基于设备本底噪声频谱特征的手机来源识别

发布时间：2018-01-10 22:06

本文关键词：基于设备本底噪声频谱特征的手机来源识别　出处：《电信科学》2017年01期 　论文类型：期刊论文

【摘要】：随着手机录音设备的普及以及各种功能强大且易于操作的数字媒体编辑软件的出现,手机来源识别已成为多媒体取证领域的热点问题。将本底噪声作为手机的"指纹",提出了一种基于本底噪声的手机来源识别方法。该方法先通过静音段录音的估计得到本底噪声;然后计算本底噪声的频谱特征在时间轴方向上的均值,将其作为手机来源识别的分类特征;最后采用主成分分析(PCA)法对特征进行降维,并采用支持向量机(SVM)进行分类。实验部分对24款主流型号的手机进行了分类,结果表明本文方法的平均识别准确率(accuracy)和平均召回率(recall)达到了99.24%,同时也验证了相比MFCC,本底噪声有更加优越的性能。
[Abstract]:With the popularity of mobile phone recording devices and the emergence of various powerful and easy to operate digital media editing software. Mobile phone source identification has become a hot issue in the field of multimedia forensics. Background noise is regarded as the "fingerprint" of mobile phone. A method of mobile phone source recognition based on background noise is proposed. The background noise is obtained by the estimation of mute segment recording. Then, the average value of the spectrum feature of background noise in the time axis is calculated, and it is regarded as the classification feature of mobile phone source recognition. Finally, the principal component analysis (PCA) method is used to reduce the dimension of the features, and support vector machine (SVM) is used to classify the features. In the experimental part, 24 mainstream mobile phones are classified. The results show that the average recognition accuracy and recall rate of this method are 99.24 and 99.24, and the comparison of MFCC is also verified. Background noise has better performance.
【作者单位】：宁波大学;
【基金】：国家自然科学基金资助项目(No.61672302;No.61300055) 浙江省自然科学基金资助项目(No.LZ15F020002;No.LY17F020010) 宁波大学科研基金资助项目(No.XKXL1405;No.XKXL1420;No.XKXL1509;No.XKXL1503);宁波大学科研创新基金资助项目(No.G16079);宁波大学王宽诚幸福基金资助项目~~
【分类号】：TN929.53
【正文快照】： Foundation Items:The National Natural Science Foundation of China(No.61672302,No.61300055),Natural Science Foundation of ZhejiangProvince of China(No.LZ15F020002,No.LY17F020010),Scientific Research Foundation of Ningbo University(No.XKXL1405,No.XKXL1420,

【相似文献】