Idea and original paper
The goal is to make changes in videos visible that cannot be seen with the naked eye. Videos contain information that are invisible to the human eye. “Eulerian Video Magnification” [1] image processing to magnify them enough to become visible.
A Laplacian or Gaussian pyramid is computed, relevant frequency bands (chosen by the user) are extracted and the extracted signal is magnified by factor α. The result is added to the original signal. How well the method works depends on the correct choice of parameters. The exact values depend on the material and on the desired effect and it may be necessary to try various values.
Implementation
Like the authors of the paper, we implemented Eulerian Video Magnification in MATLAB. A compiled version of the application can be found under Download Program. When starting the program via the sourcecode, Main.m is run, which opens the GUI. The sourcecode and its documentation are located in Documentation.
How to use the program
On starting the program, an empty GUI is loaded. All options including loading, saving and choosing parameters are located in the top section of the interface.
LOAD opens a file chooser where the original video can be selected. SAVE exports the magnified video. The magnifying factor determines how strong the magnification of the selected bandwidth is. Lower and upper frequency determine the range of the filter. Pyramid type allows the user to choose whether to use a Gaussian or a Laplace pyramid. Attenuation can be used for uneven magnification. If a value other than 0 is chosen, the alpha value (magnifying factor) is attenuated for higher pyramid levels. The higher the attenuation value, the more the magnifying factor is attenuated.
File formats that can be loaded are .mp4, .avi, .mpg’, .wmv, .mov and .mj2. Videos can be saved as .mp4, .avi or .mj2. It is advisable to use only the necessary video resolution and length as this is the main factor governing the speed of calculations.
Reference
[1] Wu, H. Y., Rubinstein, M., Shih, E., Guttag, J. V., Durand, F., & Freeman, W. T. (2012). Eulerian video magnification for revealing subtle changes in the world.