A robust intelligent audio watermarking scheme using support vector machine

作     者：Mohammad MOSLEH Hadi LATIFPOUR Mohammad KHEYRANDISH Mahdi MOSLEH Najmeh HOSSEINPOUR 

作者机构：Department of Computer Engineering Dezfoul Branch Islamic Azad University Dezfoul Iran Department of Computer Engineering Andimeshk Branch Islamic Azad University Andimeshk Iran Young Researchers and Elite Club Andimeshk Branch Islamic Azad University Andimeshk Iran 

出 版 物：《Frontiers of Information Technology & Electronic Engineering》 (信息与电子工程前沿（英文版）)

年 卷 期：2016年第17卷第12期

页      面：1320-1330页

核心收录：

学科分类：081203[工学-计算机应用技术] 08[工学] 0835[工学-软件工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：Project supported by the Dezfoul Branch Islamic Azad University Dezfoul Iran 

主　　题：Audio watermarking Copyright protection Singular value decomposition (SVD) Machine learning Support vector machine (SVM) 

摘      要：Rapid growth in information technology and computer networks has resulted in the universal use of data transmission in the digital domain. However, the major challenge faced by digital data owners is protection of data against unauthorized copying and distribution. Digital watermark technology is starting to be considered a credible protection method to mitigate the potential challenges that undermine the efficiency of the system. Digital audio watermarking should retain the quality of the host signal in a way that remains inaudible to the human hearing system. It should be sufficiently robust to be resistant against potential attacks, One of the major deficiencies of conventional audio watermarking techniques is the use of non-intelligent decoders in which some sets of specific rules are used for watermark extraction. This paper presents a new robust intelligent audio water- marking scheme using a synergistic combination of singular value decomposition （SVD） and support vector machine （SVM）. The methodology involves embedding a watermark data by modulating the singular values in the SVD transform domain. In the extraction process, an intelligent detector using SVM is suggested for extracting the watermark data. By learning the destructive effects of noise, the detector in question can effectively retrieve the watermark. Diverse experiments under various conditions have been carried out to verify the performance of the proposed scheme. Experimental results showed better imperceptibility, higher robustness, lower payload, and higher operational efficiency, for the proposed method than for conventional techniques.