Digital Watermarking Using

Wavelet Transforms.

Presented by:
Sugandhita Wadhera
072045
EE,G2
 INTRODUCTION

• These days multimedia data transforming and sharing

is common to many people.

• Multimedia data is easily copied and modified, so

necessity for copyright protection is increasing.

• Digital watermarking has been proposed as technique

for copyright protection of multimedia data.
 INTRODUCTION (continued..)
• Digital watermarking invisibly embeds copyright
information into multimedia data.

• The process involves the modification of the original

multimedia data to embed a watermark containing key
information such as authentication or copyright codes.

• The embedding method must leave the original data un-

changed, yet should impose modifications which can be
detected by using an appropriate extraction algorithm.
• In general, the digital watermarking is
classified into two classes by depending
on the domain of watermarking
embedding:

1. The spatial domain watermarking.

2. The frequency domain watermarking.

 Spatial Domain Watermarking

• The spatial domain watermarking algorithms usually

embed the watermark to the least significant bits (LSBs)
of the image pixels.

• The detection of changes in small details of the image is

based on mathematical morphology.

• Furthermore, it is able to reject small distortion

introduced by high quality image compression.

• But frequency localization of modifications is difficult.

 Frequency Domain Watermarking

• The frequency domain watermarking algorithms can

embed more bits of watermark and are more robust
against attack.

• Thus, all the watermark embedding and extracting

process is performed in frequency domain.

• The approach provides the simultaneous spatial

localization and frequency spread of the watermark
within the host image to provide robustness against
widely varying signal distortions such as cropping
and filtering.
• In general, we use:

1. Discrete cosine transform (DCT) and

2. Discrete wavelet transform (DWT).

 Why Discrete Wavelet Transform (DWT)?

• Wavelet transform (WT) allows the decomposition of the

signal in narrow frequency bands while keeping the basis
signals space limited.

• The main advantage of wavelets over Fourier analysis is

that they allow both spatial and frequency resolution.

• Also the algorithm using the largest amplitude DCT

coefficients can result in a poor image quality due to the
manipulation of the most significant portion of the image.
 Watermarking Technique

• Research into human perception indicates that the

retina of the eye splits an image into several frequency
channels each spanning a bandwidth of approximately
one octave.

• Similarly, in a multiresolution decomposition, the

image is separated into bands of approximately equal
bandwidth on a logarithmic scale.

• For this reason, the wavelet decomposition is

commonly used for the fusion of images.
• Fusion is a sensor-data-compressed information
problem in which several signals are merged into one.

• Since digital watermarking involves the merging of a

watermark with a host signal it follows that wavelets
are attractive for the watermarking of images.

• In addition, the technique is “unsupervised” since the

host image is not required for watermark extraction.
 Architecture

• The original image f in which the watermark is

embedded is referred as the host image.

• The watermark is embedded into the detail wavelet

coefficients of the host image with the use of a key.

• This key is randomly generated and is used to select

the exact locations in the wavelet domain in which to
embed the watermark.

• For each coefficient within the wavelet domain, the

key has a corresponding value of one or zero to
indicate if the coefficient is to be marked or not,
respectively.
The technique is comprised of the three stages described
below:
• Stage I: Compute the Lth-level discrete wavelet
decomposition of the host image to produce a sequence of
3L detail images, corresponding to the horizontal, vertical
and diagonal details at each of the L resolution levels.
• Stage II: Consider each resolution level l, and coefficient
location (m, n). If the associated value of the key k is one
then proceed. Otherwise do not embed a mark.
• Stage III: The corresponding Lth level inverse wavelet
trans-form the fused image components is computed to
form the water-marked image. The parameter Q is user-
defined.
 WATERMARK EXTRACTION AND
DETECTION
• The objective of the watermark extraction process is
to reliably obtain an estimate of the original
watermark from a possibly distorted version of the
watermarked image.

• The detection process requires knowledge of the

watermark ω (m, n) and the key k (m, n).

• The image for which the extraction process is to be

applied is represented by r (m n).
 Result:

• An attacker cannot easily determine the exact key

given a watermarked image if the specific wavelet
transform used in the de-composition of Step I is kept
a secret and Q is unknown.

• To destroy the watermark for certain, all of the

coefficient values may have to be randomly changed
which can cause visible degradation in the image
therefore making the image useless to the attacker.
THANK YOU.