Release of OTB 3.8 and Monteverdi 1.6

We are happy to announce the new 3.8.0 OTB release, codename “Pack Ice”. Simultaneously, we are releasing Monteverdi 1.6, the all-in one user-friendly graphical tool for remote sensing image processing and also a new release of OTB-Applications and OTB-Wrapping projects.

We are also happy to announce that starting 3.8.0, packages are available through online repositories for the 3 last Ubuntu versions as well as for OpenSuse ! These packages include the OTB library, the OTB-Applications and Monteverdi. Java and Python wrapping packages will be available very soon. Our next post will explain how to update your system repositories so as to get the last OTB installed in no time and you can already find details about the procedure on the Download page. For windows users, installers are available for Monteverdi and OTB-Applications and Wrappings. Last novelty, a Monteverdi DMG is available for our Mac users (intel only).

There are numerous improvements to the library :
– Improved local descriptors tools based on ImageFunction :
– Rework FlusserImageFunction and HuImageFunction to output all moments in one pass (FlusserMomentsImageFunction, HuMomentsImageFunction)
– Rework RealMomentsImagefilter and ComplexMomentImageFilter to output a matrix of all moments associated to p, q inferior to a given parameter (ComplexMomentsImageFunction, HuMomentsImageFunction)
– Add image function to compute a vector containing the local mean, variance, skewness and kurtosis (RadiometricMomentsImageFunction)
– Add local histogram image function (LocalHistogramImageFunction)
– Add image function to compute the local Fourier Mellin coefficients (FourierMellinDescriptorsImageFunction)
– Add a class to adapt any image function return types to itk::VariableLengthVector (ImageFunctionAdaptor)
– Add a class to build composite image functions (MetaImageFunction)
– New object detection framework (see also http://wiki.orfeo-toolbox.org/index.php/Object_detection)
– Add filter to generate negative samples (LabeledSampleLocalizationGenerator)
– Add filter to evaluate an image function on a set of point and generate ListSample (DescriptorsListSampleGenerator)
– Add filter to balance the sample number of different classes in a ListSample by generating new samples from existing ones plus noise (ListSampleToBalancedListSampleFilter, GaussianAdditiveNoiseSampleListFilter)
– Add filter to apply a shift/scale to a ListSample (ShiftScaleSampleListFilter)
– Add filter to detect object from an SVM model and an image function (ObjectDetectionClassifier)
– SVMClassifier: add hyperplanes distances as output
– GDALImageIO: support writing of non-streamable format (JPEG, PNG)
– Support reading vector images of std::complex
– BandMathFilter: add physical and image coordinates variables

Monteverdi 1.8 comes with 3 new modules and 2 modules improvements :
– Monteverdi
– New Polarimetric Synthesis module
– New DEM image extraction / HillShading module: creates an image from a DEM tiles directory, with optional hillshading
– New ColorMapping module: apply a colormap to a mono band image
– Viewer module: add multi input support, with slideshow or transparency mode, add more rendering functions, add splitted/packed layout option
– Vectorization module: new semi-automatic mode based on segmentation results proposals.

Regarding the OTB-Applications package, it has been completely revamped and offers the following new features:
– Object Detection applications (see also http://wiki.orfeo-toolbox.org/index.php/Object_detection)
– EstimateFeatureStatistics to evaluate descriptors statistics on a set of images
– TrainObjectDetector: generates an SVM model from input images and a vector data
– ObjectDetector: detects points in an image from an SVM model
– Add automatically generated GUI wrappers for OTB processing chains
– Add FLTK wrapper
– Add a generic Qt based widget class
– Add a Qt GUI wrapper based on the generic Qt widget
– Add a Qgis plugin wrapper based on the generic Qt widget

The OTB Software Guide has also been updated and is now 700 pages long.

Also, the otb-developers mailing list announced with the 3.6 release is a great success: we had lots of registrations, with very interesting contributions to the project development. Many thanks to the contributors !

On behalf of the OTB development team, we wish you a merry christmas and an happy new year,