You are on page 1of 4

2016 Fifth International Conference On Recent Trends in Information Technology

A Novel Spatial Domain Technique for Digital Image


Watermarking Using Block Entropy
Sanjay Kumar

Ambar Dutta

Department of Computer Science


Birla Institute of Technology, Mesra, Kolkata Campus,
Kolkata, India
sanjaykumar.spj@gmail.com

Department of Computer Science


Birla Institute of Technology, Mesra, Kolkata Campus,
Kolkata, India
adutta@bitmesra.ac.in

AbstractDuring last few years for copyright protection,


security and data authentication of digital media, digital
watermarking has been raised as the one of the burning research
topics due to the rapid expansion of Internet. This paper presents
a new technique for image watermarking in the spatial domain
where the concept of information theory is utilized with the
popular LSB substation technique. Here, the cover image is
segregated into a number of blocks and the watermark is
embedded into the block(s) with the maximum entropy value.
The extraction algorithm is also able to find the watermark
correctly. The proposed algorithm was evaluated with the help of
various standard performance measures like Mean Square Error
(MSE), Peak Signal to Noise Ratio (PSNR) and Structural
Similarity (SSIM) measure to verify the perceptibility and the
robustness of the algorithm. Experimental resultsdemonstrate
that the improved algorithm performs reasonably well over a
large varied datasets of cover and watermark images.
Keywords
Digital
Performance Measures

Watermarking;

Entropy;

Blind and Non-blind, (ii) Perceptible and Imperceptible, (iii)


Private and public, (iv) Robust and Fragile, and (v) Spatial and
Frequencydomain methods. On the basis of application
watermarking techniques can further be classified into the
following five classes (i) Copyright Protection Watermarks,
(ii) Data Authentication Watermarks, (iii) Fingerprint
Watermarks, (iv) Copy Control Watermarks and (v) Device
Control Watermarks.
WI

CI

Watermark
Embedder

Attack

Communication
Channel

LSB;

I. INTRODUCTION
Due to the expeditious and tremendous evolutionof
multimedia and the outspread use of the Internet, there is a
need for efficient and effective copyright protection
approaches [1, 2, 3, 4].Techniques are required to counter the
copying,falsification and unauthorized handling of images and
video. In the absenceof such methods, storingimages on a
public network place them at danger of theft andundiscovered
alteration. Cryptography, Steganography and Digital
Watermarking are the approaches which have been advanced
to defend information during the transfer of data from sources
to destination.With the help of digital watermarking, it is
possible to mask the data or identify information inner to the
digital multimedia [1, 3].Digital Watermarking techniques
include two principal steps-(i)watermark embedding, (ii)
watermark extraction.A watermarked imaged is ideal if it is
identical from the original image even if one uses maximal
quality equipment.Integrity and Security are also two crucial
requirements of ideal watermarking.
Recently, the large number of algorithm has been
developed in the field of watermarking.So it is necessary to
categorizethese algorithms for their proper understanding. In
this paper the watermarking techniques has been classified on
the basis oftheir application and characteristics [1, 3, 4].On the
basis of characteristics, watermarking can be divided as (i)
978-1-4673-9802-2/16/$31.00 2016 IEEE

Noise

Watermark

Watermark
Detector

Figure 1. Basic Watermarking Model


Image watermarking can be explained with the help of a
simplified model. The model consists of a cover image (CI),
watermark image (WI), watermark embedder, communication
channel, noise, several attacks, watermark detector and
recovered watermarks. Attacks in image watermarking can be
intentional or unintentional. The watermark image (WI) is
embedded into the cover image (CI) by the watermark
embedded, while the presence of the watermark is detected by
the watermark detector. Digital watermarking techniques must
be robust in order to endure various types of noise and
security attacks in the communication channel.
In the spatial domain digital image watermarking algorithms,
the watermark is directly added to the cover image simply by
manipulating the pixels or image characteristics. There are
different types of techniques available in the spatial domain,
some of which are described below. One of the simplest
methods to insert a watermarkin spatial domain is to add
pattern of pseudo-random noise to the gray values of image
pixels. In LSB coding based method, the LSB of the host

2016 Fifth International Conference On Recent Trends in Information Technology

image is replaced with the watermark. These bits are embedded


in the sequence which will work as the key. In order to extract
the watermark this sequence must be known. In spite of its
simplicity of LSB coding based technique,its robustness with
respect
to
noise
is
low.
In
spread-spectrum
techniques,produced energy at one or more discrete
frequencies is intentionally distributed in time.In correlationbased method, a pseudo-random noise is added to an image
with a pattern. At the decoder, the correlation between the
noise and image is calculated and if its value is found to exceed
a certain threshold, then the watermark is detected.
A. Entropy of Gray Scale Image
Entropy is a statistical measure which is used to find the
texture content in images [5, 6]. It measuresthe randomness to
describe the composition of an image. Entropy (E) is defined
as:

E = pi log 2 pi
i

where pidenotes the probability associated with the i-th


intensity level. Maximum entropy is obtained at the locations
with uniform distribution of intensities, whereas the areas of
the image with almost constant intensities have low entropy.
Human eye is less sensitive to high entropy areas.
B. Performance Measures
In order to evaluate the quality of the watermarked image and
extracted watermark, common performance measures are
Mean Square Error (MSE), Mean Absolute Error (MAE),
Peak Signal to Noise Ratio (PSNR), Structural Similarity
Measure (SSIM)etc[1, 2, 6]. MSE and MAE is one of the
widely used performance measures that makes comparison
between the two images on Pixel-by-Pixel ground.
2

1 M N
[C (m, n) W (m, n)]
MN m=1 n=1
1 M N
MAE (W , C ) =
C (m, n) W (m, n)
MN m=1 n=1

MSE (W , C ) =

where W is Watermarked image and C is Cover image, M and


N are number of rows and column of the image.
PSNR is one of the other significant performance measures
used in finding the distortion between the images.
255 X 255
PSNR = 10 log10
MSE
Lower MSE, MAE and higher PSNR values are highly
desirable.
SSIMis one of the commonly used metrics to measure the
similarity between the two images. The structural similarity
between two images x and y is calculated as
(2 x y + C1)* (2Cov( x, y) + C 2)
SSIM ( x, y ) =

2
x

)(

+ y + C1 * x + y + C 2

wherex, y denote the mean intensities of x and y; x, y


denote the standard deviation of the intensity values of x and y
and Cov(x, y) denotes the covariance of xand y. In addition,
variables C1 and C2 are used to stabilize the division in case of
weak denominator. SSIM value closer to 0 indicates better

quality of watermark image. Whereas MSE or PSNR measures


the perceived quality degradation in watermarking, SSIM
measures any perceived change in structural information.
The structure of the rest of the paper is given below. Section II
deals with improved and proposed algorithm for image
watermarking in the spatial domain using the concept of block
entropy discussed in section I and LSB substitution.
Experimental results are provided in section III with the help
different performance measures using various test data sets.
Finally, conclusion is made in section IV.
II. RELATED WORK
Least Significant Bit (LSB) is one of the elementary techniques
for digital watermarking. Wang et. al. [7] proposed an
elementary algorithm to inject fundamental data in the cover
image such that presence of the data in the image will not
noticed by the interceptor. To obtain the high quality result
they used the genetic algorithm to hide the watermark in the
right most k-LSB of the cover image. For colour image
Dharwadkar et. al.[8] proposed a spatial domain watermarking
algorithm .They stored number of bits in each pixel in variable
based on the original colour value of the pixels. In this method
the permutation of the bits of watermark and use of secret key
safeguard the security. Roduane et. al.[5] has developed the
algorithm of watermarking based on the embedding the
watermark into the blocks of the image with LSB-techniques.
In this method the entropy of the each block is calculated and
the watermark is inserted into the block which is having the
highest entropy value using LSB techniques. By, using the
LSB technique Singh et. al[9] proposed an algorithm. In this
method, they used the binary value of watermark text in LSB
and in place of the 2nd LSB the inverse of their corresponding
LSB bit. Due to the use of 2nd LSB, this method has given the
better authenticity.

III. PROPOSED ALGORITHM


The proposed algorithm is designed for the image
authentication and the copyright protection of the image based
on the combination of entropy,block,and LSB substitution
concepts.In this proposed algorithm the LSB substitution is
performed in the different bits of images for the watermark
embedding and watermark extraction. At the time of
embedding the watermark into the cover image, the block with
highest entropy is chosen in order to obtain higher
imperceptibility.
A. Embedding
1. The cover image (CI) of size MXN is first divided
into M/8 X N/8 blocks.
2. The entropy is calculated of each of the 64 blocks.
3. The watermark image(WI) of size pXq is resized to
M/8 X N/8.
4. WI is inserted in that block(s) of CI with highest
entropy by using LSB substitution method where n

2016 Fifth International Conference On Recent Trends in Information Technology

least significant bits are set to zero in the block with


maximum entropy and it is followed by right shift to
the WI by (8 n) bits.
B. Extraction
1. Do left shift to the block in a watermarked image by
(8 n) bits.
2. The recovered watermark image is resized to the size
pXq.

Dataset
3

IV. EXPERIMENTAL RESULTS


In this paper, to investigate the performance of the improved
algorithm, experimentation was conducted with the help of
large number of datasets containing cover images and
watermark images. For the sake of this paper, the results are
provided with the help of three pair of images (cover and
watermark) only.Here in this paper some of the commonly
used performancemeasuresi.e. Mean Square Error (MSE) Peak
Signal-to noise Ratio (PSNR) and Structural Similarity
(SSIM) has been used.SSIM(1) is the structural similarity
between the cover images and the watermarked images for the
different LSB bits whereas the SSIM(2) is structural
similaritybetween originalwatermarks and the recovered
watermarks for different LSBs.
Table 1: Datasets used in the paper
Cover Image

Watermark

Table 2: Performance Measures of for Dataset 1


MSE

PSNR

SSIM(1)

SSIM (2)

1stBit
Substitution

0.0078

69.2098

0.9999

0.3148

2nd
Bit
Substitution

0.0439

61.7061

0.9994

0.4287

3rd
Bit
Substitution

0.2023

55.0708

0.9980

0.4539

4th
Bit
Substitution

0.8987

48.5946

0.9945

0.4382

5th
Bit
Substitution

4.0991

42.0039

0.9893

0.3130

6th
Bit
Substitution

17.3389

35.7405

0.9847

0.2961

7th
Bit
Substitution

49.2443

31.2072

0.9837

0.2916

Dataset
1

Table 3. Performance Measures of for Dataset 2


MSE

PSNR

SSIM (1)

SSIM (2)

1 Bit
Substitution

0.0079

69.1545

0.9998

0.0909

2nd
Bit
Substitution

0.0311

63.2031

0.9996

0.2428

3rd
Bit
Substitution

0.1188

57.3826

0.9990

0.4339

4th
Bit
Substitution

0.4903

51.2261

0.9978

0.6213

5th
Bit
Substitution

2.0067

45.1059

0.9956

0.7602

6th
Bit
Substitution

7.4669

39.3994

0.9922

0.8316

7th
Bit
Substitution

27.5303

33.7326

0.9879

0.8556

st

Dataset
2

2016 Fifth International Conference On Recent Trends in Information Technology

Table 4: Performance Measures of for Dataset 3


MSE

PSNR

SSIM (1)

SSIM (2)

1 Bit
Substitution

0.0078

69.2377

0.9999

0.3155

2nd
Bit
Substitution

0.0292

63.4730

0.9995

0.7037

3rd
Bit
Substitution

0.1173

57.4365

0.9986

0.8554

4th
Bit
Substitution

0.4436

51.6610

0.9967

0.9111

5th
Bit
Substitution

1.7193

45.7773

0.9938

0.9304

6th
Bit
Substitution

7.4077

39.4339

0.9901

0.9370

st

th

7
Bit
Substitution

block based spatial domain digital watermarking technique is


proposed in which aninformation theoretic measure is used to
determine the location(s) in the cover image to which
watermark is to be embedded. The cover image is first divided
into a number of blocks of equal size and then entropy is
calculated for each of the blocks. The watermark is embedded
into that block(s) of the cover image which shows the highest
entropy. The watermark is extracted from the embedded
image with the help of LSB substitution technique. The
experimental results show the efficiency of the proposed
algorithm with the help standard performance measures using
a varied image sets. It is found from the literature that a digital
image watermarking scheme should be robust with respect to
a variety of possible attacks. Addressing different geometric
attack issues will be the future direction of research of the
present work.
REFERENCES
[1]

19.9327

35.1352

0.9877

0.9389
[2]

[3]
[4]

[5]

Figure 2: Recovered Watermarks


In the dataset 1, 2 and 3 (see Table 1), the size of the cover
images
are760X960,
3872X2592
and
2048X1536
respectively, and size of the watermarks are 622X624,
512X512 and 323X324 respectively. The experiment was
performed on a computer with 4GB RAM, Intel i5 processor
and 2.1 GHz clock speed. It is found, from the experimental
results provided in Table 2, 3 and 4, that the MSE and PSNR
values for the cover image and watermarked image are within
the allowable range (standard for 8-bit images are 30 50 dB
for PSNR, however higher values are more acceptable). The
extracted watermarks are also shown in Figure 2.
V. CONCLUSIONS
Digital watermarking is a technique to hide or embed any kind
of digital data in another multimedia data. In this paper, a

[6]

[7]

[8]

[9]

S. Kumar and A.Dutta, Implementation of a Least Significant Bit


(LSB) based Spatial Domain Watermarking Techniques, International
Journal of Advanced Research in Computer Science and Software
Engineering, Volume 5,Issue 12, December 2015.
D. Chopra, P. Gupta, G. Sanjay, A. Gupta, Lsb Based Digital Image
Watermarking For Gray Scale Image, IOSR Journal of Computer
Engineering, Volume 6, Issue 1,(Sep-Oct,2012).
F.Y Shih, Digital Watermarking and Stegnography:Fundamental and
Techniques, CRC Press, First Indian Reprint, 2012.
I. J. Cox, M.Miller, J. Bloon, J. Fridrich, T. Klaker,Digital
Watermarking and Steganography, Morgan Kaufman Publication,
Second edition, 2011.
M. Radounane, T.Boujiha, R. Messsoussi, N.Idrissi A method of LSB
substitution based on image blocks and maximum entropy ,
International Journal of Computer Science Issues, VOL 10, Issue 1, No
1, January 2013.
A.Gawri and S.Singh, A Bit Level and Entropy based Blind
Watermarking Scheme for Video Sequence using Random Frame
Selection, International Journal of Computer Applications(09758887),Volume 64- No.12, February 2013.
R. Z. Wang, Chi-Fang Lin, Ja-Chen Lin Image hiding by optimal
substitution and genetic algorithm. Pattern Recognition 34 (2001) 671683.
N. Dharwadkar and B.B Amberker Secure Watermarking Scheme for
Color Image Using Intensity of Pixel and LSB Substitution Journal of
Computing, Volume 1,Issue 1, December 2009.
I. Singh, S. Jain and A. Jain Digital Watermarking method using
Replacement of Second Least Significant Bit (LSB) with inverse of
LSB. ISSN 2250-2459, ISO 9001:2008 Certified Journal , Volume 3,
Issue 2, February 2013

You might also like