alexlocher / hpmvs

Progressive Prioritized Multi-view Stereo

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Hierarchical Progressive Multiview Stereo

Author: Alex Locher

Progressive multiview stero is an implementation of an algorihm taking a sparse 3D model (result of structure from motion software) and outputs a dense set of surface patches. Due to its progressiveness, the output gets more and more accurate the longer it runs.

Multiview Dense Stereo

Tsukuba Dataset

Progressive Output

Tsukuba Dataset

Related Publication

[1] Alex Locher, Michal Perdoch and Luc Van Gool. Progressive prioritized multi-view stereo. CVPR 2016.

[2] Y. Furukawa and J. Ponce, “Accurate, dense, and robust multiview stereopsis,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 32, no. 8, pp. 1362–1376, 2010.

Licence

HPMVS is realeased under the GPLv3 Licence.

If you use the algortihm in your academic work, please cite:

@article{locher16,
  title={Progressive Prioritized Multi-view Stereo},
  author={Locher, Alex and Perdoch, Michal and Van Gool, Luc},
  journal={CVPR},
  year={2016}
 }

Build and Install

We use the cmake framework for compilation and installation. The software is currently only tested in a Linux environment. First make sure that you install the necessary dependencies:

sudo apt-get install build-essential git cmake libeigen3-dev libjpeg-dev libgflags-dev libgoogle-glog-dev

then you should be able to get and compile hpmvs with:

git clone https://github.com/alexlocher/hpmvs
cd hpmvs
mkdir build && cd build
cmake ..
make -j4

Usage

The binary hpmvs takes an nvm-file as input and you have to specify an output directory:

./hpmvs --nvm=<nvm-file> --outdir=/tmp

About

Progressive Prioritized Multi-view Stereo

License:GNU General Public License v3.0


Languages

Language:C++ 61.8%Language:C 20.7%Language:Makefile 8.6%Language:Shell 6.3%Language:CMake 1.0%Language:Roff 0.5%Language:MATLAB 0.5%Language:M4 0.3%Language:Python 0.3%Language:Fortran 0.1%