Feature Requests

From OTBWiki
Jump to: navigation, search

This page intends to be summarize all features requests related to the OTB project in general.

After review, these requests will eventually be moved to the OTB Backlog.

The recommended procedure is to file a ticket in the OTB bug tracker, making sure to set the Severity field to feature. Your request should then appear automatically on this page. If your request does not fit into one of the declared project in the bug tracker, fell free to edit this page.

OTB Library

<rss>http://bugs.orfeo-toolbox.org/issues_rss.php?username=julienm&key=ede468243b769ad2a2c1925b9d92d565&project_id=1&filter_id=80</rss>

  • Geoeye-1 sensor model
  • Hough based and/or genetic and/or edge-following algorithms for circle/ellipse extraction
  • Bias compensation of RPC model (Willneff et al, 2005)
  • CFAR target detection in SAR intensity image with several distributions (e.g. Weibull, K), Local threshold using lookup table
  • Add skewness/kurtosis to StatisticsImageFilter
  • Histogram matching between to images (transform the histogram of one image to fit the histogram of another one, as a preprocessing for change detection for example)

Monteverdi

Requests related to the Monteverdi application and to module upgrade / new modules.

<rss>http://bugs.orfeo-toolbox.org/issues_rss.php?username=julienm&key=ede468243b769ad2a2c1925b9d92d565&project_id=4&filter_id=81</rss>

  • Shape extraction module (such as line, circle, ellipse, rectangle, parallelepiped...)
  • Multiscale segmentation module
  • Importing new features for classification in object labeling module
  • Better modularization of Monteverdi : in the current state, the Monteverdi application is hard-linked with all the modules contained in the Monteverdi sources. To develop a new module, we need to modify the Monteverdi.cxx file, recompile it and relink it with all the modules. To allow better customization of Monteverdi (including a specific module that is not intended to end up in the official Monteverdi sources, or replacing an existing module by a more specific one), we should go to a more plugin-based sheme, where modules are shared libraries that are auto-discovered by the Monteverdi executable at run time. ITK already provides all the necessary framework to do that, with the ObjectFactory : shared libraries providing an itkLoad symbol and available in the ITK_AUTOLOAD_PATH are automatically loaded, providing a way to add new factories for arbitrary ITK classes. This can be used to provide several implementations for a common interface (like ImageIO, ImageMetadataInterface, with ObjectFactoryBase::CreateAllInstance), or to replace the implementation of an existing class (with ObjectFactoryBase::CreateInstance). My guess is that we should look in that direction to provide both functionnalities (extending Monteverdi by providing new modules shipped in plugins & allow overrides of official modules). Some advantage of such a scheme would be that the Monteverdi application would become very stable (Monteverdi.cxx should not be modified a lot), that compilation of all modules would be an independent process, and so that the failure of compilation of one module would not interfere with the availability of the others. We must be aware that some modules (at least Viewer, Caching, Writer) must be known by the Monteverdi application (because they appear in the contextual menus)
  • A monteverdi module for monoplotting: 3D information extraction from a single image. This photogrammetric technique consists in using (bias-corrected) RPC model and DTM for precise 3D measurement (Willneff et al, 2005). Two ways for elevation measurement: from shadow and directly from off-nadir image (Huang & Kwoh, 2008)(maybe not the best reference).
  • Save a scenario with adding the possibility of load/save a Monteverdi scenario (store via a XML file the pipeline description)

Related to the feature request in mantis: http://bugs.orfeo-toolbox.org/view.php?id=127

  • Viewer Module or other: ablility to plot x- or y- profile from the Zoom view.
  • Polarimetric Synthesis Module: Displaying the polarimetric signature (co- cross- polarisation) for a given pixel.
  • DONE Image Fusion Module: Allowing to perform Pan sharpening or Bayesian data fusion.
  • executable to access pixel description: in OTB-Applications
  • Add functionality like visualization in Filtering -- Vectorization module of Monteverdi.Since few features can be easily discriminated using False color composite than Natural color.
  • Add a 'image/vector data properties' feature in the right click popup menu in Monteverdi main window to display image/vector data information and metadata entries (size, projRef...).

Applications

<rss>http://bugs.orfeo-toolbox.org/issues_rss.php?username=julienm&key=ede468243b769ad2a2c1925b9d92d565&project_id=2&filter_id=82</rss>


Integration with external tools

Requests related to binding to other languages as well as integration or communication with external tools, such as Google Earth or QGis.

<rss>http://bugs.orfeo-toolbox.org/issues_rss.php?username=julienm&key=ede468243b769ad2a2c1925b9d92d565&project_id=3&filter_id=86</rss>

<rss>http://bugs.orfeo-toolbox.org/issues_rss.php?username=julienm&key=ede468243b769ad2a2c1925b9d92d565&project_id=8&filter_id=84</rss>

  • I'd like to see an OTB image processing tool connected with OssimPlanet, i means a separate gui based on PyQt to select and process images, using OTB python bindings, and send the results back to ossimplanet. I'm trying to work on that direction, see PlanetSasha inside the gsoc code in ossim, for now i'm working on a GRASS integration, later i'd like to try to integrate OTB as well.
  • GRASS-OTB connection - suggestion: via Python (GRASS offers Python-ctype bindings)


Packaging

<rss>http://bugs.orfeo-toolbox.org/issues_rss.php?username=julienm&key=ede468243b769ad2a2c1925b9d92d565&project_id=9&filter_id=89</rss>