Image Registry with Legally Supported Digital Right Management System

碩士 === 國立雲林科技大學 === 電機工程系碩士班 === 94 === The progress in digital multimedia technologies enriches the methodology and tools for digital content creation, while the advances in networking and mass storage technology make the spread of digital contents just a fingertip a way. However, the convenience o...

Full description

Bibliographic Details
Main Authors: Jin-zong Hsu, 許進宗
Other Authors: Day-fann Shen
Format: Others
Language:zh-TW
Published: 2006
Online Access:http://ndltd.ncl.edu.tw/handle/88324579010481098734
Description
Summary:碩士 === 國立雲林科技大學 === 電機工程系碩士班 === 94 === The progress in digital multimedia technologies enriches the methodology and tools for digital content creation, while the advances in networking and mass storage technology make the spread of digital contents just a fingertip a way. However, the convenience of the later makes piracy of digital contents very easy, thus, copyright protection become one of the most important issues in digital contents business. Digital watermarking is currently the main stream in copyright protection for digital images, however, after years efforts and experiences, researchers and engineers have become suspicious about the ability of watermarking in copyright protection. In this paper, we propose a copyright protection mechanism, which features the process of image registry to an arbiter for possible law enforcement (similar to trademark protection) as well as active search by the copyright owner for possible piracy. The core for both image registry and active piracy search is an image identification technique, where an input query image is examined if it is an modified (attacked) version of a legally registered image in the database. We propose two image identification techniques based on two image features respectively, namely the block based Eigen-Value in pixel domain and the block based partial DCT coefficients in DCT domain. We also implemented other five algorithms in order to find the best image identification technique to be used with the proposed image copyright protection mechanism. We have set up the performance evaluation criteria and conduct extensive experiments, experimental results show that the proposed block based partial DCT coefficients (BDCT) based algorithm performs the best and the proposed Eigen-value based algorithm performs the third under 10 possible piracy attacks.