Merge branch 'develop' => v0.8.1

.travis.yml

@@ -6,7 +6,7 @@ language: cpp
 
 compiler:
   - gcc
-  #- clang: #Consider clang later, since cereal does not build fine with clang
+  #- clang: #Consider clang later, since cereal does not build fine on the clang CI version
   #  - "3.3"
 
 before_install:
@@ -33,7 +33,7 @@ before_script:
 
 script:
 # limit GCC builds to a reduced number of thread for the virtual machine
-  - if [ "$CC" = "gcc" ]; then make; else make -j 2; fi
+  - make -j 2
 # Perform unit tests only on GCC builds
   - if [ "$CC" = "gcc" ]; then make test; fi
 

AUTHORS

@@ -14,6 +14,7 @@ Michael Holroyd
 Romain Janvier
 Rory McCann
 Romuald Perrot
+rhiestan
 sergi pujades-rocamora
 sflotron
 vincentweb

BUILD

@@ -2,9 +2,9 @@
 OpenMVG (open Multiple View Geometry)
 =====================================
 
------------------
-Build instruction
------------------
+------------------
+Build instructions
+------------------
 
 Required tools:
 * Cmake 
@@ -31,12 +31,21 @@ OpenMVG_BUILD_TESTS (ON/OFF(default))=> Build openMVG unit tests
 OpenMVG_BUILD_EXAMPLES (ON/OFF(default))=> Build OpenMVG example applications.
   Does not affect binaries under 'software'
 
+--------------------------
+General informations
+for openMVG SfM pipelines
+--------------------------
+OpenMVG can export graphs as graphviz .dot files and render them as SVG files.
+If you want consider this graph visualization feature, please consider to install Graphviz.
+
 -----------------
 Linux compilation
 -----------------
 
 Setup the required external library.
 * sudo apt-get install libpng-dev libjpeg-dev libtiff-dev libxxf86vm1 libxxf86vm-dev libxi-dev libxrandr-dev
+If you want see the view graph svg logs
+* sudo apt-get install graphviz
 
  $ git clone --recursive https://github.com/openMVG/openMVG.git
  $ cd openMVG
@@ -92,6 +101,7 @@ Compile the libraries and binaries samples.
 -------------------
 Mac compilation
 -------------------
+
  $ git clone --recursive https://github.com/openMVG/openMVG.git
  $ cd openMVG
  $ ls
@@ -105,22 +115,34 @@ If you want enable unit tests and examples to the build:
  $ xcodebuild -configuration Release
 
 
- --------------------
- Using openCV sample
- --------------------
+--------------------
+Using openCV sample
+--------------------
+
  Add -DOpenMVG_USE_OPENCV=ON to your cmake command line and set OpenCV_DIR variable to your openCV build directory
 => i.e.: -DOpenCV_DIR="/home/user/Dev/github/itseez/opencv_Build" -DOpenMVG_USE_OPENCV=ON
 
-------------------------------------
-Using as library dependency in cmake
-------------------------------------
-Adding following lines to your CMakeLists.txt should provide OpenMVG usable as
-static library:
+------------------------------------------------------------
+Using OpenMVG as a third party library dependency in cmake
+-------------------------------------------------------------
+
+OpenMVG can be used as a third party once it have been installed.
+Because it can use it's own ceres version, it's better to install it locally and not in system files.
+So please consider using the CMAKE_INSTALL_PREFIX cmake variable to specify a local installation directory.
+
+Here the syntax to add the variable to the cmake command line:
+-DCMAKE_INSTALL_PREFIX:STRING="./openMVG_install"
+
+Perform "make" and "make install"
+
+Once the library has been installed, go to your project that want use OpenMVG as an external library and add:
 
- add_subdirectory(openMVG/src)
- include_directories(${OpenMVG_INCLUDES})
- target_link_libraries(target ${OpenMVG_LIBS})
+FIND_PACKAGE(OpenMVG REQUIRED)
+INCLUDE_DIRECTORIES(${OPENMVG_INCLUDE_DIRS})
+ADD_EXECUTABLE(main main.cpp)
+TARGET_LINK_LIBRARIES(main ${OPENMVG_LIBRARIES})
 
-Information about required dependencies, standalone build and platform 
-specificity can be found below.
+Specify to CMAKE where OpenMVG have been installed by using the cmake OpenMVG_DIR variable that must point to:
+-DOpenMVG_DIR:STRING="YourInstallPath"/share/openMVG/cmake
 
+A message will be displayed if OpenMVG is found or not at the cmake configure step.

README.md

@@ -78,5 +78,16 @@ publications are relevant:
 [5] Moisan Lionel, Moulon Pierre and Monasse Pascal. IPOL 2012.
 [Automatic Homographic Registration of a Pair of Images, with A Contrario Elimination of Outliers.](http://dx.doi.org/10.5201/ipol.2012.mmm-oh)
 
-Please cite [3] if you use openMVG library in your work.
+[6] Moulon Pierre, Monasse Pascal and Marlet Renaud. ICCV 2013.
+[Global Fusion of Relative Motions for Robust, Accurate and Scalable Structure from Motion.](http://imagine.enpc.fr/~moulonp/publis/iccv2013/index.html)
+
+or cite it as:
+
+```
+  @misc{openMVG,
+    author = "Pierre Moulon and Pascal Monasse and Renaud Marlet and Others",
+     title = "OpenMVG",
+    howpublished = "\url{https://github.com/openMVG/openMVG}",
+  }
+```
 

docs/sphinx/rst/bibliography.rst

@@ -60,6 +60,9 @@ Bibliography
 .. [CMPMVS] **Multi-View Reconstruction Preserving Weakly-Supported Surfaces.**
     M. Jancosek, T. Pajdla, CVPR 2011.
 
+.. [Waechter2014] **Let There Be Color! Large-Scale Texturing of 3D Reconstructions.**
+    M. Waechter, N. Moehrle, and M. Goesele, ECCV 2014.
+
 .. [FLANN] **Fast Approximate Nearest Neighbors with Automatic Algorithm Configuration.**
     Muja, Marius, and David G. Lowe.  VISAPP (1). 2009.
  

docs/sphinx/rst/openMVG/cameras/cameras.rst

@@ -79,22 +79,6 @@ A 3D point is projected in a image with the following formula (homogeneous coord
 OpenMVG Pinhole camera models
 ------------------------------
 
-* Simple pinhole camera models (intrinsic + extrinsic(pose))
-* Pinhole camera models with radial distortion (intrinsic + extrinsic(pose))
-
-.. code-block:: c++ 
-
-  // Setup a simple pinhole camera at origin
-  // Pinhole camera P = K[R|t], t = -RC
-  Mat3 K;
-  K << 1000, 0, 500,
-       0, 1000, 500,
-       0, 0, 1;
-  PinholeCamera cam(K, Mat3::Identity(), Vec3::Zero());
-
-  // Radial distortion (brown model -> polynomial distortion (K1 to k3))
-  BrownPinholeCamera(K, Mat3::Identity(), Vec3::Zero(), k1, k2 , k3);
-
 * Pinhole intrinsic
 
   * :class:`Pinhole_Intrinsic : public IntrinsicBase` 
@@ -111,3 +95,14 @@ OpenMVG Pinhole camera models
     * classic pinhole camera (Focal + principal point and image size) + radial distortion by three factors.
     * can add and remove distortion
 
+* Simple pinhole camera models (intrinsic + extrinsic(pose))
+
+.. code-block:: c++ 
+
+  // Setup a simple pinhole camera at origin
+  // Pinhole camera P = K[R|t], t = -RC
+  Mat3 K;
+  K << 1000, 0, 500,
+       0, 1000, 500,
+       0, 0, 1;
+  PinholeCamera cam(K, Mat3::Identity(), Vec3::Zero());

docs/sphinx/rst/openMVG/openMVG.rst

@@ -16,7 +16,6 @@ openMVG provide a collection of tiny libraries that allow to solve computer visi
   robust_estimation/robust_estimation.rst
   matching/matching.rst
   tracks/tracks.rst
-  bundle_adjustment/bundle_adjustment.rst
   geometry/geometry.rst
 
 
@@ -34,7 +33,7 @@ Todo:
 
 * **graph** - Manipulation of graphs (connected CC, triplet listing...)
 
-* **exif_IO** - Collection of tools to extract and use image EXIF data.
+* **exif** - Collection of tools to extract and use image EXIF data.
 
 * **matching_image_collection** - Functions to perform local matching and geometric filtering of putative correspondences in image collections.
 

docs/sphinx/rst/openMVG/sfm/sfm.rst

@@ -110,9 +110,9 @@ Non linear refinement, Bundle Adjustment
 
 OpenMVG provides a generic bundle_adjustment framework to refine or keep as constant the following parameters:
 
-* internal parameters,
-* external parameters,
-* 3D structure.
+* internal orientation parameters (intrinsics: camera projection model),
+* external orientation parameters (extrinsics: camera poses),
+* structure (3D points).
 
 .. code-block:: c++
 
@@ -128,6 +128,21 @@ OpenMVG provides a generic bundle_adjustment framework to refine or keep as cons
 
   const double dResidual_after = RMSE(sfm_data);
 
+Bundle Adjustment (ajustement de faisceaux), is a non linear optimization problem.
+It looks to minimizing the residual error of a series of user cost functions (the reprojection errors of the structure :math:`X_j` to the images measures :math:`x_j^i`).
+According:
+
+* :math:`X_j` the Jnth 3D point of the structure of the scene,
+* :math:`x_j^i` the observation of the projection of the 3D point :math:`X_j` in the image :math:`i`,
+* :math:`P_i` the projection matrix of the image :math:`i`
+
+From a user provided initial guess the vector of parameters: :math:`\{X_j,P_i\}_{i,j}`: camera parameters :math:`\{P_i\}_i` and the scene structure :math:`\{X_j\}_j` are refined in order to minimizes the residual reprojection cost:
+
+.. math::
+  \underset{ \{P_i\}_i, \{X_j\}_j}{minimize} \left\| \sum_{j=0}^{m} \sum_{i=0}^{n} x_j^i - P_i X_j \right\|_2
+
+OpenMVG proposes options in order to tell if a parameter group must be kept as constant or refined during the minimization.
+
 SfM Pipelines
 ==============
 

docs/sphinx/rst/software/SfM/ExportUndistortedImages.rst

@@ -0,0 +1,42 @@
+
+********************************************
+openMVG_main_ExportUndistortedImages
+********************************************
+
+This application export undistorted images from known camera parameter intrinsic.
+
+Algorithm of the application
+
+.. code-block:: c++
+
+  Require: internal + camera calibration
+  Require: images
+  Ensure: undistorted images
+  for each view
+      if view has valid intrinsic
+        undistort and save the undistorted view
+
+Information and usage
+========================
+
+The chain is designed to run on a sfm_data.json file.
+The sfm_data file should contains:
+- valid view with some defined intrinsics,
+
+  .. code-block:: c++
+
+    $ openMVG_main_ExportUndistortedImages -i Dataset/out_Reconstruction/sfm_data.json -o Dataset/out_Reconstruction/undistortedImages
+
+Arguments description:
+
+**Required parameters:**
+
+  - **[-i|--input_file]**
+
+    - a SfM_Data file with valid intrinsics and poses and optional structure
+
+  - **[-o|--outdir]**
+
+    - path where the undistorted images will be stored
+
+

docs/sphinx/rst/software/SfM/MVS.rst

@@ -55,6 +55,11 @@ Export to MVE (Multi-View Environment)
 
 You will need to compile MVE tools and `FSSR <http://www.gris.informatik.tu-darmstadt.de/projects/floating-scale-surface-reco/>`_.
 
+Export to MVS Texturing
+=======================
+
+If you don't want to use the full MVE pipeline but only `MVS Texturing <http://www.gris.tu-darmstadt.de/projects/mvs-texturing/>`_ [Waechter2014]_ to project a set of oriented images on a mesh, one solution is to use the openMVG_main_openMVG2MVSTEXTURING binary. This binary converts your SfM_Data file into one format used by MVS Texturing. In addition, you may need to undistort your images with openMVG_main_ExportUndistortedImages as it's not handled by the openMVG_main_openMVG2MVSTEXTURING tool.
+
 Export to CMPMVS
 ========================
 

docs/sphinx/rst/software/SfM/SfM.rst

@@ -82,6 +82,7 @@ OpenMVG SfM pipelines run as a 4 step process:
 
    ./ComputeSfM_DataColor.rst
    ./ComputeStructureFromKnownPoses.rst
+   ./ExportUndistortedImages.rst
 
 
 5. Optional further processing (3rd party)
@@ -131,6 +132,7 @@ To know more about each tool visit the following link and read the doc below:
    ./GlobalSfM.rst
    ./ComputeSfM_DataColor.rst
    ./ComputeStructureFromKnownPoses.rst
+   ./ExportUndistortedImages.rst
 
 .. toctree::
    :maxdepth: 1