Part of my final year studies at the Department of Computing and Imperial College London.
This is in an investigation into particular motion stabilisation techniques and their efficiancy when applied to video-guided keyhole heart surgery situations. For much greater detail please see the paper to be presented late June.
This project was written in C++ and takes advantage of the following libraries.
- QT
- Matlab
- OpenCV
- CoinLP Solver
Thanks to Professor Daniel Rueckert: http://www.doc.ic.ac.uk/~dr/ Thanks to Matthias Grundmann http://www.cc.gatech.edu/cpl/projects/videostabilization/ (whose work inspired this undergraduate investigation)
This will not be readily usable in its current state. It has been configured for my Labs and for my personal computer. When the project is finished I will investigate tidying things up so that it can be easily installed elsewhere.
There is a BUG in OpenCV 2.3 whereby not all frames are read in. Use 2.4 and above. This project requires QT5.
Be aware of the video codecs when you use when using Motion. Ensure your input video plays on Matlab's Movie Player BEFORE running Motion if you want to do post-stabilisation evalution.
There is also a PERL script for batch processing which cycles through all possible parameters (defined in the script).