The OTB Applications package makes available a set of simple software tools, which were designed to demonstrate what can be done with Orfeo Toolbox . Many users started using these applications for real processing tasks, so we tried to make them more generic, more robust and easy to use. Orfeo Toolbox users have been asking for an integrated application for a while, since using several applications for a complete processing (ortho-rectification, segmentation, classification, etc.) can be a burden. Recently, the OTB team received a request from CNES’ Strategy and Programs Office in order to provide an integrated application for capacity building activities (teaching, simple image manipulation, etc.). The specifications included ease of integration of new processing modules.
The application is called Monteverdi , since this is the name of the Orfeo composer. The application allows you to build interactivelly remote sensing processes based on the Orfeo Toolbox . This is also in remembering of the great (and once open source) Khoros/Cantata software.
Installation of Monteverdi is very simple. Standard installer packages are available on the main platforms thanks to OTB-Developpers and external users. These packages are available few days after the release. Get the latest information on binary packages on the Orfeo ToolBox website in the section download.
We will discribe in the following sections the way to install monteverdi on:
- Windows platform (XP/Seven)
- Ubuntu 12.04 and higher
- OpenSuse 12.X and higher
- MacOSX 10.8
If you want build from source or if we don’t provide packages for your system, some informations are available into the OTB Software Guide , in the section (Building from Source)
For Windows XP/Seven/8.1 users, there is a classical standalone installation program for Monteverdi, available from the OTB download page after each release.
Since version 1.12, it is also possible to get Monteverdi package
through OSGeo4W for Windows
XP/Seven users. Package for Monteverdi is available directly in the
OSGeo4W installer when you select the otb-monteverdi package. Follow
the instructions in the OSGeo4W installer and select the
otb-monteverdi. The installer will proceed with the installation of
the package and all its dependencies. Monteverdi will be directly
installed in the OSGeo4W repository and a shortcut will be added to your
desktop and in the start menu (in the OSGeo4W folder). You can now use
directly Monteverdi from your desktop, from the start menu and from an
OSGeo4W shell with command monteverdi. Currently, you should use the
32bit OSGeo4W installer but we will soon distribute monteverdi package
for 64 bit installer.
A standard DMG package is available for Monteverdi for MacOS X 10.8. Please go the OTB download page . Click on the file to launch Monteverdi. This DMG file is also compatible with MacOSX 10.9.
For Ubuntu 12.04 and higher, Monteverdi package may be available as Debian package through APT repositories.
Since release 1.14, Monteverdi packages are available in the ubuntugis-unstable repository.
You can add it by using these command-lines:
sudo aptitude install add-apt-repository sudo apt-add-repository ppa:ubuntugis/ubuntugis-unstable
Now run:
sudo aptitude install monteverdi
If you are using Synaptic, you can add the repository, update and install the package through the graphical interface.
apt-add-repository will try to retrieve the GPG keys of the repositories to certify the origin of the packages. If you are behind a http proxy, this step won’t work and apt-add-repository will stall and eventually quit. You can temporarily ignore this error and proceed with the update step. Following this, aptitude update will issue a warning about a signature problem. This warning won’t prevent you from installing the packages.
For OpenSuse 12.X and higher, Monteverdi packages is available through zypper.
First, you need to add the appropriate repositories with these command-lines (please replace 11.4 by your OpenSuse version):
sudo zypper ar http://download.opensuse.org/repositories/games/openSUSE_11.4/ Games sudo zypper ar http://download.opensuse.org/repositories/Application:/Geo/openSUSE_11.4/ GEO sudo zypper ar http://download.opensuse.org/repositories/home:/tzotsos/openSUSE_11.4/ tzotsos
Now run:
sudo zypper refresh sudo zypper install Monteverdi
Alternatively you can use the One-Click Installer from the openSUSE Download page or add the above repositories and install through Yast Package Management.
There is also support for the recently introduced ’rolling’ openSUSE distribution named ’Tumbleweed’. For Tumbleweed you need to add the following repositories with these command-lines:
sudo zypper ar http://download.opensuse.org/repositories/games/openSUSE_Tumbleweed/ Games sudo zypper ar http://download.opensuse.org/repositories/Application:/Geo/openSUSE_Tumbleweed/ GEO sudo zypper ar http://download.opensuse.org/repositories/home:/tzotsos/openSUSE_Tumbleweed/ tzotsos
and then add the monteverdi packages as shown above.
[fig:mainwindow]
This is Monteverdi’s main window (figure [fig:mainwindow]) where the menus are available and where you can see the different modules, which have been set up for the processing. Input data are obtained by readers. When you choose to use a new module, you select its input data, and therefore, you build a processing pipeline sequentially. Figure [fig:inputswindow] shows the generic window which allows to specify output(s) of Monteverdi’s modules.
[fig:inputswindow]
Let’s have a look at the different menus. The first one is of course the “File” menu. This menu allows you to open a data set, to save it and to cache it. The “data set” concept is interesting, since you don’t need to define by hand if you are looking for an image or a vector file. Of course, you don’t need to do anything special for any particular file format. So opening a data set will create a “reader” which will appear in the main window. At any time, you can use the “save data set” option in order to store to a file the result of any processing module.
The application allows to interactively select raster/vector dataset by browsing your computer. Monteverdi takes advantage of the automatic detection of images’ extensions to indicate the dataset type (optical, SAR or vector data).
The input dataset is added to the “Data and Process” tree, which describes the dataset content and each node corresponds to a layer.
This module allows to visualize raster or vector data. It allows to create RGB composition from the input rasters. It is also possible to add vector dataset which are automatically reprojected in the same projection of the input image or Digital Elevation informations.
The viewer offers three types of data visualisation:
- The Scroll window : to navigate quickly inside the entire scene
- The Full resolution window: the view of the region of interest selected in the scroll window
- The Zoom window
- The Pixel description: give access to dynamic informations on the
current pixel pointed. Informations display are:
- The current index
- The pixel value
- The computed value (the dynamic of hte input image is modified to get a proper visualization
- The coordinates of the current pixel (longitude and latitude)
- In case where there is a Internet connection available, Monteverdi displays the estimate location of the current pixel (country + city)
[fig:viewerpixeldescription]
The Visualization offers others great functionnalities which are available in the detached window. It is for example possible to superpose vector dataset to the input image (see figure [fig:viewervectordata]).
[fig:viewervectordata]
The “Setup Tab” allows to modify the RGB composition or use the grayscale mode to display only one layer.
[fig:rgbcomposition]
The “Histogram Tab” get access to the dynamic of the displayed layers. The basic idea is to convert the output of the pixel representation to a RGB pixel for rendering on conventional displays. Values are constrained to 0-255 with a transfer function and a clamping operation. By default, the dynamic of each layer is modified by clamping the histogram at min + 2\% and max - 2\%.
[fig:histogram]
There is also possible to select pixel coordinates and get access to all the informations available in the “Pixel description Box”.
[fig:pixeldescriptioninformations]
The “cache data set” (see figure [fig:cachingmodule]) is a very interesting functionality. As you know, Orfeo Toolbox implements processing on demand, so when you build a processing pipeline, no processing takes place unless you ask for it explicitly. That means that you can plug together the opening of a data set, an orthorectification and a spleckle filter, for example, but nothing will really be computed until you trigger the pipeline execution. This is very convenient, since you can quickly build a processing pipeline and let it execute afterwards while you have a coffee. In Monteverdi , the process is executed by saving the result of the last module of a pipeline. However, sometimes, you may want to execute a part of the pipeline without having to set the file name to the obtained result. You can do this by caching a data set. That is, the result will be stored in a temporary file which will be created in the “Caching” directory created by the application. Another situation in which you may need to cache a data set is when you need that the input of a module exists when you set its parameters. This is nor a real requirement, since Monteverdi will generate the needed data by streaming it, but this can be inefficient. This for instance about visualization of the result of a complex processing. Using streaming for browsing through the result image means processing the visible part every time you move inside the image. Caching the data before visualization will generate the whole data set in advance allowing for a more swift display. All modules allow you to cache their input data sets.
[fig:cachingmodule]
The aim of Monteverdi is to provide a generic interface which is based on the definition of the internal processes. In this frame, the way that you have to manage modules are identical during the definition of a new process. Selecting a module on the upper main window, open automatically the “Inputs definition Window” wich allows to select data which are inputs of the current module. Monteverdi module can manage single or multiple inputs and these inputs can be images on your computer or results of previous module already registered in the “Data and Process” tree.
Management of image formats in Monteverdi works in the same manner as in the Orfeo Toolbox . The principle is that the software automatically recognize the image format. Communication between modules follow also the same principle and the Input definition of modules request to all available outputs of the same type in the “Data and process” tree. Internally, all the treatments in Monteverdi are computed in float precision by default. It is also possible to switch to double precision by compiling the application from source and set the CMAKE option compile float to ON.
It allows to extract regions of interest (ROI) from an image. There are two ways to select the region:
- By indicating the X and Y coordinatres of the upper-left coordinates and the X-Y size of the regions.
- By interactivelly selecting the region of interest in the input image.
With Monteverdi , you could generate a large scale of value added informations from lots of inputs data. One of the basic functionnality is to be able to superpose result’s layers into the same dataset. Concatenating images into one single multiple-bands image (they need to have the same size), and to be able to create for example RGB composition with the inputs layer.
[fig:concatenate]
Monteverdi allows to export raster or vector dataset to a file to your system. In the case of raster images, it is possible to cast output pixel type. In Monteverdi all the processes are done in floating point precision. On large remote sensing dataset, saving your result in float data type could lead to file too large(more than 25 Go for pan-sharpened 8 bands WorldView2 with a resolution of 46 centimeters). Since the module allows to cast pixels in other types :
- unsigned char (8 bits)
- short (16 bits)
- int (32 bits)
- float (32 bits)
- double (64 bits)
- unsigned short (16 bits)
- unsigned int (32 bits)
[fig:exportdataset]
In the frame of remote sensing process, one common operation is to be able to superpose and manipulate data which come from different sources. This section gives access to a large set of geometric operations. It performs re-projection and orthorectification operations on Optical or SAR dataset using the available sensor models (image informations available in the meta-data are automatically read by the application).
The application is derived from the otbOrthorectificationApplication in the OTB Applications package and allow to produce orthorectified imagery from level 1 product. The application is able to parse metadata informations and set default parameters. The application contains 4 tabs:
- Coordinates: Define the center or upper-left pixel coordinates of the orthorectified image (the longitude and latitude coordinates are calculated through meta-data informations. It is also possible to specify the map projection of the output.
- Output image: The module allows to only orthorectified a Region Of interest inside the input dataset. This tab allows to set the size of the ROI around the center pixel coordinate or from the upper left index. The orthorectified imagery can also be resampled at any resolution in the line or column directions by setting the “Spacing X” and the “Spacing Y” respectively, and choosing interpolation method.
- DEM: Indicate path to a directory containing SRTM elevation file. The application is able to detect inside the direcory which DEM files are relevant in the process. You can find detailed informations on how to get a usable DEM
- Image extent: Compare the initial image extension with the preview the orthorectified result. This preview is automatically updated if the user change the “Size X” or “Size Y” values in the “Output Image” tab.
[fig:ortho]
This module allows to take ground control points on a raster image where no geographic informations are available. This GCPs list is making correspondence between pixel coordinate in the input image and physical coordinates. The list allows to derive a general function which convert any pixel coordinates in physical positions. This function is based on a RPC transformation (Rational Polynomial Coefficients). As a consequence, the module enriches the output image with metadata informations defining a RPC sensor model associated with the input raster. There are several ways to generate the GCPs:
- With Internet access: dynamically generate the correspondance on the input image and Open Street Map layers.
- Without Internet access: Set manually Ground control points : indicate index position and cartographic coordinates in the input image.
It is also possible to import/export the list of Ground Control points from/to an XML file.
Moreover, if the input image has GCPs in its metadata, the module allows to add or remove points from the existing list, which is automatically loaded.
In the solar spectrum, sensors on Earth remote sensing satellites measure the radiance reflected by the atmosphere-Earth surface system illuminated by the sun. This signal depends on the surface reflectance, but it is also perturbed by two atmospheric processes, the gaseous absorption and the scattering by molecules and aerosols.
In the case of the Optical calibration, the basic idea is to be able to retrieve reflectance of the observed physical objects. The process can be split in 3 main steps:
- Derived luminance from the raw value in the input image.
- Convert the luminance to reflectance to produce the TOA image(Top Of Atmosphere).
- Inverse a radiative transfer code, which simulates the reflection of solar radiation by a coupled atmosphere-surface system. This step produce the TOC (Top of Canopy) imagery, which is the final result of the optical calibration module.
The calibration and validation of the measurement systems are important to maintain the reliability and reproducibility of the SAR measurements, but the establishment of correspondence between quantities measured by SAR and physical measure requires scientific background. The SAR calibration module allows to estimate quantitative accuracy. For now only calibration of TerraSARX data is available.
The Band Math module allows to perform complex mathematical operations over images. It is based on the mathematical parser library muParser and comes with a bunch of build-in functions and operators (listed here ). This home-brewed digital calculator is also bundled with custom functions allowing to compute a full expression result simply and really quickly, since the filter supports streaming and multi-threading. The Monteverdi module provides an intuitive way to easily perform complex band computation. The module also prevents error in the mathematical command by checking the expression as the user types it, and notifying information on the detected error:
Figure [fig:bandmathndviwithres] presents an example on how the band math can produce a threshold image on the NDVI value computed in one pass using built-in conditional operator “if” available in the parser.
An other operational example, on how this simple module can produce reliable information. Figure [fig:ndwi2] shows the result of the subtraction of the Water indice on 2 images which was taken before and during the crisis event. The difference was produced by the band math module and allows to get a reliable estimation of the flood events.
[fig:bandmathndviwithres]
[fig:ndwi2]
The Connected Component Segmentation module allows segmentation and object analysis using user defined criteria at each step. This module uses muParser library using the the same scheme as it is done in Band math module (see [Band:sub:math module] for a detailled explanation). It relies on three main steps process :
- Mask definition :
This mask is used as support of Connected Component segmentation (CC) . i.e zeros pixels are not taken into account by CC algorithm. Binarization criteria is defined by user, via muparser. This step is optional, if no Mask is given, entire image is processed. The following example creates a mask using intensity (mean of pixel values) parameter :
intensity > 200
- Segmentation :
Connected Component Segmentation driven by user defined criteria. Segmentation process can be followed by a small object rejection step. The following example use distance (pixel intensity value difference ) parameter to define acception/rejection criteria between two adjacents pixels :
distance < 10
- Object analysis post processing :
This step consists in post processing on each detected area using shape and statistical object characterization. The following example use elongation parameter to test labeled objects :
SHAPE_elongation > 2
A detailled presentation of parameters and variables, can be found on the wiki .
Results are then exported in shape file format. Graphical user interface is presented on Figure [fig:connected:sub:componentmodule]. At each step intemerdiate output can be seen using Display item list. Viewing windows are updated by clicking on Update button. Available display outputs are :
- Input image :
- input image.
- Mask Output :
- mask image created using formula.
- Masked Image :
- input image multiplied by mask image.
- Segmentation Output :
- output of Connected Component segmentation filter.
- Segmentation after small object rejection :
- output of Connected Component segmentation after relabeling and small object rejection.
- Filter Output :
- final output after object based analysis opening post processing.
[fig:connected:sub:componentmodule]
Available variables for each expression can be found using item list variables names. available functions can be found in help windows by clicking on Help button. The module also prevents error in the mathematical command by checking the expression as the user types it. Background value is set to green if formula is right, in red otherwise. If mask expression is left blank entire image is processed. If Object Analysis expression is left blank the whole set of label objects is considered.
After segmentation step, too small objects can be rejected using Object min area input. Eliminating too small objects at this step is needed to lighten further computation. min area is the pixel size of the label object.
When a first pass have been done, Specific label object properties can be displayed. Select the “Filter Output” visualization mode, Update the visualization. Then use right click on selected object in image to display object properties.
Clicking on Save and Quit button export output to Monteverdi in vector data format.
A detailled presentation of this module, and examples can be found on the wiki .
A boat detection example is presented on Figure [fig:boat detection]. Results can be seen on Figure [fig:boat detection result].
[fig:boat detection]
[fig:boat detection result]
Under the term Feature Extraction, it include several techniques aiming to detect or extract infor- mations of low level of abstraction from images. These features can be objects : points, lines, etc.They can also be measures : moments, textures, etc.
For a given pixel, the Mean-shift algorithm will build a set of neighboring pixels within a given spatial radius and a color range. The spatial and color center of this set is then computed and the algorithm iterates with this new spatial and color center. The Mean-shift can be used for edge-preserving smoothing, or for clustering.
[fig:meanshift]
Supervised classification is a procedure in which individual items are placed into groups based on quantitative information on one or more characteristics inherent in the items and based on a training set of previously labeled items.
The supervised classification module is based on the Support Vector Machine method which consists in searching for the separating surface between 2 classes by the determination of the subset of training samples which best describes the boundary between the 2 classes. This method can be extended to be able to classify more than 2 classes.
The module allows to interactivelly describe learnings samples which corresponds to polygons samples on the input images.
Then a SVM model is derived from this learning sample which allows to classify each pixel of the input image in one of the defined class.
The non supervised classification module is based on the Kmeans algorithm. The GUI allows to modify parameters of the algorithm and produce a label image.
This section give access to specific treatments related to the SAR (Synthetic Aperture Radar) functionnalities.
SAR images are generally corrupted by speckle noise. To suppress speckle and improve the radar image interpretability lots of filtering techniques have been proposed. The module implements to well-known despeckle methods: Frost and Lee.
Compute the derived intensity and log-intensity from the input SAR imagery.
In conventional imaging radar the measurement is a scalar which is proportional to the received backscattered power at a particular combination of linear polarization (HH, HV, VH or VV). Polarimetry is the measurement and interpretation of the polarization of this measurement which allows to measure various optical properties of a material. In polarimetry the basic measurement is a 2x2 complex scattering matrix yielding an eight dimensional measurement space (Sinclair matrix). For reciprocal targets where HV=VH, this space is compressed to five dimensions: three amplitudes (|HH|, |HV|, and |VV|); and two phase measurements, (co-pol: HH-VV, and cross-pol: HH-HV). (see grss-ieee ).
Allow to construct an image that would be received from a polarimetric radar having selected transmit and receive polarizations. The Synthesis module waits for real and imaginary part (real images) of the HH, VV, VH and HV images. The reciprocal case where case VH=HV, is not properly handled yet, for now the user has to set the same input for the two HV and VH.
As we saw in the previous main section, the basic measurement is a 2x2 complex scattering matrix yielding an eight dimensional measurement space. But other measurements exist:
- covariance matrix and with its reciprocal specific case
- coherency matrix and with its reciprocal specific case
- circular coherency matrix and with its reciprocal specific case
- Mueller matrix and with its reciprocal specific case...
Modules in the Conversion subsection allow to proceed these conversions between matrix representations. Allowed conversion and input images types are described in the following figure [fig:sarpolconv].
[fig:sarpolconv]
This module allows to perform some of classical polarimetric analysis methods. It allows to compute:
- The polarimetric synthesis:
- input: 4 bands complex image
- output: mono channel real image
- parameters: the synthesis parameters (incident and reflected \psi and \chi angles)
- The reciprocal H alpha image:
- input: 6 bands complex image
- output: 3 bands real image