چكيده به لاتين
Abstract:
In recent years, due to ever increasing development of telecommunication and computer networks, the digital multimedia contents i.e. Audio, Image and Video are easily accessible. Therefore, In order to prevent illegal copying of these digital contents, watermarking has been proposed.
In this thesis, we propose two methods of watermarking; blind and robust audio watermarking schemes, based on Fibonacci numbers properties and spread spectrum in transform domain. Watermark embedding is performed in the low-frequency band in which Human Auditory System has less sensitivity using Discrete Wavelet Transform. The key idea is dividing audio signal spectrum into small frames and modificating magnitude of 6th level of Discrete Wavelet Transform approximation coefficients by using Fibonacci numbers in the first proposed scheme and modificating magnitude of 6th level of Discrete Wavelet Transform approximation coefficients by using spread spectrum in the second proposed scheme. To improve given results, psychoacoustic model as well as decorrelation approach has been used.
The experimental results illustrate that the proposed methods have high robustness against various common audio signal processing attacks i.e. Gaussian noise addition, Low pass filter, Resampling, Re-quantizing, MP3 compresstion, Amplitude Scaling, Echo addition without signifant perceptual disturtion (ODG is about -1). The first proposed scheme has a capcity about 686 bps data payload and the second scheme has about 83 bps data payload. Also we can attain trade-off between capacity, robustness and impressibility, by choosing various kinds of sequence numbers in watermarking approach using Fibonacci numbers.
Keywords: Audio watermarking, Fibonacci Numbers, Discrete Wavelet Transform, Spread Spectrum, Psychoacoustic, Blind.
