CRWCR

A Continuous Random Walker Model with Explicit Coherence Regularization for Image Segmentation (TIP 2019).

---

This is a tiny software which implements our CRWCR algorithm.

---

Performance

With 10 1D initializations and 10 2D PR iterations (enough for most applications), the running time on my laptop ( with Intel i7-6700HQ CPU and GTX965M GPU) are shown in the table below.

Image Size	481X321	680X669	1024X1024	1443X963	1924X1284
Time (ms)	32	64	123	190	327

---

Building requirement

• The project can be compiled with Visual Studio (>= 2017.15.8) and gcc on Ubuntu 16.
• The building through CMake requires a version number >= 3.8.
• The primary GUI is based on Qt (Supported >= 5.10).
• GPU is supported, please check the USE_CUDA option if you have a GPU device. The GPU version is based on CUDA (Supported >= 9.0).
• The image rendering is based on OpenGL (>= 4.0).

Citing CRWCR:

If you use our code in your research, please cite with:

@article{li2019continuous,
  title={A Continuous Random Walk Model With Explicit Coherence Regularization for Image Segmentation},
  author={Li, Mengfei and Gao, Hong and Zuo, Feifei and Li, Hongwei},
  journal={IEEE Transactions on Image Processing},
  volume={28},
  number={4},
  pages={1759--1772},
  year={2019},
  publisher={IEEE}
}

Other

The code has been tested under Windows 10 with Visual Studio 2017 and Ubuntu 16 with gcc. Any feedback is welcome.