Research, Database Watermarking Algorithm Based On Audio Information

A wealth of digital resources are provided for learners in modern distance education, which help greatly in reducing the cost of education, expanding the scale of education and achieving life-long education. These resources are easy to copy, modify and delivery, on one hand, they are convenient to share; on the other hand, pirate may occur. So, how to protect the copyright of digital resources is becoming a hotspot. As a major carrier of distance education resources, database copyright protection need be resolved urgently. Currently, watermarking is the most popular technology of database copyright protection.This paper reviews the development of watermarking relational database and the researches on it, and outlines basic theory of watermarking relational database, including the concept of digital watermarking, general models, the characters, attacks and research fields. Most watermarks used in current watermarking algorithms are less meaningful, and don't have close correlationship with copyright owner. Therefore, speech signals are introduced as watermark to be embedded into the relations for the first time. The main researches are as follows:(1) Discusses the necessity and feasibility of introducing speech as watermarks. Speech signals carry abundant information, which can improve the meaning of watermarks. Speech signals are unique and stabile, the correlativity of watermark can be improved by using both meaning of speech and feature of speech when identifying copyright owners. At the same time, speeches are easy to get and process, which can guarantee the feasibility of introducing speech as watermarks.(2) Proposes a speech based algorithm for watermarking relational databases. Considering the significance, identification and stability of watermark, we choose speech of copyright owners as watermark information. The embedding watermarks are generated after a certain signal processing. And the corresponding watermark insertion algorithm and watermark detection algorithm are described in detail.(3) Proposes a multiple watermarking relational database algorithm for the protection of multiple owners'. Watermark insertion algorithm inserts the watermark which is mixed by a number of copyright owners' speech signal into suspicious database. We get a number of copyright owners' speech signal using independent component analysis(ICA) in watermark extraction algorithm.To sum up, the main contributions of this paper include: demonstrating the necessity and feasibility of introducing speech as watermarks, proposing a speech based algorithm for watermarking relational database and a speech based algorithm for multiple watermarking relational database. Analyses and experimental results show that the proposed algorithms have excellent performance, good application in protecting copyright of relational databases.