Structure from Motion workflows ============

A series of shell and python scripts for Structure from motion processing using the MicMac library.

The scripts

Clone/download/unzip this repo to wherever you wish
Add the script folders to your path e.g your .bashrc or .bash_profile

#micmac
export PATH=/my/path/micmac/bin:$PATH

#sfm scripts
export PATH=/my/path/Sfm:$PATH

export PATH=/my/path/Sfm/substages:$PATH

Make them executable

chmod +x Sfm/*.sh Sfm/*.py Sfm/substages/*.py Sfm/substages/*.sh

Update your paths

Dependencies

Sfm requires:

GNU/Linux or Mac OS
Python 3
MicMac
OSSIM

https://micmac.ensg.eu/index.php/Accueil

Dependency installation ~~~~~~~~~~~~~~~~~

MicMac

See MicMac install instructions here:

https://micmac.ensg.eu/index.php/Install

For ref only - I don't recommend using GPU-aided processing with MicMac as it appears to be incomplete. With reference to GPU supported compilation specifically, the following may help:

Replace the GpGpu.cmake file with the one supplied here as I have added the later Pascal 6.1 architecture
Make sure you install and use an older gcc compiler such as 5 or 6 for the cmake bit
Replace k with no of threads

cmake -DWITH_OPEN_MP=OFF
      -DCMAKE_C_COMPILER=/usr/bin/gcc-5
      -DCMAKE_CXX_COMPILER=/usr/bin/g++-5
      -DCUDA_ENABLED=1
      -DCUDA_SDK_ROOT_DIR=/path/to/NVIDIA_CUDA-9.2_Samples/common 
      -DCUDA_SAMPLE_DIR=/path/to/NVIDIA_CUDA-9.2_Samples 
      -DCUDA_CPP11THREAD_NOBOOSTTHREAD=ON ..

make install -j k

OSSIM

Install OSSIM via tha ubuntu GIS or equivalent repo

Ensure the OSSIM preferences file is on you path, otherwise it will not recognise different projections
see here https://trac.osgeo.org/ossim/wiki/ossimPreferenceFile

All in one scripts

These process the entire Sfm workflow

sfm.sh

Sub-stage scripts

These divide the workflow into Orientation, dense cloud/DSM processing and mosaic generation. All are internal to the complete workflows.

Orientation.sh

This performs feature detection, relative orientation, orienation with GNSS and sparse cloud generation
outputs the orientation results as .txt files and the sparse cloud

dense_cloud.sh

Processes dense cloud using either the PIMs or Malt-based algorithms, ortho-mosaic, point-cloud and georefs everything

orthomosaic.sh

Orthomosaic the output of any of the above including the batch scripts

Use

type -h to get help on each script e.g. :

sfm.sh -help

Thanks

Thanks to developers and contributors at MicMac and it's forum, particularly L.Girod whose work inspired the basis of the shell scripts

Ciaran1981/Sfm

Dependencies

Contents

All in one scripts

Sub-stage scripts

Use

Thanks