Binary image watermarking on audio signal using wavelet transform

Digital watermarking approaches have recently become an important field of study in a number of practical applications. Generally, digital watermarking is the process of embedding particular information data into other signal data in such a way that the quality of the original information data is maintained. Watermarking can be performed on images, video, text, or audio to protect them from copyright violation. Among all of these types of watermarking, audio watermarking is difficult since the quality of such signals is highly affected by the watermark code. This thesis introduces some approaches that have features and may be applicable for audio watermarking. Generally, the digital audio watermarking approaches may be performed in either the special domain or the transform domain. The approaches that are based on the transform domain are gaining more attention due to their robustness or resistance to attackers. The most commonly used transforms for achieving digital audio watermarking are discrete cosine transform (DCT), short-term Fourier transform (STFT), and digital wavelet transform (DWT). In this research work, the main attention is paid toward digital wavelet transforms such as Haar and Daubechies-4. In addition, this thesis investigates the process of embedding a binary image into various audio signals such as speech and music signals. The values of the pixels in the binary images are either 1 or 0. These pixels are inserted into samples of the speech signal in such a way that the quality of the speech signal is preserved. One main reason to consider the audio watermarking approaches in the wavelet domain is that many modern multimedia standards such as JPEG2000 and MPEG-4 are based on the discrete wavelet transform. These common standards have brought new requirements such as low bit-rate transmission, coding, and many other advantages. Finally, experimental results are presented to validate the performance of the proposed watermarking approach and to compare its performance with other related watermarking approaches. The results show that the performance, in terms of signal-to-noise ratio (SNR) and normalized correlation (NC), of the proposed watermarking approach outperforms other approaches based on DCT. In addition, the results show that the extraction process of the hidden code becomes complex when considering any signal processing operations such as MP3 compression and high pass filtering. The worst case is when using a high pass filter where the recovered binary image is not satisfactory.