Correlation Flow
Steps to run the demo-
- Clone the repository
cd catkin_ws/src
git clone https://github.com/akathpal/correlation_flow
- Install Dependencies as mentioned.
- Compile the ROS workspace using catkin_make
- Give permissions to py files.
cd catkin_ws/src/correlation_flow/script
chmod u+x imagepub_video.py
chmod u+x imagepub_2frames.py
- Run correlation flow launch file
roslaunch correlation_flow correlation_flow.launch
- To run the video from a camera,
rosrun correlation_flow imagepub_video.py
- To view the optical flow output using 2 frames,
rosrun correlation_flow imagepub_2frames.py
Correlation Flow: Robust Optical Flow using Kernel Cross-Correlators
Velocity Estimation in 3-D space $v_x, v_y, v_z, \omega_z$
This repo contains source codes for the following paper, which is accepted by ICRA-18:
Chen Wang *, Tete Ji *, Thien-Minh Nguyen, and Lihua Xie, "Correlation Flow: Robust Optical Flow Using Kernel Cross-Correlators", IEEE International Conference on Robotics and Automation (ICRA), 2018.
Platform
Codes have been tested on Ubuntu 16.04 with ROS kinetic.
Install Dependencies:
- Install FFT library:
sudo apt-get install libfftw3-dev libfftw3-doc
Install Intel Math Kernel Library (MKL) (optional):
- Download MKL from Intel website
- Extract downloaded file
tar -zxvf [name_of_downloaded_file]
- Go to extracted folder, give permission:
sudo chmod +x install.sh
- Run installation
./install.sh
- Link library, add to .bashrc:
source /opt/intel/bin/compilervars.sh intel64
- Try compile in ROS workspace
If this repo is useful for your projects, you may cite it as:
@inproceedings{wang2018correlation,
title={{Correlation Flow: Robust Optical Flow using Kernel Cross-Correlators}},
author={Wang, Chen and Ji, Tete and Nguyen, Thien-Minh and Xie, Lihua},
booktitle={International Conference on Robotics and Automation (ICRA)},
year={2018},
organization={IEEE}
}
Correlation Flow is built on following work:
Chen Wang, Le Zhang, Lihua Xie, Junsong Yuan, Kernel Cross-Correlator, In AAAI Conference on Artificial Intelligence (AAAI-18), 2018 (PDF available here) (source codes available here)
Works using Correlation Flow
Thien-Minh Nguyen, Abdul Hanif Zaini, Chen Wang , Kexin Guo, and Lihua Xie, "Robust Target-relative Localization with Ultra-Wideband Ranging and Communication", IEEE International Conference on Robotics and Automation (ICRA 2018), 2018. (Video available here)
*The above work applys correlation flow to improve the performance of localization accuracy.