Progressive 3D Reconstruction All the Way

Author: Alex Locher

This is an implementation of the method presented in the paper 'Progressive 3D Reconstruction All the Way'. The code bridges the gap between progressive sparse 3D modelling and progressive dense 3D modelling by adapting an existing dense 3D model to an updated sparse scene. This allows to reuse the parts of the dense model which did not changed while still incorporate and account for all the changes in internal and external camera calibration.

Input and Output of the algorithm

Related Publication

[1] Alex Locher, Michal Havlena and Luc Van Gool. Progressive 3D Modeling All the Way. 3DV 2016. pdf

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

Licence

PATW is realeased under the GPLv3 Licence.

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

@article{locher163dv,
  title={Progressive 3D Modeling All the Way},
  author={Locher, Alex and Havlena, Michal and Van Gool, Luc},
  journal={3DV},
  year={2016}
 }

Disclaimer

The software is research code and should not be taken as an example for good coding style ;-)

Build and Install

The software can be build with the cmake framework and has some dependencies

If you don't have the dependencies installed yet, you can try to execute the install_deps.sh scritp included in the repository.

git clone https://github.com/alexlocher/patw
cd patw
git submodule init && git submodule update
# optinally install dependencies
./install_deps.sh
mkdir build && cd build
cmake ..
make -j4

Usage

The binary patw takes two nvm files nvm-file for the sparse models a ply file for the dense model before as input and the output can be specified:

./patw --nvm1=<nvm-file-before> --nvm2=<nvm-file-after> --patches=<dense-patches-before> --outdir=/tmp/patw

shidahe / patw