About

This software delivers tools to build surface reconstructions from point cloud data and a simple viewer to display the results. Additionally, the found surfaces will be classified into predefined categories. The main aim of this project is to deliver fast and accurate surface extraction algorithms for robotic applications such as tele operation in unknown environments and localization.

Download and Compilation from Source

Step 0: Get the source code from our Github repository:

https://github.com/uos/lvr2

Linux (Ubuntu 18.04)

Step 1: Install all required package dependencies:

sudo apt-get install build-essential \
     cmake cmake-curses-gui libflann-dev \
     libgsl-dev libeigen3-dev libopenmpi-dev \
     openmpi-bin opencl-c-headers ocl-icd-opencl-dev \
     libvtk7-dev libvtk7-qt-dev libboost-all-dev \
     freeglut3-dev libhdf5-dev qtbase5-dev \
     qt5-default libqt5opengl5-dev liblz4-dev \
     libopencv-dev libyaml-cpp-dev

A C++17 compiler is required, e.g., g++7, gcc7 need bo installed. If CUDA is installed you also need g++6, see “Optional for NVIDIA graphics cards users”

Step 2: Configure and build from sources:

mkdir build
cd build
cmake .. && make

Optional for NVIDIA graphics cards users:

If you want to compile with CUDA support install the latest version of the CUDA toolkit, which you can find on NVIDIAs CUDA download site:

To enable CUDA support, you need to compile the software with a compatible GCC version. We have testet compilation with CUDA 9.1 and GCC 6. To use this compiler for compilation of CUDA generated code, set the CUDA_HOST_COMPILER option to g++-6 is forced to g++-6. Please ensure that this version is installed on your system. /

Step 3: Installation

After successful compilation, you will find the generated example tools in the ./bin/ directory. Optionally, you can install the library and header files to your system:

sudo make install

MacOS

Install the required libraries using Homebrew:

brew install boost boost-mpi cmake eigen flann gcc glew gsl hdf5 opencv lz4 qt vtk 

mkdir build
cd build
cmake .. && make

Usage

Your can experiment with the software using the provided dataset. For a simple reconstruction call in yout build directory:

bin/lvr2_reconstruct ../dat/scan.pts

in the root directory of the project. This will create a file called “triangle_mesh.ply” which can be displayed using the viewer application:

bin/lvr2_viewer triangle_mesh.ply

For more information, build the Doxygen documentation by calling

make doc

in the build directory.

Changelog for package lvr2

20.11.3 (2021-10-08)

• add libgdal-dev to package.xml, resolves #10

20.11.2 (2020-12-03)

• fix install dirs for lvr2 targets

20.11.1 (2020-11-11)

• fix for vtk8 issue
• add missing include and clean up using std
• use double for precision and refactor cl sor tool
• adds sor filter tool based on gpu knn
• added hdf5features example
• faster loading. map buffers instead of insert next cells
• added embree dep to cmake.in
• merged raycaster and cleanup
• working OpenCL Raycaster with better structure
• fix colors in filtering
• some performance optimizations
• migrated dmc approach

20.7.1 (2020-07-09)

• removed cl2.hpp, FeatureProjector.cpp from lib
• added meta information factory with slam6d and yaml support
• add debian folder from deb-lvr2 repo.
• add read vertex normals and colors from MeshBuffer, e.g., ply files
• add std arrays to conversion proxy for attribute map to channel conversion
• the large scale reconstruction chunk size is now dictated by the chunk manager
• make most things configurable for chunked mesh visualization
• hyperspectral meta information can be saved and loaded
• implemented basic load and save function of new IO features
• added function to save the tsdf-values in the chunk manager
• switched to new hdf5 io scheme for chunks and added a chunking pipeline with a multiple layer support
• integrated unoptimized hdf5-input for partial reconstruction
• fully integrated existing approach of dmc
• Contributors: Alexander Mock, Bao Tran, Benedikt Schumacher, Kevin Rüter, Lennart Niecksch, Malte kl. Piening, Marcel Wiegand, Raphael Marx, Sebastian Pütz, Thomas Wiemann, Timo Osterkamp, Wilko Müller

19.12.1 (2020-01-04)

• Initial release of lvr2
• Contributors: = Adrien Devresse, Aleksandr Ovcharenko, Alexander Altemöller, Alexander Löhr, Alexander Mock, Ali Can Demiralp, Ann-Katrin Haeuser, Bao Tran, Benedikt Schumacher, Christian Swan, Daniel Nachbaur, David J. Luitz, Denis Meyer, Devresse Adrien, Dmitri Bichko, Dominik Feldschnieders, Felix Igelbrink, Fernando Pereira, Florian Otte, Henning Deeken, Henning Strüber, Isaak Mitschke, Jan Philipp Vogtherr, Jan Toennemann, Jochen Sprickerhof, Johan M. von Behren, Juan Hernando Vieites, Kim Oliver Rinnewitz, Kristin Schmidt, Lars Kiesow, Lennart Niecksch, Lukas Kalbertodt, Malte kl. Piening, Marcel Mrozinski, Marcel Wiegand, Martin Günther, Matthias Greshake, Michael Görner, Michael V. DePalatis, Mike Gevaert, Nils Niemann, Pablo Toharia, Raphael Marx, Rasmus Diederichsen, Sabine Rast, Sebastian Pütz, Sergei Sobol, Simon Herkenhoff, Stefan Eilemann, Steffen Hinderink, Sven Albrecht, Thomas Wiemann, Tristan Igelbrink, Wilko Müller, Wolf Vollprecht