OverviewTeaching: 5 min
Exercises: 15 minQuestions
What more advanced features are there?Objectives
Create more advanced motion videos, using optical flow and video magnification.
Extract some statistical features.
The MGT contains several more advanced analysis features, some with limited documentation, and several that are still in development. We will here look at some of them.
Optical flow is a different motion estimation method, which not only looks at whether pixels change, but also in which directions they change. Thus a motion video created with optical flow, can also show motion vectors.
The downside to this, is that it is much more computationally heavy than plain frame differencing.
Eulerian Video Magnification is a method for amplifying parts of the image that are seemingly “invisible”. It requires quite some fine-tuning of the parameters to work properly. Try for example:
mg = mgvideomagnify('standstillcrop.avi','IIR',0.4,0.05,10,16,0.1);
mg = mgvideomagnify('standstillcrop','Butter',3.6,6.2,60,90,30,0.3)
Optical flow shows the direction of motion, but is much more CPU-consuming.
MGT is the starting point for a lot of more advanced analyses.