Hough line detection algorithm on AEM of ENF signal for audio authentication

Electric network frequency (ENF), a signature in multimedia application, is been utilized as a significant tool to check the authenticity of media recordings. In this type of authentication systems the ENF signals are extracted from the tampered recording and compared with reference ENF database. This establishes a popular application of audio timestamp verification. In this paper the authentication system incorporates the tampering detection of the audio recordings along with the audio timestamp verification. The system utilizes the Absolute Error Map (AEM) between the ENF signals obtained from the reference database and the testing audio recordings. The AEM serves as an ensemble of the raw data associated with the ENF matching process. Through the analysis of AEM, a new Hough Transform based Line Detection algorithm is introduced in this paper. This can be utilized for jointly deal with timestamp verification and tampering detection. The authentication scheme first checks whether the tampering has occurred or not If testing data is tampered with, it outputs the tampered type and region along with the timestamp verification. The effectiveness of the system is demonstrated by analyzing it under different level of AWGN and false positives. Finally the performance of Hough Transform based Line Detection algorithm is analyzed with that of the Exhaustive search algorithm utilized in an AEM based authentication system.