基于奇异值分解与融合的音频盲认证算法  

Blind Audio Authentication Algorithm Based on SVD and Amalgamation

作　　者：王潇[1] 柏森[1] 赵波[1] 

机构地区：[1]重庆通信学院,重庆400035

出　　处：《中山大学学报（自然科学版）》2004年第A02期188-191,共4页Acta Scientiarum Naturalium Universitatis Sunyatseni

基　　金：重庆市应用基础研究资助项目(8028)

摘　　要：提出一种适用于检验音频真实性和完整性的盲音频认证方法。它采用奇异值分解(SVD)与融合技术相结合,将音频奇异值分解所得的奇异值作为不变特征,利用融合将其嵌入到原始音频中。验证时无需原始音频数据,利用融合的逆过程提取不变特征,进行匹配,达到音频认证的目的。实验结果表明,该算法不但可以有效地检测出音频中被篡改内容的位置,而且能将随机噪声同恶意篡改区分开。To verify the authenticity and integrity of audio, we present an audio authenticaion method, in which singular value decomposition (SVD )is used to extract singular value as invariable features of host audio. Combining with amalgamation, we insert these extracted features into the host audio at the sender. Then at the receiver, we extract inserted features as authentication code, and compare them with those features from SVD. After setting proper threshold, we can make our judgments at last. Experimental results show that the proposed method not only can distinguish Gausssian noise from malicious tampering but also can detect where tampered regions are.

关 键 词：音频认证 奇异值分解 融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]