By: Jonathan Farache
If you’ve ever been let down by the idea human vision is limited to what can be seen by the “naked eye”, you’re in for a jolt. A technique by the name of “Eulerian Video Magnification” allows us to see beyond the confines of our own sight. The idea is pretty straightforward: a video or a picture is processed by honing in on its composing pixels. Each pixel (either in a still image or through a video) goes through microscopic changes over time. After processing the image, each of these variations is amplified by a factor of up to a 100, resulting in a much more vivid transformation. As an example, and so far the most crucial application of this technology, take a new-born baby. Anyone who has encountered these young infants knows that sometimes, when they just lie there, it’s hard to tell whether they are breathing. Eularian Video Magnification can solve this in a couple of different ways: first, it could amplify the change in skin color to exaggerate the variation of passing blood cells, thereby effectively measuring the pulse. On the other hand, it could literally amplify the height the baby’s chest rises when he/she breaths to make the respiration obvious-beyond-doubt.
This technology is a classical manifestation of computer vision. Most of us can see just fine, but sometimes that doesn’t quite cut it. Computer vision can be as simple as using human vision to look at several things simultaneously to make a precise analysis, and as complicated as transcending the limitations of our sight. Since Eularian Video Magnification can be done in real-time, we may well be looking at a technologically-supplemented super-power in the near future. We’ve all heard of Google glass – what if they incorporated this sort of vision-enhancement into their spectacles? We could essentially walk around and see an entirely different world. Superman’s X-ray vision may be impressive, but does that compare to a firefighter in the field being able to see the pulse of someone that’s trapped under a pile of debris? The implications are huge.
Consider an explosion in a battlefield. If the blast or shrapnel is somehow avoided, the actual blast-wave can still be lethal. Detecting internal bleeding (the likely cause of a blast-wave) without sophisticated assistance can be difficult. If, alternatively, a medic could take a better look at the blood-flow or blood-concentration of the soldier in real time, he would be in a better place to make a correct assessment, and thus a more useful decision.
The ability to pick up on minute movements and alterations in the real world means a general rise in precise measurements of tangible things. Apply this freely to shuttle-launches, lie-detection, respiration analysis, and any other number of activities and you’ve got a fairer, more accurate view of the world.
Jonathan Farache is a contributing writer for Computer Vision Online.
Hao-Yu Wu, Michael Rubinstein, Eugene Shih, John Guttag, Frédo Durand, and William Freeman. 2012. Eulerian video magnification for revealing subtle changes in the world. ACM Trans. Graph. 31, 4, Article 65 (July 2012), 8 pages. [PDF]
MIT: Eulerian Video Magnification for Revealing Subtle Changes in the World (link)
NYTimes: Scientists Uncover Invisible Motion in Video (link)
Image credit: MIT/Michael (Miki) Rubinstein