icra2018/logos repository overview

⁠LOGOS

LOGOS is a method of outlier removal specifically designed for image features such as SURF.

Details on LOGOS can be found in the paper LOGOS: Local geometric support for high-outlier spatial verification⁠.

⁠Dependencies

The demo can be run with images from the Oxford City Centre dataset⁠. This dataset was generated by M. Cummins and P. Newman for the paper FAB-MAP: Probabilistic Localization and Mapping in the Space of Appearance⁠ (2008).

If available, the VLFeat library will be used via the MATLAB API. This is only used for the demo, not for LOGOS itself. The VLFeat library can be downloaded from the VLFeat homepage⁠.

The demo also depends on methods for detecting and extracting SURF features from the MATLAB Computer Vision System Toolbox.

⁠Running the demo

Run

demo

in MATLAB for a usage example.

This example uses images 1448.jpg and 388.jpg from the Oxford City Centre dataset⁠.

⁠Reference

If you use LOGOS in an academic work, please cite:

@INPROCEEDINGS{LOGOS:LA:ICRA2018, 
author={S. Lowry and H. Andreasson}, 
booktitle={2018 IEEE International Conference on Robotics and Automation (ICRA)}, 
title={LOGOS: Local geometric support for high-outlier spatial verification}, 
year={2018}, 
volume={}, 
number={}, 
pages={}, 
doi={}, 
ISSN={}, 
month={May},}
}

⁠How to Run with Docker

⁠Linux

⁠Prerequisites

• MATLAB installed on local host.

Tested on Ubuntu 16.04.6 with Docker 18.06.1-ce, MATLAB R2017a.

1. Open a terminal and run the command (replacing the MATLAB folder and MAC address with your own):

docker run --rm -p 8888:8888 -v /usr/local/MATLAB/R2017a:/usr/local/MATLAB/R2017a \
    -v /usr/local/lib/python3.5/dist-packages/matlab:/usr/local/lib/python3.5/dist-packages/matlab \
    --mac-address=2c:60:0c:d6:50:36 icra2018/logos:latest

1. Run a web browser and open the link: http://localhost:8888/lab/tree/README.ipynb⁠