Orfeo ToolBox: a Free and Open Source Solution for Research and Operational Remote Sensing Projects
Savinaud, Mickael
CS-SI, FRANCE

The Orfeo Toolbox (OTB) is an open source (CeCILL license, similar to GPL), remote sensing-oriented, image processing library [1]. It has been initiated by the French Space Agency (CNES) in the frame of the ORFEO accompaniment program [2]. Based on the medical image processing library Insight ToolKit (ITK), OTB provides to its users an extensive set of algorithms and functionalities dedicated to remote sensing data exploitation. More specifically, it embeds approaches to handle large data using advanced streaming and multi-threading strategies. Thus, OTB-based processing chains take advantages of both optimized Input/Output access and streamed/multithreaded filtering to perform efficient processing.

OTB is based on a large set of open source libraries which contribute to expand the OTB ecosystem. The main part of this ecosystem is described in the figure 1. Through the use of these libraries, it offers the possibility to manipulate and process the main remote sensing data from optical to hyperspectral through SAR sensors. The processing capabilities of OTB library cover a large set of uses cases from optical calibration and data projection to image segmentation and classification. An exhaustive list can be found in the documentation.


Figure 1: Orfeo ToolBox ecosystem: enhance and improve gold standard libraries in large data image processing framework.

Currently Orfeo ToolBox ships the following tools to expose the main functionalities of the library:

  • Monteverdi the Orfeo composer which provides a GUI which can handle large data and their processing,
  • A command-line launcher,
  • A graphical launcher based on a Qt interface which provide ergonomic parameters setting, display of documentation, and progress reporting,
  • A SWIG interface,which allows to use any application into a high-level language such as Python or Java for instance.
    Moreover, through the Sextante project, OTB applications are also available into the QGIS software. QGIS users have therefore an easy way to combine these processing with other tools. We can illustrate these functionalities into the following figure.

    Figure 2: Example of large scale segmentation result from a Pleiades image displayed in QGIS (extract from [3]).

    As an open source project, OTB exposes its source code via a public official repository. Moreover, the library is built and tested on a nightly basis. The results are available publicly which ensures multi-platform consistency and continuous validation. Last, OTB encourages full access to the details of all the algorithms through extensive documentation. This documentation is divided between class documentations, the software guide [4] which gives some technical remote sensing processing background along with code examples and a cookbook [5] dedicated to non-developers.

    Since the beginning in 2006, OTB is uses to promote the development of new methodologies in the field of remote sensing based on the requirements express by the Orfeo community for Pleiades data. Moreover it was intensively used in different R&D studies funded by the CNES for general purpose in image processing. For example we can notice the work done about multi-temporal series analysis by the CESBIO [6] with OTB. In the field of processing chain, Venµs L2 and L3 product generation have been the first operational use case of the library. The entire framework and the image processing algorithms (cloud and aerosol detection, atmospheric corrections, temporal analysis) of this project have been built on OTB. The robustness and performance of the solution have given the possibility to extend the processing chain to other sensor products (Landsat 5-7, Formosat, Sentinel-2) into the MACCS project (Multi-mission Atmospheric and Cloud Correction Software).

    More recently the OTB has been integrated into the operational Sentinel-2 Instrument Processing Facility (S2-IPF) which produces L1 products to make all radiometric processing. In the same context, geometric quality assessment of the Sentinel-2 Mission Performance Assessment project (S2-MPA) will be based on the OTB correlation framework. The strong processing framework of OTB has been also underline by EUMETSAT for this study about definition and prototyping of a new methodology for the Geometry Quality Assessment of the next MTG satellite.

    The large spectrum of possibility offers by OTB covers also some end-users applications. Different projects begin to use OTB into their core to provide geospatial analysis:

  • PEPS: a project created to define the new use of remote sensing data into the definition of public policy.
  • SATERRE: a project created to define new added value service for farmers and to help regional authority into their land use policy.
    This two projects have made the choice of OTB as image processing framework mainly because its open source policy give a better control on the methodology. Moreover the methodology can be easily open to users and guarantee a strong viability of the solution.

    Other projects are probably done with OTB in the field of remote sensing or outside without formal feedback to the project. We encourage the use of the library wherever possible with a strong documentation and with open-source policy about the project. The capacity of OTB library to provide numerous image processing algorithms and processing frameworks is one of the key points of its success. The support of large data for new earth observation satellites (Sentinels, Pleiades, next Spot constellation) is another key point of the development of the library. We hope that OTB will continue to encounter the user needs to develop a strong community.

    [1] Inglada, J., and Christophe, E. The Orfeo ToolBox remote sensing image processing software. In IGARSS 2009, IEEE, pp. IV-733.
    [2] Tinel, C., Fontanaz, D., De Boissezon, H., Grizzonet, M., and Michel, J. The Orfeo accompaniment program and Orfeo ToolBox. In IGARSS 2012, IEEE, pp. 7102-7105.
    [3] Michel J. et al. Large-scale segmentation of Very High Resolution satellite images using Orfeo ToolBox, submitted to JOSIS, 2013
    [4] Team, O. D. The ORFEO Tool Box Software Guide - Updated for OTB-3.16. 2013.
    [5] Team, O. D. The Orfeo ToolBox Cookbook, a guide for non-developers - Updated for OTB-3.16. 2013.
    [6] Inglada, J., Dejoux, J., Hagolle O. and Dedieu, G. Multi-temporal remote sensing image segmentation of croplands constrained by a topographical database, In IGARSS 2012, IEEE, pp. 6781-6784.