ABSTRACT
Along with the large production of the multimedia, the copyright protection of them
is becoming more and more important. Digital watermarking has received great
attention recently as a kind of new technique to protect the copyright. The watermarking
technique embeds invisible codes in digital data, and it can be extracted to identify the
copyright owner when there are some arguments about the real owner. Although the
code could also be used to protect the digital production, it will lose such function if the
production is decoded, while the watermark in the protected data is hard to remove.
Hence, there are more advantages of digital watermarking in protecting the copyrights.
Based on large research results of the researchers working on digital watermarking,
this paper focuses on the algorithm for audio watermarking. The mainly contributions of
it are as follows:
(1) An algorithm for audio watermarking based on wavelet decomposition is
proposed. The algorithm first deals with the audio signal using wavelet decomposition
to get the time-variant low frequency components of the audio signal. Then, the
spectrum of the low frequency components is spread and the watermark is embedded
into the low frequency components spread by using the MDCT (Modified Discrete
Cosine Transform). Finally, the better veracity of extracting the watermark is assured by
using error correction. By using the method, the amount of the embedded watermark
bits is increased; the security and imperceptibility of the embedded watermark is
ensured. In the mean time, the watermarking scheme is also robust to the common
signal processing.
(2) A new binary audio watermarking algorithm is studyed. As watermark message, a
liner frequency modulation (LFM) signal is embedded in an audio signal. Since LFM
signal is localized as a straight line in the time-frequency (TF) plane, we can recover the
original watermark sequence by detecting the line in the image of the TF plane.
Although Hough-Radon transform (HRT) can correctly find the embedded watermark
message after common signal operations for bit error rates up to 20% by detecting the
line in the image, it needs to search for the parameters in a global space, and therefore it
has high computational capacity and costs large amount of time, which makes the