03-07-2013, 04:53 PM
AN ALGORITHM FOR WATERMARKING MOBILE PHONE COLOUR IMAGES USING BCH CODE
AN ALGORITHM.pdf (Size: 386.81 KB / Downloads: 19)
ABSTRACT
This paper proposes a scheme for embedding phone
numbers into color images captured by a mobile phone
camera. Firstly, the phone number digits are transformed
using BCD encoder and the generated binary vector is
appended by the phone number checksum represented in
binary format. Then, this binary vector is coded by BCH
code before inserting it in the DCT blocks of the image.
The coded watermark information are embedded into a
predefined low frequency coefficient in the DCT domain.
The robustness of the scheme is assessed by using
“Stirmark” software package. The new algorithm
survived JPEG compression and 3x3 2D filtering.
INTRODUCTION
Image watermarking is a technique for inserting
hidden information into an image. The hidden
information can be used to protect the copyright
ownership of the image. The insertion of the extra
information should cause imperceptible degradation to
the host image. Different techniques were proposed in the
literature to tackle this research problem. Good reviews
on image watermarking techniques can be found in [1, 2].
The insertion of the watermark can be done either in
the spatial domain or in a transformed domain. However,
it was found that the transformed domain techniques
offer better robustness against image manipulation and
compression techniques. One type of transformed domain
is by using the discrete cosine transform (DCT). This
process employs the embedding of the watermark
information into certain coefficients in the frequency
domain. Usually, the insertion is done in the low
frequency coefficients so that it will survive attacks such
as low-pass filtering and JPEG compression.
THE PROPOSED ALGORITHM
The mobile phone number together with the international
code is used as a watermark for images. The process of
converting this sequence of decimal digits into a binary
vector is shown in Figure 1. This figure shows how to
convert a UAE phone number which consists of 14
decimal digits. The same process can be used with any
other country phone number. The first step in this process
is to convert the decimal digits into binary representation
using binary coded decimal (BCD) encoder. Each one of
the 14 digits is converted into a 4-bit binary number.
Then the checksum of the phone number is added to the
generated binary vector. This checksum can be used
during the decoding process to verify the correctness of
the extracted phone number.
The extraction process
The embedded information can be extracted by
performing 8x8 DCT transform on the Y-component of
the watermarked image. Although the proposed
algorithm is blind, it requires information such as the size
of the image, the embedding scale , and the security
key used during the embedding process.
Firstly, the watermark bits are extracted based on the
formula described in Equation(8). Basically, the
predefined frequency component in each 8X8 block is
divided by the scale and rounded to the nearest integer.
The extracted watermark will depend on the value of
resultant integer.
CONCLUSIONS
A new blind watermarking algorithm for color images
captured by using mobile cameras was presented. The
low frequency component of the DCT is used to embed
the phone number. Several watermarking strengths have
been examined and the best watermarking strength is
16 . The algorithm was tested with several mobile
phone numbers from several countries. In general, the
proposed watermarking algorithm has shown to be
resistant mainly to JPEG compression and 3x3 2D lowpass,
and median filtering attacks.