Advanced Background subtraction

http://bit.do/dfQf7 Hidden discount at AgarwalInnoSoft.
State-of-the-art (as of 2011) background subtraction. Robust Principal Component Analysis (RPCA) is used to separate components of a video. On the left is the original video (200 frames taken from the "Subway" clip at http://perception.i2r.a-star.edu.sg/bk_model/bk_index.html ). Using the RPCA formulation with the TFOCS software (http://tfocs.stanford.edu), the frame on the left have been separated into a low-rank portion (middle) and sparse portion (right). The low-rank portion contains *predictable* elements of the video (the stationary background, and the periodic movement of the elevator), while the sparse portion contains the non-predictable elements (the people walking). The bottom row is similar to the top row except the RPCA formulation relaxed the constraints to take into account the quantization of the original video (since each pixel takes a value between 0 and 255). More information here: http://www.cs.caltech.edu/~srbecker/research.shtml