03-05-2013, 02:07 PM
Peak Detection based Spread Spectrum Audio Watermarking using Discrete Wavelet Transform
Peak Detection based Spread.pdf (Size: 203.57 KB / Downloads: 45)
ABSTRACT
This paper proposes a robust spread spectrum based audio
watermarking scheme using Discrete wavelet transform (DWT).
Here we use a peak detection algorithm to obtain high
robustness. In this watermarking scheme, watermarks are
embedded into the peak value of detail coefficients of,
transformed audio signal. It is a blind watermarking technique
which is used for providing copyright protection and content
authentication for digital content. Tests shows that the proposed
watermarking scheme is robust against signal processing attacks
like resampling, requantization, MP3 compression, noise
addition. Performance of this technique is analyzed by
calculating the percentage match and SNR values.
INTRODUCTION
As the use of digital data like audio, vedio and images had
increased, illegal copying and distribution of these digital data
had also increased. To overcome these problems, people have
come up with many techniques like hardware based copy
protection system etc. As these traditional methods do not serve
the purpose, therefore some new techniques have been
developed. Digital watermarking [1] [2] is one of the very
important method, that have been widely used to prevent from
illegal copying and distribution.
Audio watermarking [3] [4] is the process in which
information is embedded into the audio signal. In other words
digital audio watermarking involves the hiding of data with in a
digital audio file, so that later it can be used to trace the
authenticity of audio file. Audio watermarking techniques
satisfy three characteristics which are inaudibility, robustness
and bit rate. These Requirements are represented by the magic
triangle [5] shown in figure 1. It is called magic triangle because
inaudibility is the main requirement of audio watermarking
process. Therefore it is presented in the upper portion of the
triangle but the other two requirements cannot achieve together
which means if data rate is high robustness is low and
robustness is high data rate is low. Therefore it is presented in
two corners of the triangle.
EXISTING WORK
There exists many SS audio watermarking techniques that have
already been developed. Here we will discuss few of the
techniques with more focus on SS based watermarking
technique. Malvar et al [7] proposed two techniques one is
general spread spectrum audio watermarking technique and
second one is improved spread spectrum audio watermarking
technique.
PROPOSED WORK
In this paper an audio watermarking scheme using spread
spectrum technique in DWT domain is proposed. Here the audio
signal is first decomposed upto the third and fourth level of
detail coefficients using DWT transformation. After that
watermark is embedded into the highest peak values of detail
coefficients. A collection of 0s and 1s are used as watermark
bits. The mechanism we have used for watermark embedding
and extraction purpose is explained below.
SIMULATION RESULTS AND
ANALYSIS
In this section the simulation results of watermarking scheme
are presented. The performances of the scheme are analyzed
against signal processing attacks such as MP3 compression,
resampling, requantization, noise addition. Here the number of
watermark bits after spreading process is 1000. A set of wave
files are used such as pop, rock, country, speech etc. We have
performed simulations using MATLAB 7.9. In our proposed
watermarking scheme Daubechies-2 wavelet transform are used
for decomposition of audio files.
CONCLUSION
A secured audio watermarking based on spread spectrum
technique is presented in this paper. A peak detection method is
used for embedding of watermark bits. A set of experiments
have been performed and their performances are evaluated on
the basis of SNR and percentage match. Here the performance
of existing method is compared with our proposed one.
In our proposed scheme the watermark bits are extracted without
using the original signal. From experimental results we observe
that our proposed technique provides low bit error rate and high
SNR in comparison to existing techniques.