Toronto Metropolitan University
Browse
- No file added yet -

Sparse Signal Decomposition Techniques For Multimedia Fingerprinting

Download (10.47 MB)
thesis
posted on 2021-05-22, 09:27 authored by Xiaoli Li
This dissertation focuses on digital multimedia content protection from the copyright point of view. Several approaches aiming to resolve the challenge to some extent in the emerging area of multimedia protection were proposed and studied. This study proposes an approach to secure the authorized media sharing in Peer-to-Peer (P2P) networks. The P2P networks was initially designed for bandwidth saving, but its file sharing property was later on put to use for pirate. This situation has not been improved effectively until now. The approach aims to embed an unique-mark (fingerprint) into each authorized copy in the P2P networks so that it can be used to track the pirate initiator. This study also proposes another scheme for protecting the ownership of digital media files that have been circulated without copyright mark embedded. To protect this type of files, the ownership of each file needs to be stored associated with its meta-data (such as the ownership, title and artist) and can be identified correctly later on. Since the size and the number of the media files to be stored are extremely large, the mini versions (fingerprints) of the files become necessary to be derived. The common criteria of designing these two approaches are to ensure the fingerprint is compact, robust, discriminative, and ease of computation. To well balance the criteria, the sparse decomposition techniques play a very important role. The results of the tests under various distortions show the proposed fingerprinting schemes are very promising for real applications.

History

Language

English

Degree

  • Doctor of Philosophy

Program

  • Electrical and Computer Engineering

Granting Institution

Ryerson University

LAC Thesis Type

  • Dissertation

Year

2011