Today, we are looking at a dramatic scene. In industrial settings, there are numerous situations where the venting of tanks is essential, in particular because of temperature differential and the state of the liquid inside. Here is a dramatic example:
Here is what happens for the most dramatic moment of the video through a Robust PCA implementation (GoDec). The video is decomposed in three components, a low rank one (ideally a background image in the case of rank 1), a sparse and noisy component.
Let us note that the low rank component was some sort of average of the two tank configurations. It also stayed longer in a full configuration when the real tank had already imploded. Let us also note the contour of the tank drawn early in the process in the noise component (are we seeing some sort of tiny vibrations?) Had we stopped the tape earlier, the rank of the video could have stayed at one (leaving only one static picture in the low rank component) and have most of the action in the sparse and noisy component of the scene,. The sparse component seems to show the most lit section of the scene.
Other Advanced Matrix Factorization Algorithm can be found in the Matrix Factorization Jungle page.
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.