Stochastic Systems Group
Home Research Group Members Programs  
Demos Calendar Publications Mission Statement Alumni

SSG Seminar Abstract


Wyner-Ziv Video Compression


Dmitry Malioutov - SSG, MIT


Conventional video coders achieve high compression of video sequences by exploiting strong spatial and temporal correlation between pixels. To achieve state of the art performance, beyond that of still-image coders, it is necessary to take full advantage of the temporal correlation by using motion estimation. However, motion estimation is a computationally intensive task, restricting the encoders to be very complex. Motivated by the information-theoretic result of Wyner-Ziv on coding with side-information, distributed video compression attempts to shift motion estimation from the encoder to the decoder with none or minor degradation on the performance. The success of distributed video compression would allow for a new generation of light-weight video encoders.

In the talk we give an overview of prior work on distributed video compression, and present a new approach for distributed source coding using multi-level serially-concatenated-accumulate codes.



Problems with this site should be emailed to jonesb@mit.edu