Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

CHAPTER V
IMAGE ENCRYPTION USING FUZZY SETS THEORY

5.1. Introduction

The soft computing algorithms are the fusion of paradigms designed to present
solutions to the real world problems, which are not modelled or very difficult to be
modelled mathematically. They are consortium of methodologies that work
synergistically and provide in one form or another, flexible information and
processing capability for handling real-life ambiguous situations. [148, 149] Their
aim is to exploit the tolerance for imprecision, uncertainty, logic and partial truth in
order to achieve close resemblance with human-like decision making. Soft computing
is basically optimization technique to find solutions for problems which are very hard
to solve.
Fuzzy sets theory is one of the most important branches of soft computing
algorithms. Fuzzy sets were introduced by Lotfi A. Zadeh in 1965 to represent data
and information possessing non-statistical uncertainties. It has been developed
through various ways and in many disciplines. Fuzzy logic is an approach related to
the fuzzy sets theory of soft computing algorithms based on “degree of truth” rather
than the usual “true of false” (1 or 0) and Boolean logic on which the modern
computer is based. This chapter presents two new algorithms for image encryption
which use the fuzzy set theory to provide high protection to the images data from
illegal intrusions. It is fast in the process of encryption and decryption. The
decryption process does not induce any loss of image data, and it has ability of
dealing with different format of images as well. The first algorithm based on fuzzy
logic to provide security levels and their corresponding processing levels by
generating random sequences for the encryption/decryption process. The second
algorithm based on fuzzy graph to obtain highly encrypted image.

106
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

5.2. Basics Concepts of Fuzzy Sets Theory

The fuzzy sets theory can be defined as the theory which is related to the classes of
objects with un-sharp boundaries in which membership is a matter of degree. Since
1992, the fuzzy set theory is known as the theory of neural nets and the field of
evolutionary programming have become one of the important areas of ‘computational
intelligence’ or ‘soft computing’. The strong relationship between these areas has
naturally become close. In this section, however, the focus primarily will be on the
notion of fuzzy set theory, fuzzy logic and fuzzy graph. The notion of a fuzzy set
stems from the observation made by Zadeh (1965a) that “more often than not, the
classes of objects encountered in the real physical world don’t have precisely defined
criteria of membership”. [150-152] Fuzzy logic is brilliant engineering and data
analysis approach proposed by Lotfi A. Zadeh. This section deals with the basic
concept of fuzzy logic, and it includes some basic examples of fuzzy logic. Fuzzy
logic is an approach based on ‘degrees of truth’ on which the modern computer is
based. In addition, the fuzzy logic has two different meanings; one is wide and the
other is narrow.
In the wide sense, fuzzy logic (FL) is almost equivalent with the theory of fuzzy
sets. On the other hand, the narrow sense shows fuzzy logic as a logical system,
which is an extension of multi-valued logic. In this perspective, fuzzy logic in its
narrow sense is a branch of FL. Even in its more narrow definition, fuzzy logic differs
both in concept and nature from traditional multi-valued logical systems. [153]
Fuzzy graph theory was introduced by Azriel Rosenfeld in 1975. Though he was
very young, he has been growing fast and has numerous applications in various fields
[154]. Later, Bhattacharya [155] gave some remarks on fuzzy graphs. Mordeson and
Peng introduced some operations on fuzzy graphs. A new concept namely, graph
structure was introduced by E. Sampathkumar [156].
Definition: A fuzzy graph is a pair where a fuzzy subset of S is, is a
symmetric fuzzy relation on . The elements of S are called the vertices or nodes of G
and the pair of vertices as edges in G. The underlying crisp graph of the fuzzy graph
is denoted as G*: (S, E) where . The crisp graph (S, E) is a special
case of the fuzzy graph G with each vertex and edge of (S, E) having degree of
membership.

107
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

5.3. Mechanism of Fuzzy Logic

The use of fuzzy logic needs at least two input magnitudes. If they are more, they
would be better for fuzzy logic. Three basic steps are needed for applying fuzzy
analysis:

 Fuzzification of the inputs

 Fuzzy-Inference
 Defuzzification of obtaining crisp output

For fuzzification, a fuzzy set is required for each input. This set is a kind of
linguistic variables such as relative terms: far, close, small, normal, high, cold, warm,
hot, etc. Once fuzzy set is defined for each of the variables, the degree of membership
should be calculated for every input. Minimum degree of membership is zero, and
maximum is 1 or 100%. To calculate degree of membership, it is easy to use linear
functions as presented in figure 5.1 bellow. [157]

Figure 5.1 Calculation degree of membership using linear functions

However, calculation of the degree of membership is not limited to linear functions

only. Gaussian or some arbitrary functions can be used as it is shown in figure 5.2
bellow. In that sense, the number of possibilities is theoretically infinite.

108
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

Figure 5.2 Calculation degree of membership using Gaussian function

After calculating the degrees of membership for every input and every fuzzy set, the
step of (fuzzy-inference) should be started. All numerical data obtained in
fuzzification process need to be mutually compared in ‘if-then’ fashion. This process
is called fuzzy-inference. Overall number of ‘if-then’ rules is equal to the outputs of
fuzzy set numbers for each input. For example, if we have two input values;
input X has three fuzzy sets (negative, optimal and positive), and input Y has five
fuzzy sets (negative, negative-optimal, optimal, optimal-positive and positive), the
(15) ‘if-then’ rules are required to compare negative degree of membership of X with
negative degree of membership of Y, negative degree of membership of X with
negative-optimal degree of membership of Y and so forth. In fact, there is a need for
comparing at least two numbers in every ‘if-then’ rule, but there are a lot of numbers
and ‘if-then’ rules to be compared. There are various options to show how to
determine output from compared numbers in each ‘if-then’ rule. Options are as
follows:

 Choose minimum: μ=min{X, Y, Z}

 Choose maximum: μ=max{X, Y, Z}
 Choose complement μC =1-μ

Defuzzification is used for obtaining crisp output from numbers obtained in step
fuzzy-inference. In order to calculate crisp output, we can use “centre of gravity” or
some other method. Together with numbers from fuzzy-inference, we need to have
additional parameters called “singleton values”. Singleton values can be achieved
from experiments.

109
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

5.4. Application of Fuzzy Logic

Fuzzy logic addresses such applications perfectly as it resembles human decision
making with an ability to generate precise solutions from certain or approximate
information. [158-160] It fills an important gap in engineering design methods left
vacant by purely mathematical approaches (e.g. linear control design), and purely
logic-based approaches (e.g. expert systems) in system design. Fuzzy logic can be
applied in various fields.

5.4.1. Fuzzy Logic in Image Processing

Fuzzy logic has rapidly become one of the most successful technologies for
developing sophisticated image processing because of its simple use. The applying of
fuzzy modelling to image processing is not new, a fact that Pal and King. [161] were
quick to realize. It has been used to enhance digital image and signals processing. The
Canon camera’s autofocus system is the best example of the efficient implementation
of fuzzy technology. The autofocus, auto zooms and auto exposure systems for
Minolta cameras use fuzzy controllers. Using fuzzy logic techniques, Sanyo and
Canon camcorders are better in auto-white balancing system, auto exposure and
autofocus system. Image stabilizer for camcorder of Matsushita has been
implemented with fuzzy technology. Fuzzy inference was also used to improve image
quality. Because of Fuzzy system technology, electro photography process of
photocopying machine has been improved. The image quality of Sanyo copies has
been improved by better toner supply control based on fuzzy control. Matsushita
copies are also improved by better auto exposure fuzzy based control. Some other
successful applications are hand written language recognition and voice recognition.

5.4.2. Fuzzy Logic in Control of Complex System

The aim of the application is to establish proper protection as well as to minimize

the power used to protect the pipeline. Fuzzy control used as modelling the different
soil conditions along the pipeline would be complex for a practical control system.
The fuzzy control model was made with 3 anodes followed by 126 rules, which
resulted in the controller having adequate performance in maintaining protection.
[162]

110
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

5.4.3. Fuzzy Logic in Medical Diagnosis

Fuzzy methods are used in the automation of the medical diagnosis application.
Using these methods, the medical result yield contains likelihood estimation rather
than confirmation of presence or absence of disease. In practice, the
recommendations for action suggested in guidelines are vague and conflicting. But
these difficulties are well handled by the fuzzy methods [163, 164].

5.4.4. Fuzzy Logic in Automotive Industry

Most of the automotive manufacturers are perusing fuzzy control concepts. Fuzzy
control has been applied to control automatic transmission system, suspension
system, engine system, climate system and antilock brake system. These systems are
used to make the vehicle better, more efficient and safer to ride. [165]

5.4.5. Fuzzy Logic in Healthcare Industry

Fuzzy system technology has larger impact on healthcare industry. The biomedical
applications are less due to inherent complexity and uncertainty of the systems as
well as the risk involved. In biomedicine science, human knowledge, skills and
experience are more important in diagnosis and treatment of diseases. This bio -
medical system is difficult to be managed because it has time to delay. World’s first
fuzzy control in medicine is drug delivery system, which was developed and
implemented to regulate blood pressure in post-surgical open heart patients at
cardinal intensive care unit (ICU). Some other applications include determining the
disease risk using fuzzy expert systems, medical control system and determination of
drug dose. Classification of tissue and structure in electrocardiograms, classification
of normal and cancerous tissues in brain magnetic resonance image are the other
successful applications of fuzzy systems in medicine [166, 167].

5.4.6. Fuzzy Logic in Data Mining

Fuzzy logic is applied at various stages of Data Mining to impart human type
reasoning in decision making. The human behaviour is always fuzzy in nature and it
is very difficult to model human behaviour patterns with crisp Boolean Data Mining
systems.

111
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

5.5. Properties of Fuzzy Logic

Some of the essential characteristics of fuzzy logic are related to the following
properties:
• In fuzzy logic, exact reasoning is viewed as a limiting case of approximate
reasoning.
• In fuzzy logic, everything is a matter of degree.
• In fuzzy logic, knowledge is interpreted a collection of elastic or, equivalently,
fuzzy constraint on a collection of variables.
• Inference is viewed as a process of propagation of elastic constraints.
• Any logical system can be fuzzified.
There are two main characteristics of fuzzy systems that give them better
performance for specific applications.
• Fuzzy systems are suitable for uncertain or approximate reasoning, especially for
the system with a mathematical model that is difficult to derive.
• Fuzzy logic allows decision making with estimated values under incomplete or
uncertain information.
Elements of a fuzzy set may belong to the set, may not belong to the set, or may
belong to a degree.
Membership of a crisp set is described by a bivalent condition; the membership of a
fuzzy set is described by a multivalent condition.
Fuzzy sets are used to deal with uncertainty on one hand and to represent knowledge
on the other. [168]

112
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

5.6. Proposed Algorithms Based on Fuzzy Sets Theory

5.6.1. Fuzzy Logic Algorithm

In this approach, a new algorithm is proposed to encrypt an image using the concept
of fuzzy logic based on fuzzy pseudorandom number PN bit generator. This
algorithm provides more secure image encryption, minimum loss of data, maximum
speed and maximum distortion in encryption, with the addition of extra degree of
protection.
The Steps of Encrypting an Image Using Fuzzy Logic Algorithm

1. Loading an image.
2. Determining the size of image (height and width).
2.1. Checking the result of (H mod 48 ) and ( W mod 48 ).If they are equal to zero
then go to 3rd step otherwise do

H =H+(48- (H mod 48)) and W = W+[48- (W mod 48)]. (5.1)

3. Dividing the image into a set of blocks, each block size is (48).
4. Doing pixels permutation procedure.
4.1. The permutation process for 48 successive pixels as:

4.2. The permutation procedure is as follow:

Start:

End

113
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

5. Encryption process for RGB

5.1. Encryption Process for the RED colour value:
The three sequences will be used in this process, where will
determine the direction of the R binary value rotation.
If
If
Number of rotated position of R will be determined according to:
(5.2)
The resulting rotated R will be bitwise XORed with the decimal value
of
5.2. Encryption Process for the GREEN colour value:
The value of will determine the direction of the G binary value
rotation.
If
If
Number of rotated position of G will be determined according to:
(5.3)
The resulting rotated G will be bitwise XORed with the decimal value
of
5.3. Encryption Process for the BLUE colour value:
The value of will determine the direction of the B binary value rotation.
If
If
Number of rotated position of B will be determined according to:
(5.4)
The resulting rotated B will be bitwise XORed with the decimal value of
6. Repeating the 4th and 5th steps for all blocks
7. Getting encrypted block
8. Getting encrypted image

114
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

5.6.2. Fuzzy Graph

In this approach, a new algorithm is proposed to encrypt an image using the concept
of fuzzy graphs obtained from a matrix of image’s pixels, and then they are used to
encrypt an image. This algorithm provides more secure image encryption, minimum
loss of data, maximum speed and maximum distortion in encryption with the addition
of extra degree of protection.
The Steps of Encrypting an Image Using Fuzzy Graph Algorithm

1. Loading an image.

2. Determining the size of image (height and width).

2.1. Checking the result of (H mod 5 ) and ( W mod 5 ).If they are equal to zero
then go to 3rd step otherwise do

H = H+(5- (H mod 5)) and W = W+[5- (W mod 5)]. (5.5)

3. Dividing the image into a set of blocks, each block size is (8×8).

Considering a block B(w × h) where w and h are width and height of B.

4. Extracting the RGB components and store them as R, G and B matrix

R = R(r * c), G = G(r * c), B= R(r * c). (5.6)

4.1. Finding the membership for each pixel as: P(i,j)/255.

5. Doing interval-valued fuzzy graphs for each block as:

(5.7)

Where X and Y are an interval-valued fuzzy set, are the pixels in the
block, PR, PG and PB are the pixels of RED, GREEN and BLUE components in each
block, the membership of pixels in each block and
is the membership of RED, GREEN and BLUE components which

115
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

are calculated by dividing the summation of pixels for each component dividing by
number of pixels in each block.
6. Obtaining the fuzzy graph for each component in each block.
6.1. Considering the vertices of block as the entries of matrix and connecting the adjacent
edge to make grid graph like graph.
6.2. Applying the sigmoid function as: sig (X; Y,C) = (5.8)

Where X and Y are an interval-valued fuzzy set and C is the center of graph

7. Encrypting the block:

. (5.9)

8. Repeating 4th - 7th steps for all blocks to get encrypted image

5.7. Results and Discussion

In this section, several experiments are performed to test the effect of the proposed
algorithms on image encryption. The proposed approaches are implemented and
applied to three images, which have different types and sizes, to verify its
adaptability, quality, security and speed. The results are illustrated in this section to
confirm the properties of proposed approaches based on genetic algorithm and to
compare them with the quality of image encryption. The security analysis performed
in this approaches is visual analysis, histogram analysis, correlation coefficient. All
these experiments prove that the proposed approaches in this chapter fulfil all the
results completely and rightly. In this subsection, the discussions and analyses of the
encryption algorithms based on fuzzy logic and fuzzy graph are presented. The
following discussions and analyses are performed on original, encrypted and
decrypted images. In these experiments, the algorithm parameters are set as follows:
the images sizes are different; the length of images blocks is (48×48) pixels in fuzzy
logic and (8×8) pixels in fuzzy graph. Image encryption and decryption tests have
been carried out using standard images of different sizes in gray scale and colour.

116
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

Encrypted and decrypted outputs have been obtained from fuzzy logic and fuzzy
graph algorithms. It may be noted that the encrypted image obtained from fuzzy logic
and fuzzy graph algorithms does not resemble the original image.

Figure 5.3 Encryption and decryption of Image ‘Car’ by fuzzy logic and fuzzy graph

5.7.1. Visual Analysis:

The purpose of visual testing is to highlight the presence of the similarities between
plain image and its cipher. Figure 5.3 shows that the encrypted images do not contain
any features of the plain images. After the visual testing had been performed on some
images, which have different sizes and formats, it showcased that there is no
perceptual similarity.

5.7.2. Statistical Analysis

A statistical analysis of encrypted images provides much information about the
security of a cipher with reference to statistical attacks that could be launched. This is
shown by a test performed on the histograms of the enciphered images and on the
correlations of the adjacent pixels in the ciphered image.

[Link]. Histogram Analysis

The histogram analysis is used to illustrate the confusion and diffusion properties in
the encrypted image for testing purposes. The histogram of the plain image ’Lena’
and that of the encrypted image by using proposed approaches are shown in figures

117
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

5.4 – 5.7. While comparing the two, the histogram of the encrypted image is fairly
uniform and is significantly different from that of the original image. The encrypted
images transmitted are not affected by any attacker.

Figure 5.4 Original Image of Lena and Its RGB Histogram

Figure 5.5 Encrypted Image of Lena and Its RGB Histogram by fuzzy logic

118
Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

Figure 5.6 Original Image of Lena and Its RGB Histogram

Figure5.7 Encrypted Image of Lena and Its RGB Histogram by fuzzy graph

119
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

[Link]. Correlation Coefficient Analysis

The correlation coefficient displays the relationship between two neighbouring
pixels. If correlation between two pixels is nearly 1, the image Pixels are highly
correlated, but if it is nearly 0, the image pixels are highly uncorrelated. In the
proposed method, the experiments on the images prove that the original image
correlation is nearly one, and cipher image correlation is nearly zero. So, this saves
the system from statistical attacks. Tables 5.1 and 5.2 show the results of
experiments. The formula of correlation coefficient of two adjacent pixels is as
follows:

(5.10)

Where x and y are the values of two adjacent pixels in the image and N is the total
number of adjacent pixels selected from the image.
The equation (3.5) is used to calculate the correlation coefficients in the horizontal,
vertical and diagonal planes of the image. The calculated results are listed in Tables
5.1 and 5.2.

Table 5.1 Adjacent pixels correlation coefficients analysis by fuzzy logic

Direction Plain image of Lena Cipher image of Lena

Red Green Blue Red Green Blue
Horizontal 0.9878 0.9860 0.9869 0.0134 0.0151 0.0156

Vertical 0.9868 0.9897 0.9917 0.0288 0.0257 0.0221

Diagonal 0.9871 0.9795 0.9726 0.0212 0.0196 0.0196

Table 5.2 Adjacent pixels correlation coefficients analysis by fuzzy graph

Direction Plain image of Lena Cipher image of Lena

Red Green Blue Red Green Blue
Horizontal 0.9785 0.9754 0.9772 0.0264 0.0249 0.0244

Vertical 0.9811 0.9764 0.9838 0.0399 0.0353 0.0318

Diagonal 0.9763 0.9645 0.9652 0.0342 0.0274 0.0251

120
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

5.7.3. Sensibility Analysis

[Link]. NPCR and UACI Analysis

The NPCR is measures the change rate of the number of the cipher-image pixels
when only one pixel of the plain-image is modified. The (UACI) means unified
average changing intensity. Index measures the average intensity of differences
between two images. Let us assume two ciphered images and whose
corresponding plain images have only one-pixel difference. The gray-level values of
ciphered images and at row i, column j are labeled as and
respectively. NPCR and UACI are defined as:

(5.11)

(5.12)

Where W and H are the width and height of the image. In this example, the selected
pixel is the last pixel of the plain-image. The NPCR and UACI at previous
approaches are calculated and listed in Tables 5.3and 5.4. The data show that the
performance is satisfactory for the proposed approaches.

Table 5.3 NPCR and UACI by fuzzy logic

NPCR and UACI by fuzzy logic
Lena Image Red Green Blue
NPCR 99.5793 99.6247 99.5990
UACI 0.1178 0.0527 0.0298

Table 5.4 NPCR and UACI by fuzzy graph

NPCR and UACI by fuzzy graph

Lena Image Red Green Blue
NPCR 99.6049 99.6168 99.6049
UACI 0.01689 0.0540 0.0322

121
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

[Link]. Peak Signal to Noise Ratio (PSNR) and Mean Square Error (MSE)
Peak signal to noise ratio (PSNR) and mean square error (MSE) are very useful
measurement of performance the quality of reconstruction of encryption codes (e.g.,
for image encryption). The signal in this case is the original image, and the noise is.
When comparing the plain image, cipher image and decrypted image. PSNR is an
approximation to human perception of reconstruction quality. Although a higher
PSNR generally indicates that the reconstruction is of higher quality PSNR is most
easily defined via the mean square error (MSE) as:

(5.13)

Where the mean square error (MSE) is defined as:

(5.14)

Where M and N are the number of pixels in the frame, I(I,j)and are the
number of pixels in the original and decrypted image, respectively. Greater PSNR
value (>30dB) reveals better image quality. For encrypted image, smaller value of
PSNR is expected. The results of PSNR and MSE for different images by using fuzzy
logic and fuzzy graph algorithm are listed in Tables 5.5 and 5.6.

Table 5.5 PSNR and MSE by fuzzy logic

PSNR and MSE by fuzzy logic
Image Lena Bird Car
PSNR 8.4569 7.8987 6.6568
MSE 3.1830e+03 2.6229e+04 1.3579+04

Table 5.6 PSNR and MSE by fuzzy graph

PSNR and MSE by fuzzy graph
Lena Image Lena Bird Car
PSNR 9.0017 9.3096 7.2987

MSE 3.2765e+03 1.0549e+04 1.0651e+04

122
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

5.7.4. Average Time of Encryption and Decryption

The executing time of encryption and decryption process by proposed approaches
based on genetic algorithms has been implemented and conducted using Matlab-
7.10.0 (R2010a) in AMD V 140 CPU @ 2.30 GHz with Windows-7 operating system.
In this case, three separate images having the different type’s formats of JPG, TIF and
BMP respectively and sizes 225 × 225 pixels, 194 × 259 pixels and 1153 × 2050
pixels have been used to measure the encryption time for each image by using the
suggested approaches. The encryption time obtained using these images are given in
tables 5.7. The average time of encryption is achieved by fuzzy logic and fuzzy graph
while encrypting different images of different sizes and formats. This shows that the
fuzzy graph algorithm is 1.07 times less than fuzzy logic algorithms. The result of
average time is shown in Table 5.7 and Figure 5.8.

Table 5.7 Average time of encryption and decryption by using fuzzy logic and fuzzy graph algorithms
with three images different in size, type and formats
Average time of encryption and decryption by using fuzzy logic and
fuzzy graph algorithms with Three Different Images
Image / Approach Fuzzy Logic Fuzzy Graph
Lena 16.477 14.936
Bird 15.741 14.525
Car 56.645 53.812

Average time of encryption and decryption by using fuzzy logic

and fuzzy graph algorithms
60

50

40
Fuzzy Logic
30
Fuzzy Graph
20

10

0
Lena Bird Car

Figure 5.8 Average time of encryption and decryption images by using fuzzy logic and fuzzy graph
algorithms

123
 Encryption and Decryption Image Using Multiobjective Soft Computing Algorithm

According to the above mentioned figures and tables, the proposed approaches can
categorize the encryption algorithms on the basis of its ability to secure the
confidential data against security attacks and its processing speed.

5.8. Conclusion
This chapter proposed two approaches of image encryption based on fuzzy sets
theory. The first is described in section 5.6.1. The second is described in section 5.6.2.
These approaches have been tested by different analysis measurements and showed
their capability and quality for encrypting and decrypting images as demonstrated in
tables and figures above.

124