토널 마스커 특성을 이용한 오디오 워터마킹
(An) audio watermarking method using the attribute of the tonal masker
토널 마스커 오디오 워터마킹 컴퓨터공학;
- 원문 URL
In this dissertation we propose the tonal masker as a new audio watermarking attribute for audio watermarking and show the method of audio watermarking using the audio watermarking attribute and evaluate the method. This dissertation describes that it is possible to modulate tonal maskers imperceptibly on the basis of the theory of tonal masker composition in MPEG psychoacoustic model 1. And according to the existing researches, we prove that the modulation for the energies of the frequencies that compose a tonal masker is imperceptible. The tonal masker is one of the important components of the human perception, and is very stable after various signal processing which make a little change of audio quality. We showed the stability of the signals that compose tonal maskers by the experiment result and analyzed whether the tonal maskers to embed watermark were distributed sufficiently or not. Thus, tonal masker attribute is able to change imperceptibly and is the strong watermarking attribute against various signal processing. The proposed watermark encoding method in this thesis modulates the energies of the constitution signals of the tonal masker that are selected by a key. The modulation of the energies mean that a watermark bit is embedded. In the detecting method, we search the modulated tonal masker using the key without the original audio and detect the embedded watermark bit In order to evaluate robustness of the proposed method objectively, we compared the proposed method with the existing method that have the embedding of the spread spectrum signal and the watermark extracting using LP analysis. Before comparison of robustness, we adjusted audio quality of two methods for comparison similarly. After each attack of noise addition, compression, band-pass filtering, re-sampling, echo transform, and equalization, the detection ratios of the proposed method were more improved than those of the existing method. And also the proposed method is possible to detect the watermark for the error of frame synchronization to some degree when it was detected At the subjective evaluation of sound quality, the result of the proposed method showed average 0.31 SDG(Subjective Diff-Grades). This result is considered as a practicable audio quality. However, it is impossible to detect the watermark after time scale change attacks. Therefore a new algorithm for synchronization of frame is required. We introduced the tonal masker that is one of the important audio components of the human perception into audio watermarking. And we analyzed the tonal masker in various aspects as the attribute of audio watermarking. When the results of our research is taken into account, the Audio watermarking system using the attribute of the tonal masker was performed at a good efficiency for imperceptibility and robustness. And we will continue researches of the faster method for the application of real time detection and of a new algorithm for synchronization of frame.