This document discusses image authentication techniques using digital watermarking. It defines digital watermarking as a technique for inserting information like a watermark into an image that can later be extracted or detected to protect copyright and ensure tamper resistance. The process involves embedding a watermark during insertion, potential attacks on the watermarked image, and detection of the watermark. Various domains for embedding watermarks are discussed like the spatial domain and transform domains like DCT and DWT. Properties of good watermarks and classifications of watermarks like robust and fragile are also summarized.

1. Image Authentication
Techniques
based on Watermarking
Nawin Kumar Sharma
CDS16M009@IIITDM.AC.IN

2. Introduction
 Image authentication techniques have become very important due to
the emergence of wide range multimedia applications.
 Images have to cross non-secure channels like the Internet for
transmission.
 Images must be protected against manipulation attempts, especially
in the case of military and medical applications.
 Unauthorized replication is also a problem, which leads copyright
dispute for actual ownership.

3. Image Authentication
 A mechanism to ensure
 The image is not faked
 The image is originated from a trusted sender
 Authentication
 Data Authentication (Content protection)
 Owner authentication

4. Digital Watermarking
 Digital Image Watermarking is a technique for inserting
information (the watermark) into an image, which can
be later extracted or detected for protecting digital
content copyright and ensuring tamper-resistance, which
is hard to remove by unauthorized persons.

5. Digital Watermarking Process
A watermarking system is
usually divided into three
distinct steps.
• Embedding
• Attack
• Detection

6. • Embedding F : Watermarked Image = Function (Cover, Watermark, Key)
• Extraction F : Watermark = Function (Watermarked Image, Cover, Key(
• Non-blinded detection: Original host image(Cover) is used in detection part.
• Blinded detection: When Cover is not used in detection part.
Digital Watermarking Process
(Non-blinded and with key)

7. Embedding
In embedding, an algorithm accepts the host and the data to be
embedded, and produces a watermarked signal.
• Inputs to the scheme are the watermark, the cover data and an
optional public or secret key. The output are watermarked data.
The key is used to enforce security.
• Image Content
 Spatial Domain (Ex-Least Significant Bit Watermarking)
 FFT - Magnitude and Phase
 Wavelet Transforms
 DCT Coefficients
7

8. Extraction/Detection
Detection (often called extraction) is an algorithm which is applied to
the attacked signal to attempt to extract the watermark from it. If the
signal was unmodified during transmission, then the watermark still is
present and it may be extracted.
 Inputs to the scheme are the watermarked data, the secret or public
key and, depending on the method, the original data and/or the
original watermark.
 The output is the recovered watermarked W or some kind of
confidence measure indicating how likely it is for the given
watermark at the input to be present in the data under inspection.
8

9. Watermark Detection
* = 
Suspected Image
Extracted
Watermark,W2
Original
Watermark,W1
Correlation
•Watermark Extracted from Suspected Image
•Correlation, =E[W1*W2]/{ E[W12]E[W22]}
•Compute correlation of Extracted and Original Watermark
•Threshold correlation to determine watermark existence

10. Attacks on Watermarking
 Unintentional
 All image manipulations commonly used to prepare images for print
publication.
 For example: Resizing, rotation, sharpening, contrast modification,
compression, etc.
 Intentional (Malicious)
 All the well-known intentional attacks include: Disabling, altering,
embedding new watermark etc.
10

11. 11
Attacks on the Watermarked image

12. SVD Domain Watermarking
 [U1,S1,V1] = svd(A); // A is cover image
 temp = S1+(a * W); // a is scaling factor, and W is watermark
 [Uw,Sw,Vw] = svd(temp);
 Aw = U1*Sw*(V1.'); // Aw is watermarked image
 Extraction of watermark in SVD domain
 [Uw1,Sw1,Vw1] = svd(Aw);
 D = Uw*Sw1*(Vw.') //temp;
 W = (D-S1)/a;

17. Watermark Properties
 Watermark should appear random, noise-like
sequence
 Appear Undetectable
 Good Correlation Properties
 High correlation with signals similar to watermark
 Low correlation with other watermarks or random
noise
W=[1 0 0 1 0
0 1 1 0 1
1 1 0 1 0
0 1 1 1 1
0 1 0 0 0]

18. Steganography vs. Watermarking
 Steganography is to hide a message m in data d such eavesdropper cannot
detect the presence of m in d.
 Watermarking is to hide a message m in data d, to obtain new data s0 an
eavesdropper cannot remove or replace m in d.
 Main aim of Steganography is imperceptibility of human senses, whereas main
aim of watermarking is robustness.

19. Cryptography vs. Watermarking
 Cryptography is the most common method of protecting digital content and is
one of the best developed science.
 However, encryption cannot help the seller monitor how a legitimate
customer handles the content after decryption.
 Digital watermarking can protect content even after it is decrypted.

20. Classification of Watermarking

21. Classifications of digital watermarking
o Robust watermark
 It sticks to document ( image, audio, video or text) to which it is embedded.
 Removing it destroys the quality of signal.
 It is used for copyright protection.
 Frequency based techniques are very robust against attack involving image compression
and filtering because the watermark is actually spread through the image.
 As high frequency component is not resistant to JPEG compression so low frequency
component are used for watermarked embedding.
o Fragile Watermark
 It breaks very easily on modifying host signal.
 It is used for temper detection, finger printing and digital signature.
 Hash code of the host image can be used as fragile watermark for tamper detection.
o Semi Fragile Watermark
 It is sensitive to signal modification and gives information about nature and location of
attack also, it provides data authentication.

22. Limitations / Conclusions
 Rapidly growing field of digitized images, video and audio has
urged for the need of protection.
 Watermarking is a key process in the protection of copyright
ownership of electronic data (image, videos, audio, ...).
 Digital watermarking does not prevent copying or distribution.
 Digital watermarking alone is not a complete solution for
access/copy control or copyright protection.
 Digital watermarks cannot survive every possible attack.22

23. References
1. Mohmmad Ali M. Saiyyad, Nitin N. Patil (2014). Authentication and Tamper
Detection in Images Using Dual Watermarking Approach IEEE
2. Qingtang Su, Gang Wang, Xiaofeng Zhang, Gaohuan,Beijing Chen (2017). A
new algorithm of blind color image watermarking based on LU decomposition
Multidim Syst Sign Process, Springer
3. Wikipedia article on Digital watermarking