24-07-2012, 01:36 PM
COMPARISON BETWEEN TWO WATERMARKING ALGORITHMS USING DCT COEFFICIENT, AND LSB REPLACEMENT
COMPARISON BETWEEN TWO WATERMARKING.pdf (Size: 177.32 KB / Downloads: 61)
ABSTRACT
Digital watermarking is a method through which we can authenticate images, videos and even texts. Watermarking
functions are not only authentication, but also protection for such documents against malicious intentions to change
such documents or even claim the rights of such documents. In this paper two watermarking algorithms are
simulated. The first algorithm is based on the Discrete Cosine Transform (DCT)(in the frequency domain) and the
second algorithm is based on the least significant bit (LSB)replacement (in the spatial domain). The results are
shown and compared under different kinds of attacks.
INTRODUCTION
Digital watermarking is a technique which allows
an individual to add hidden copyright notices or
other verification messages to digital audio, video,
or image signals and documents. Such a message
is a group of bits describing information pertaining
to the signal or to the author of the signal (name,
place, etc.). The technique takes its name from
watermarking of paper or money as a security
measure. Digital watermarking can be a form of
steganography [1], in which data is hidden in the
message without the end user's knowledge.
Watermark Extraction Process:
To obtain the extracted watermark from
watermarked image, the following procedure was
performed:
Step 1. Original image is used for watermark
retrieval, as in the embedding process, original
image and watermarked image are divided into a
number of 4×4 blocks.
Step 2. Calculate the DCT transform coefficients
for each block in both original image and
watermarked image.
Gaussian Noise:
In this section, Gaussian noise is applied over the
watermarked image with zero mean and different
variances, where the variance of the noise is a
function of the image intensity values in the
watermarked image. From table (1), and figure (3),
it is found that as long as variance increases the
PSNR decreases and so the NCC, also when the
variance increases up to 0.001 the PSNR decreases
down to 29.9212 but the extracted watermark still
can be distinguished.
Cropping:
In this section, a square at the image center is
cropped out from the watermarked image then the
watermark is extracted after cropping with
different dimensions, table (1) illustrates the effect
of applying cropping with different dimensions on
the watermarked image.
It is clear that the NCC gradually decreases which
means that this algorithm is robust against
cropping, as the recognition of a degraded
watermark is easy.