Towards Design and Implementation of

Image Authentication / Secret Message

Transmission Technique using

Steganographic Approaches

Thesis submitted for the Degree of Doctor of Philosophy (Engineering) in the

Faculty of Engineering, Technology and Management

University of Kalyani

By

Nabin Ghoshal

Under the Supervision of

Prof. Jyotsna Kumar Mandal

Department of Computer Science and Engineering

University of Kalyani

Kalyani, Nadia, West Bengal, India

January 2011

Steganography or “hidden writing” is a technique that allows one to hide binary data within an

image with the addition of very few noticeable changes. Its goal is to achieve secrete

communication through multimedia carrier e.g. image, video and audio files between two parties

those are interested in hiding. To overcome the point of attack here is always concealing the very

existence of the embedded data into the carrier. The hiding is designed to achieve efficient trade-

offs among the three conflicting goals of maximizing rate of information embedding, minimizing

distortion between the host and composite signal, and maximizing the robustness of embedding.

For decades people strove to develop innovative methods using steganographic approaches for

secrete communication. Steganography have a long and exciting history that goes back to

antique. It can be traced back to 440 BC in ancient Greece [80]. They were used a technique for

writing a secrete message on a wooden panel cover it by a wax, and then write a message on the

wax. In another situation where steganography was used by spies, prisoners, and solders during

World War II because mail was seriously inspected. So many instances were happened, in case

of postal censors crossed out anything that looked like sensitive information and prosecuted

individuals for suspicious mail. To prevent the secrete message from being delivered, censors

even randomly deleted innocent-looking sentences or entire paragraph.

Recently the interest is increasing in the digital steganography on carrier multimedia image.

Image trafficking for commercial applications across the network is increasing day by day due to

the proliferation of internet. For the exponential growth of potential computer user, secrete

communication and image authentication using steganographic approaches are the demanded

area of research. Image authentication techniques have recently gained great attention due to its

importance for a large number of multimedia applications. Digital images are increasingly

transmitted over non-secure channels such as the Internet. Therefore, military, medical and

quality control images must be protected against attempts to manipulate them; such

manipulations could tamper the decisions based on these images. To protect the authenticity of

multimedia images, several approaches have been proposed. These approaches include

conventional cryptography, fragile and semi-fragile watermarking and digital signatures that are

based on the image content. Data hiding in the image has become an important technique for

image authentication and identification. Ownership verification and authentication is the major

task for military people, research institute, and scientist. Image authentication is a technique for

inserting information into an image for identification and authentication. Fig. 1.1 and Fig. 1.2 are

shows the typical steganographic scheme and the different embodiment of data hiding

respectively.

Key Key

Cover image Stego. image Source image

Message/

Image Message/Image

Sender Receiver

Fig. 1.1: A typical image authentication process

Fig. 1.2: A different embodiment of data hiding

1.1 Cryptography, Watermarking and Provable Security

The primary tool available for data protection is encryption. Steganography and cryptography are

counter parts in digital security the obvious advantage of steganography over cryptography is

that messages do not attract attention to themselves, to messengers and to recipients also. In

cryptosystem the security can be provided by scrambling the original content i.e. ciphertext as a

result it generates the attention to eavesdroppers. Digital watermarking is the process of hiding

the watermark imperceptibly in the content. This technique was initially used in paper and

Hiding

Algorithm

Extraction

Algorithm

Fragile

Fingerprint

Data Security system

Cryptography Information Hiding

Steganography Watermarking

Linguistic Technical Robust

Imperceptible Visible

Text Audio Video

Digital

Image

Song

In Spatial

Domain

In Frequency

Domain

currency as a measure of authenticity. Data hiding primarily refers to a digital watermark which

is a piece of information hidden in a multimedia content, in such a way that it is imperceptible to

a human observer, but easily detected by a computer. The principal advantage is that the

watermark is inseparable from the content. So, information security and image authentication has

become very important to protect digital image document from unauthorized access.

The main contributions of this thesis are divided into two categories one is image authentication

under spatial domain and another is under frequency domain. The results which establish these

works fall under several categories: one bit steganography in spatial domain, two bits

steganography in spatial domain, three bits steganography in spatial domain, one bit

steganography in frequency domain, two bits steganography in frequency domain and three bits

steganography in frequency domain. Here I summarise the results in each categories.

1-bit-steganography in spatial domain (1bss)

A 1-bit-steganography for grayscale images allows two parties with a shared secrete to send

hidden messages undetectably over a public channel. The presented work emphasizes on image

protection or authentication by hiding message/image using a hash function [ ] with a session key.

The scheme uses efficient insertion within a byte, which may be conform proper unauthorised

access while passing across the networks. Another one bit steganography technique is developed

for image authentication for colour images. This work emphasizes on information and image

protection against potential enemy while being transmitted across the networks. The

authentication or ownership verification is done by hiding secret data within the source image

byte by selecting 3 x 3 masks[ ]. In each image byte within the mask one bit of secrete data is

embedded at the randomly selected position among 1st to 5

th form LSB which conform proper

authentication and identification of the image.

2-bits-steganography in spatial domain (2bss)

2-bits-authentication technique AI/HLVD[ ] for grayscale images for shearing more secrete data

between two parties. Two bits of the authenticating message/image are inserted per byte of the

source image. Here during the embedding process the embedding is done on the size of

authenticating message/image, content of the authenticating message/image and the 128 bits

message digest (MD-5) generated from the authenticating message/image. A bit wise XOR

operation has also been performed with the inserted bits and hence a 128 bits message digest

generated from source image to enhance the security. Another method is presented an image

authentication [ ] and secures message transmission technique by embedding message/image into

colour images. Authentication is done by embedding message/ image by choosing image blocks

of size 3 x 3 called mask from the source image in row major order. The dimension of

authenticating image followed by MD-5 key and then the content of authenticating

message/image are also embedded. This is followed by an XOR operation of the embedded image

with another self generated MD-5 key obtained from the source image.

3-bits-steganography in spatial domain (3bss)

The 3-bits-steganography technique FBIA [] emphasizes on information and image protection

against unauthorized access and to insert large amount of messages/image data along with

message digest MD in to the source grayscale image for image identification and also to transmit

secure message within the image. Image authentication is done by embedding message / image in

spatial domain by choosing image blocks of size 3 × 3 from the source image in row major order.

Three bits of authenticating message /image/message-digest are fabricated within each source

image byte of each image block where the position is chosen randomly using a hash function. The

second method presented [] an image/legal-document authentication and secures message

transmission technique by embedding message/image/message-digest into colour images. Image

authentication is done by embedding message/image within the image pixels of source image.

Legal document authentication is done by embedding the authenticating image and self generated

message digest (generated from signed document part) into the image part of the legal document.

Techniques proposed are implemented on the digital multimedia images in spatial domain.

Results, discussion and comparison are drawn with respect to the popular existing authentication

technique S-Tools. Comparison of these techniques with S-Tools with respect to Histogram,

Noise analysis, Standard Deviation analysis, Pick Signal-to-Noise Ratio (PSNR), Image Fidelity

(IF) and Mean Square Error (MSE) are also done. The equations are given below those are used

to calculate different parameters for comparison. Histogram analysis is mainly the visual

interpretation of original and stego. images using different techniques. Noise is calculated using

the equation 1.1, where piE and pi

S are pixel values of the i

th pixel in both the embedded and

source image respectively. Noise is computed by finding the average of 4 direct neighbour pixels

in the 3 x 3 mask (Fig. 1.3) around the pixel pi. pjE and pj

S are 4 direct neighbour of pixel pi.

Standard deviation is calculated using the equation 1.2. In equation 1.2, x is the arithmetic mean

and N is the total number of different values and xi(s) are different level of quantum values.

∑∑∑×

=

==

×

+

−

+

=nm

i

j

S

j

S

i

j

E

j

E

i pppp

Noise1

4

1

4

1

3355

--------------------------------- (1.1)

Fig. 1.3: Noise detecting mask 4 neighbour pixels

( )∑=

−=N

i

i xxN 1

2

1σ ---------------------------------------- (1.2)

��� � �

��∑ �, � ��, �

��, ---------------------------------------- (1.3)

���� � 10 ��������/���� ------------------------------------ (1.4)

� � 1 � ∑ �, � ��, ��

/ ∑ �, �

�, �, ------------------------ (1.5)

Equation 1.3 is for calculating the Mean Square Error for measuring the amount of noise

integration in the stego. image after authentication. In equation 1.3, MN is the dimension of

original image Im,n and ��, is the stego. image whose coordinates are (m,n). PSNR is calculated

by equation 1.4 where R is max (�, � ) of original image �, . For higher PSNR value in dB

indicates minimum noise integration. IF is calculated using the equation 1.5. The calculated value

is for differentiating the original and stego. images, here �, and ��, are original and stego.

images.

1-bit-steganography in frequency domain (1bsf)

P2

Pi

P4

P3

P1

This novel steganographic schemes DFTMCIAWC based on Discrete Fourier Transformation

(DFT) to authenticate multimedia colour images in frequency domain and two parties can share

secrete data via wireless communication. Authentication is done through embedding secrete

message/image into the transformed frequency components of the source image at message

originating node. The DFT is applied on sub-image block called mask of size 2 x 2 in row major

order where authenticating message/image bit is fabricated within the real frequency component

of each source image byte except the first frequency component of each mask.

2-bits-steganography in frequency domain (2bsf)

This novel technique for Image Authentication in Frequency Domain using Discrete Fourier

Transformation Technique (IAFDDFTT) has been proposed to authenticate a gray level PGM,

TIFF image by embedding a message/image where 2 x 2 submatrix is taken as source matrix from

the image matrix and transform into the frequency domain. Two bits of authenticating

message/image are fabricated within the real part of each pixel, excluding the 1st pixel of each

submatrix where the position is chosen using a hash function. The process is repeated for each

submatrix on row major order to insert authenticating message/image content and 128 bits

Message Digest (MD-5), generated from authenticating message/image. Inverse DFT is

performed to transform the embedded image from frequency to spatial domain as final step of

encoding. In the second steganographic technique which demonstrates the colour image

authentication process in frequency domain based on the Discrete Fourier Transformation (DFT).

Image authentication is done by hiding secrete message/image into the transformed frequency

component of source image. The DFT is applied on sub-image block called mask of size 2 x 2 in

row major order. Secrete message/image bit is fabricated within the transformed real frequency

component of each source image byte except the first frequency component of each mask. Here a

control technique is applied to minimise the noise integration.

3-bits steganography in frequency domain (3bsf)

This novel data embedding technique in frequency domain has been proposed using Discrete

Fourier Transform (DFT) for image authentication and secured message transmission based on

hiding a large volume of data into gray images. Image authentication is done by embedding

message/image in frequency domain by choosing image blocks of size 2 × 2, called mask, from

the source image in row major order and transform it into the frequency domain using DFT.

Three bits of authenticating message/image/message-digest are fabricated within the real parts of

each source image byte except first frequency component of each mask. The dimension of

authenticating image followed by message digest (MD) and the content of authenticating

message/image are also embedded. Inverse DFT (IDFT) is performed on embedded data to

transform embedded frequency component to spatial component. The proposed IATFDDFT

emphasizes on information and image protection against unauthorized access in frequency

domain to achieve a better trade-off between robustness and perceptibility.

The outlines of frequency domain algorithms are stated in sections 1.4.4, 1.4.5 and 1.4.6. All

techniques are implemented using Discrete Fourier Transform (DFT) for spatial to frequency

domain transformation and Inverse Discrete Fourier Transform (IDFT) for frequency to spatial

domain. Comparative study has been made by calculating MSE, PSNR and IF using equation 1.3,

1.4 and 1.5 respectively. The following equations are used DFT and IDFT for domain

transformation. Equation 1.6 and 1.7 are for DFT and IDFT respectively.

( ) ( )∑ ∑−

=

+−−

=

=1

0

21

0

, 1

,M

x

N

vy

M

uxjN

y

eyxfMN

vuFπ

---------------------- (1.6)

( ) ( )∑ ∑−

=

+−

=

=1

0

21

0

, 1

,M

u

N

vy

M

uxjN

v

evuFMN

yxfπ

---------------------- (1.7)

Where u = 0 to M – 1 and v = 0 to N – 1 and x, y are spatial image variable.

Organization of the thesis

Chapter 2 discusses the technique (BLIA/SMTT) and stated results on 1-bit-steganography on

gray images and mask based colour image authentication technique (MDHIAT) is described in

the chapter 3. Chapter 4 and 5 are discusses the techniques of AI/HLVD and IAHLVDSMTTM

for grayscale and colour image authentication respectively using 2-bit-authentication. 3-bits-

steganography techniques FBIA and ATILD for grayscale and colour images are stated in

chapter 6 and 7 respectively. Chapter 8 discusses the technique IAFDDFTT for grayscale image

authentication using 2-bit-steganography in frequency domain and another authentication

technique IATFDDFT discusses and stated result in chapter 9 using 3-bit-steganography.

Chapter 10 and 11 discusses the technique and results for colour image authentication using 1-

bit-steganography and 2-bit-steganography. In the final and concluding chapter 12 discusses the

application area of all above algorithms.