πŸ“· Threaded depth-map cleaning and inpainting using OpenCV
Switch branches/tags
Nothing to show
Clone or download
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
docs Init Jun 24, 2018
ext Init Jun 24, 2018
src Adding comments Jun 24, 2018
.gitignore Init Jun 24, 2018
.gitmodules Init Jun 24, 2018
CMakeLists.txt Init Jun 24, 2018
LICENSE Create LINCENSE Jun 24, 2018
README.md Adding a video example Jun 28, 2018

README.md

Threaded Depth Cleaner

Threaded depth-map cleaning and inpainting using OpenCV.

  1. Qucikstart
  2. Depth Cleaning
  3. Video
  4. Dependencies
  5. License

Depth-map image Left-to-right: Original depth-map visualized using a color-map, filtered depth-map visualized using a color-map

Quickstart

The tool is setup to use the Intel's librealsense SDK and a RealSense depth sensor (415/435).

  1. Clone the repository with it's submodules:
git clone https://github.com/juniorxsound/ThreadedDepthCleaner --recursive
  1. Install OpenCV (Windows, macOS, Linux)
  2. Install CMake (Instructions)
  3. cd into the repository folder and run:
mkdir build && cd build && cmake ../ && make -j4
  1. Run it by running ./DepthCleaner

Tested on macOS (10.13.2) using CMake (3.10.0)

Depth Cleaning

  1. Use the SCALE_FACTOR macro to determine how much to scale the depth-map for the inpainting (1 means no scaling and less performent, 0.2 means 1/5 of the original depth-map is used for inpainting and more performent).

Dependencies

  1. librealsense
  2. OpenCV

Depth coloring and cv::Mat queuing functionality was implemented thanks to @PKLab and @Catree πŸ™ŒπŸ»

Thanks to @Dodiku for help with the cover image πŸ’ͺ🏻

License

All code in this repository is released under the MIT license found here