIMS Ingestion Service), which permits to ingest PIMS-DLR metadata and products into DIMS

The Differential Query Service, which permits to query both DIMS and PIMS-DLR catalogues in order to identify DIMS products not yet ingested in the PIMS-DLR system.

The New Product Notification Service, the User Authentication Service and the DIMS Catalogue Access Services aim at adapting DIMS to the outer world, allowing authenticated users to gather access to DIMS products and to be notified when new products (e.g. from the TerraSAR-X mission) are available. The Feature Enrichment Service will permit, on the other hand, to add new information (e.g. feature maps extracted from images using IIM techniques) into the DIMS catalogue.

The Ingestion Management Service and the Differential Query Service allow the interaction between PIMS-DLR and DIMS. The former adapts the KEO ingestion process to the needs of PIMS-DLR; the latter permits to discover DIMS products not yet ingested into PIMS-DLR and to automatically trigger the ingestion, if necessary.

In addition to this, the PIMS-DLR system provides all basic processing functionalities of KIM (e.g. collections creation and management, interactive feature labels definition, catalogue searches, etc.) through a new advanced user interface, which also allows authorised users to easily publish derived services on the ESA Service Support Environment (SSE) or to deploy them into the KEO environment.

How it works   Top


The KAOS client application permits (through an easy user interface) the system administrator to control and manage all PIMS-DLR functionalities:

  Image Access and browse the DIMS catalogue (for collections available to PIMS-DLR)
  Image Create new collections into PIMS-DLR for IIM activities 
  Image Ingest new images (either automatically or manually) into already available PIMS-DLR collection
  Image Perform the interactive training for identifying relevant features
  Image Make significant identified features accessible to DIMS and to SSE / KEO

Moreover, some PIMS-DLR functionalities are also available to authenticated SSE users through specific SSE services, which will permit:

  Image Access, browse and order products from the DIMS catalogue (for collections available to PIMS-DLR)
  Image Access to features extracted by IIM activities, either through PIM activities or via specific Feature Extraction Algorithms

In the quasi-operational configuration of the system, when a new product becomes available into the DIMS Product Library, the Ingestion Management Service will automatically upload it into an image collection (specified by the system administrator) of PIMS-DLR. The ingestion of products (from DIMS into PIMS-DLR) can be also user-driven, if necessary, i.e. resulting from a differential query on both systems. Moreover, the PIMS-DLR system will permit to upload significant features, extracted by users through IIM, into DIMS.


Figure 2: Snap shot of the Image Browser panel provided by the KAOS client application. The window permits to browse collections into the PIMS-DLR system, to select products and possibly order them from the DIMS Product Library.

Authenticated users may also access and browse the DIMS catalogue via the SSE DIMS Catalogue Access Service, which allows to perform catalogue searches and to order products from DIMS collections published on SSE.




Figure 3: Snap shot of the DIMS Catalogue Access Service on SSE.

Output   Top

The PIMS-DLR project aims at developing a quasi-operational system to fully exploit Image Information Mining capabilities, permitting to apply Probabilistic Information Mining and Feature Extraction Algorithms to a selected set of EO data available at the German Aerospace Center (DLR). The system allows to:

  Image Access to available data and create specific collections for data analysis
  Image Perform the interactive training for identifying relevant features on ingested images
  Image Create SSE services derived from significant extracted features
  Image Create KEO processing chains based on developed Feature Extraction Algorithms

The PIMS-DLR Final Presentation has been held on June 23, 2009.

Results of analysis and validation activities carried out in PIMS-DLR are available at the following link. Such activities also led to the definition of a new validation procedure for assessing the functionalities of the KIM system



Contributors to this page: andreadv .

Page last modified on Wednesday 22 of December 2010 16:07:04 CET by andreadv.