10/15/2003 I will be presenting my ongoing work on super-resolving video sequences through hallucination

Goksel Dedeoglu 15 Oct

I will be presenting my ongoing work on super-resolving video sequences through hallucination. After a review of Baker and Kanade’s face hallucination, I will propose a graphical model that can justify the energy function minimized by this algorithm. Subsequently, I will extend this framework into time, and show the regularizing role of temporal models in obtaining 16-fold super-resolved videos of a 6×8 pixel-wide talking face.