Audio Digital Watermarking Based On Auditory Perception

The digitalization of multimedia information and popularization of network techniques facilitate storage and transmission of the multimedia information. At the same time, the issue of safety and copyright of multimedia information are urgent to resolve. Traditional encryption techniques can not solve the problems. So digital watermarking was brought forward in 1990s. Digital watermarking refers to embedding symbolic information into digital information by some algorithm, the digital information usually are image, audio and video etc carriers. The influence of watermark embedding on these carriers can not be perceived, also the watermark can stand against kinds of attacks.Firstly, idea, model, performance, request, classification and application of audio watermarking are introduced in this thesis. Then kinds of present audio watermarking algorithms are reviewed and their advantages and disadvantages are summarized. Based on these summarizes, an improving algorithm of audio watermarking based on auditory perception is researched and put forward. The main works in this thesis are as follows:This thesis focuses on the problem how to improve robustness under premise of ensuring imperceptibility. The algorithm calculates masking threshold by auditory frequency masking at first. Then quantization steps are controlled by masking threshold. So the watermarking can be adaptively embedded with maximum intensity on premise of ensuring imperceptibility. At the same time, watermarks are embedded in those points whose values are great than the masking threshold for fighting against compression. Also one bit watermark information is repeatedly embedded by the idea of repeat coding in low and middle band coefficients of ever frame, so that the watermarking would have better robustness. A method is proposed to get the best repeat embedding times to solve the problem that how to confirm the best repeat times, so that we can get balance of robustness and calculation. Finally watermarking is attacked, the results of the experiment show that the embedded watermarking can withstand kinds of attacks such as noise, low-pass filtering, re-sampling and compression and has good robustness and imperceptibility, meantime it can be blindly detected, so it can be applied in protection and authentication of music and audio on the internet.