Home

Awesome

Introduction

This code is adapted from PlaneSweepLib in order to reconstruct with satellite images. It can also work on ground-level images.

Project page: https://kai-46.github.io/VisSat/

Compilation

The following libraries are required:

NVIDIA CUDA </br> Boost (system filesystem program_options) </br> OpenCV </br> Eigen3 </br>

The script 'install_opencv.sh' aims to assist installing opencv on your computer.

Compile the program via:

mkdir build && cd build && cmake .. && make

General Usage

SatellitePlaneSweep/build/bin/satellitePlaneSweep \
    --imageFolder {} \
    --imageList {} \
    --refImgId {} \
    --outputFolder {} \
    --matchingCost {} \
    --windowRadius {} \
    --nX {} --nY {} --nZ {} \
    --firstD {} --lastD {} --numPlanes {} \
    --filterCostVolume {} --guidedFilterRadius {} --guidedFilterEps {} \
    --saveCostVolume {} \
    --debug {} \
    --saveBest {} --filter {} --filterThres {} \
    --saveXYZMap {} \
    --savePointCloud {}

Brief explanation of the options

Interfacing with Python

scripts/binary_grid_io.py can be used to read the output files of the C++ program as numpy array.

License

This software uses the GPLv3 license.