Date of Award:

5-2016

Document Type:

Dissertation

Degree Name:

Doctor of Philosophy (PhD)

Department:

Computer Science

Committee Chair(s)

Stephen W. Clyde

Committee

Stephen W. Clyde

Committee

Nicholas Flann

Committee

Curtis Dyreson

Committee

Amanda Lee Hughes

Committee

Bedri Cetiner

Abstract

The ultimate goal of any video delivery system is to allow any user to watch any video of any kind on any display device over any type of network with a desired output, i.e., viewing codec with preferred quality of service. This could theoretically require 1032 video transcoding functions that convert any original video to any desired output. Guaranteeing a required format and preferred quality of service of the perceived video requires selecting or composing a set of transcoding functions that satisfy the requested format and preferred quality of service. An effective way to accomplish this is by allowing the selection and composition mechanisms to take place based on a model that accurately assesses the impact of each transcoding function on video quality. Using such a model, each user will receive the requested video based on the required format and preferred quality of service.

The main contribution of this research is the development and substantiation of such a model, called Video Transcoding Objective-quality Model (VTOM), that provides an extensible video transcoding service selection mechanism, which takes into account both the format and characteristics of the original video and the desired output. VTOM represents a mathematical function that uses a set of media-related parameters for the original video and desired output, including codec, bit rate, frame rate, and frame size to predict the quality of the transcoded video generated from a specific transcoding function.

VTOM includes four quality sub-models, each describing the impact of one of these parameters on objective video quality, as well as a weighted product aggregation function that combines these quality submodels with four additional error sub-models in a single tool for assessing the overall video quality. I compared the predicted results from the VTOM with quality values generated from an objective video quality metric. These extensive comparisons yielded results that showed good correlations, with low error values.

Because no suitable metrics exist in the literature that evaluate the relative quality of two videos that have different frame rates, this research presents and develops such a metric, called Frame Rate Metric (FRM). FRM uses any frame-based objective quality metric to compare two videos. This research also presents a strategy that helps in evaluating the relative quality of two videos that have different frame sizes.

This research also presents and adapts four QoS-aware video transcoding service selection algorithms. Each of them selects the "best-fit" video transcoding service from a pool of available ones. This selection satisfies the requested format and desired quality of service. The evaluation results showed the effectiveness and the efficiency of these candidate algorithms.

As a consequence from this dissertation, the researchers and developers of video delivery systems and applications have a model to calculate the degradation that each transcoding function can cause rather than statistically evaluate it. Current statistical methods consider only the desired quality of service. This can lead us to conclude that VTOM can improve video transcoding selection and composition algorithms.

Checksum

6c1ea692db5313c536d9cada4118e419

Download

Share

COinS

Copyright for this work is retained by the student. If you have any questions regarding the inclusion of this work in the Digital Commons, please email us at .