Here we publish a set of image segmentation libraries developed for the research in a Vision System Laboratory at IITP RAS. Currently this not very big collection includes the following:
- remseg — region adgacency graph (RAG) framework
- colorseg — color-based segmentation, details have been published in Linear colour segmentation revisited, SPIE, 2019.
Practically, the code is not platform specific but it is tested only on Linux. The official support is curently provided only for Linux too.
To configure and compile the project run the following command:
git clone https://github.com/Visillect/segmentation
mkdir build
cd build
cmake .. -DCMAKE_BUILD_TYPE=Release
make
For manipalutions with images the MinImg data structure is used provided in
minsubsystem libraries.
For compatility with OpenCV image types the vi_cvt package is provided.
The convertion from OpenCV cv::Mat to MinImg could be done as
#include <vi_oct/ocv.h>
cv::Mat cv_image = ...
...
MinImg min_image = vi::cvt::ocv::as_minimg(cv_image);
To convert MinImg to cv::Mat use
MinImage* min_image = ...
...
cv::Mat cv_image = vi::cvt::ocv::as_cv_mat(min_image);
Also, for MinImg smart pointers wrappers are provided in mximg package.
Please, use the GitHub issue tracker at https://github.com/Visillect/segmentation for bug reports, feature requests, etc.
Contributions (bug reports, bug fixes, improvements, etc.) are very welcome and should be submitted in the form of new issues and/or pull requests on GitHub.
This project (except remseg package) is licensed under the 2-Clause BSD License -- see doc/license.txt files in vi_packages and minsubsystem packages.
The remseg package is licensed under MIT license.