A Python source code implementation of motion magnification based on the paper: Phase Based Video Motion Processing by Neal Wadhwa, Michael Rubinstein, Frédo Durand, William T. Freeman, ACM Transactions on Graphics, Volume 32, Number 4 (Proc. SIGGRAPH), 2013. project.
Note: this follow up code can also handle large motion https://acceleration-magnification.github.io/
- python 2.7
- numpy
- perceptual (Complex steerable pyramid, install with: sudo pip install perceptual)
phasebasedMoMag.py # Main file
pyramid2arr.py # Help class to convert a pyramid to a 1d array
media/guitar.mp4 # Example video
./media/guitar.mp4
When you run the code 'python phasebasedMoMag.py' it expects an example video in the 'media' folder. Here we use the http://people.csail.mit.edu/mrub/evm/video/guitar.mp4 video from the motion magnification website.
The pbMoMA implementation is based only on the paper. It was developed independent of the source code that can be requested from the paper authors (this pyton code was written without having access to that code). Therefore, the results from the pbMoMA code may differ from the results by the paper authors. Differences include: using a sliding window, only an Ideal filter, no sub-octave pyramid, and no color.
The code was implemented during the Lorentz Center workshop ICT with Industry: motion microscope. Participants: Joao Bastos, Elsbeth van Dam, Coert van Gemeren, Jan van Gemert, Amogh Gudi, Julian Kooij, Malte Lorbach, Claudio Martella, Ronald Poppe.