Structure Mediated Use of Local-Global Duality in Image and Video Processing

An image can have multiple representations, each of which has its own advantages and disadvantages from the perspective of image processing operations. This thesis investigates how to combine multiple representations with efficient optimization techniques so as to gather together their combined strengths in improving both objective and perceptual quality of the processed image. In many of the applications that we consider, global smoothness properties of the image are captured by linear transform-based representations while local edge-based properties are captured with a segmentation algorithm. Applications used to justify the efficacy of such an approach include lossless image compression, lossy video compression, denoising AWGN corrupted images and image magnification. It is found that many of the traditional unirepresentational processing techniques are special cases of the general multirepresentational formulations that are presented in this thesis.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Electrical Engineering

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2015

Author dc:creator

• Ratakonda, Krishna Chaitanya

Contributors dc:contributor

• Ahuja, Narendra

• Engineering, Electronics and Electrical

Language dc:language

eng

Identifier

(MiAaPQ)AAI9944980

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/81297