Foveation for Low Bitrate Video Coding

Mr. Hamid R. Sheikh

Dept. of Electrical and Computer Engineering
The University of Texas at Austin

Friday, February 16th, 3:00 PM, ENS 302

sheikh@ece.utexas.edu

Abstract

Data compression algorithms rely on modeling of signal source as well as the receiver for transmitting the source information with reduced number of bits. Lossy compression schemes typically use a receiver model to discard any information that is least important to the receiver. Our knowledge about the abilities and limitations of the Human Visual System (HVS), for example, would enable us to design lossy image and video compression schemes that introduce distortion in the signal in such as way that they are least perceptible to the human eye while retaining the fidelity of those aspects of the image to which the eye is more sensitive. Quantization matrices in MPEG, for example, make use of the fact that the sensitivity of the human visual system (HVS) is different for different spatial frequencies. "Foveation" is another layer of HVS modeling that allows us to model the inability of the HVS to observe the entire image that is being viewed at full resolution because of non-uniform spacing of sensor neurons on the retina. Since the HVS cannot perceive the full resolution information that is present in a video sequence, we can hope to remove the information that cannot be perceived by the HVS and obtain an increase in compression gain without sacrificing perceived quality.

This talk will introduce the concept of foveation for low bitrate video coding exploring the trade-offs, the costs involved as well as look at computational feasibility for deployment.

Biography

Hamid R. Sheikh graduated with his B.S.E.E in 1998 from the University of Engineering and Technology, Lahore, Pakistan. He joined the University of Texas at Austin in Fall 1999, for his M.S.E.E degree. He is currently employed as a Graduate Research Assistant, under Prof. Alan C. Bovik, at the Laboratory for Vision Systems which is part of the Laboratory for Image and Video Engineering. He has been involved in several Digital Signal and Image Processing projects.

A list of Telecommunications and Signal Processing Seminars is available at from the ECE department Web pages under "Seminars". The Web address for the Telecommunications and Signal Processing Seminars is http://anchovy.ece.utexas.edu/seminars