A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

Embed Size (px)

Citation preview

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    1/35

    A HYBRID GENETIC ALGORITHM AND

    CHAOTIC FUNCTION MODEL FOR IMAGE

    ENCRYPTION

    SADIQUE NAYEEM,M.Tech.

    Pondicherry University

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    2/35

    Introduction

    A new method based on a hybrid model is proposed for

    image encryption which is composed of a genetic

    algorithm and a chaotic function.

    In the first stage, a number of encrypted images are

    constructed using the original image and the chaotic

    function.

    In the next stage, these encrypted images are used as

    the initial population for the genetic algorithm.

    In each stage of the genetic algorithm, the answer

    obtained from the previous iteration is optimized to

    produce the best-encrypted image ( with the highest

    entropy and the lowestcorrelation coefficient among

    adjacent pixels).

    2

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    3/35

    Terminology

    Genetic algorithm

    Entropy of an image

    Histogram of an image

    Chaotic function Correlation coefficient among adjacent pixels

    3

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    4/35

    Genetic algorithms

    Genetic algorithms which mimic principle of evolution in nature,

    search for final solutions among a population of potential solutions.

    The best solutions are chosen in each generation.

    These solutions, after mating, produce a new set of offspring.

    At the beginning, the initial population is randomly formed, and thefitness function evaluates every population.

    Optimum result is not met, new generations are produced through

    the process of recombination and mutation.

    The more suitable generations are selected to produce newgenerations until the optimum result is obtained.

    4

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    5/35

    Entropy of an image

    Entropy is a statistical measure of randomness that can be used to

    characterize the texture of the input image.

    It is one of the prominent features in randomization and can be

    calculated as:

    where N is the number of gray levels used in the image (for gray level

    images, N is 8), and P(Si) shows the probability of having a ith gray

    level in the image.

    Matlab Code: E = entropy(I) In images that are produced completely randomly, N will be 8, and

    this N is considered as an ideal value.

    5

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    6/35

    Histogram of an image.

    The histogram of an image refers to the frequency of

    any gray level in it.

    This feature is one of the most important statistical

    features of an image.

    6

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    7/35

    Chaotic function

    Chaotic functions are similar to noise signals, but they

    are completely certain; that is, if we have the primary

    quantities and the drawn function, the exact signal can

    be reproduced.

    Advantages:

    Sensitivity to primary conditions

    Apparently accidental feature

    Deterministic work

    7

    Logistic Map Signal

    r=3.999 (Completely Chaotic)

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    8/35

    Correlation coefficient among adjacent

    pixels

    The correlation coefficient matrix represents the normalized

    measure of the strength of linear relationship between variables.

    The variable can be the two adjacent pixel. Correlation coefficient is

    given by:

    8

    The correlation coefficients range from -1 to 1,

    where

    Values close to 1 suggest that there is a positive

    linear relationship between the data columns.

    Values close to -1 suggest that one column of

    data has a negative linear relationship to another

    column of data.

    Values close to or equal to 0 suggest there is no

    linear relationship between the data columns.

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    9/35

    THE PROPOSED METHOD

    FORMATION OF THE INITIAL

    POPULATION

    GENETIC OPTIMIZATION

    9

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    10/35

    Formation of the initial population

    To form the initial population from the plain-image, first the plain-

    image is divided into four equal parts.

    The chaotic function logistic map is employed to separately

    encrypt all of the pixels present in each of these four parts as

    follows:1. Five pixels are selected as the encryption key for forming the

    initial value of the logical map function and for encrypting that

    part of the image.

    10

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    11/35

    Formation of the initial population

    cont..

    2. The initial value of the chaotic function logistic map is

    determined by taking five pixels as under:

    P = [P1, P2, P3, P4, P5] (gray scale).

    Convert P into an ASCII code: B = [P1,1, P1,2, P1,3, . . .,

    P5,7, P5,8]

    The initial value of the logistic map function(U0k), which lies in the interval 01,is obtained.

    11

    3. For each part of the plain-image, step 2 is repeated.

    Therefore, in the end, there will be 4 different initial values for

    each image (one value for each part of the image).

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    12/35

    Formation of the initial population

    cont

    4. For encrypting the pixels in each part of the image, the initial value

    of that part and Eq. is used as follows:

    The symbol represents XOR.OldValue stands for the current value of the pixel.

    NewValue shows the new value of the pixel after it is encrypted.

    The value ofUik refers to the ith value of the chaotic function in the

    kth part of the original image, determined for each step using Eq.

    All of the pixels in each part, except the five pixels used as the key,

    are sequentially (row-by-row) encrypted in this manner.

    To build the rest of the population, steps one through four are

    repeated.

    12

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    13/35

    First generation of the genetic population

    13

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    14/35

    Genetic optimization

    Crossover:

    After forming the initial population, the genetic algorithm is

    used to optimize the encrypted images.

    (a)&(b) are the input images.

    (c)&(d) are images after crossover.

    14

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    15/35

    Fitness function

    The fitness function used is the correlation coefficient

    between pairs of adjacent pixels of the image.

    At each stage, the new generations are produced andthe previous ones are evaluated using the fitness

    function.

    The generation is required to have the lowest correlationcoefficient for the encrypted image to be as the final

    cipher-image.

    15

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    16/35

    Selection

    50% of the population with the minimum correlation coefficient +

    10% of the remaining population are selected for the next

    generations.

    Non-Elites are not selected.

    Therefore, 10% of the remaining population are chosen arbitrarilyto help the selected elites from 90% to produce new generations.

    This process is repeated until the correlation coefficient of the best

    generation produced does not significantly change in two

    successive stages.

    After this stage, the generation that has the lowest correlationcoefficient is encrypted as the final cipher-image.

    16

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    17/35

    Example

    17

    Plain-image Plain-image divided into four equalparts.

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    18/35

    Example cont..

    A matrix with dimensions 12 12 is used in which the numbers are

    in the interval [0, 7].

    The first five pixels of the first row of each part are used in

    determining the initial value of the chaotic function logistic map.

    The first five pixels of the first part of the image are (3, 1, 7, 6, 2) P=[ 2, 6, 7, 1, 3 ]

    18

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    19/35

    Example cont..

    19

    First generation Second generation

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    20/35

    Example cont..

    20

    Third generation after crossover Forth generation aftercrossover.

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    21/35

    EXPERIMENTAL

    RESULTS21

    1. ANALYSIS OF THE ENTROPY OF THE

    IMAGE

    2. HISTOGRAM ANALYSIS3. ANALYSIS OF THE CORRELATION

    COEFFICIENT

    4. ANALYSIS OF THE FITNESS FUNCTION

    5. KEY ANALYSIS

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    22/35

    Test conducted

    Gray level images with dimensions of 256 256.

    The initial number of the population for GA = 128.

    The crossover operation rate = 90%

    Mutation operation rate = Zero.

    22

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    23/35

    1. ANALYSIS OF THE ENTROPY OF THE

    IMAGE

    Entropy is a

    statistical measure of

    randomness of the

    pixels

    23

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    24/35

    2. HISTOGRAM ANALYSIS

    Fig. 8. (a) Plain-image, (b) cipher-image, (c) histogram of plain-image, and (d)

    histogram of cipher-image.

    24

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    25/35

    2. HISTOGRAM ANALYSIS

    CONT25

    Fig. 9. (a) Plain-image, (b) histogram of plain-image, histogram of image after, (c) 2nd iteration, (d) 25th iteration, and

    (e) 70th iteration.

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    26/35

    3. ANALYSIS OF THE CORRELATION

    COEFFICIENT

    In a good encryption algorithm, the correlation coefficient between

    pairs of encrypted adjacent pixels in the horizontal, vertical, and

    diagonal positions are as small as possible.

    To test the correlation coefficient between two adjacent vertical

    pixels, two adjacent horizontal pixels, and two adjacent diagonal

    pixels in a cipher-image, the following procedure was used: first,

    2500 pairs of pixels were randomly selected. Then, the

    correlation coefficients were obtained using Eq.

    26

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    27/35

    3. ANALYSIS OF THE CORRELATION

    COEFFICIENT27

    Fig:(a) Correlation analysis of plain-image and (b) correlation analysis of

    cipherimage.

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    28/35

    3. ANALYSIS OF THE CORRELATION

    COEFFICIENT

    In another test, the results of the correlation coefficients of two

    adjacent pixels in the vertical, horizontal, and diagonal positions

    were investigated in each of the different iterations.

    The results obtained for the correlation coefficients in this method

    are far better than those of the methods of [22,24,25]

    28

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    29/35

    4. ANALYSIS OF THE FITNESS

    FUNCTION

    Selection of the fitness function in genetic algorithms is one of themost important and most influential elements for achieving the

    desired result.

    Tests conducted showed that the correlation coefficient is a

    better optimizerthan the function of the entropy of the image.

    As observed in this table, when the correlation coefficient is used as

    the fitness function, the degree of entropy and the correlation

    coefficient of the image are better than when entropy is used as the

    fitness function.

    29

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    30/35

    5. KEY ANALYSIS

    For a good encryption algorithm, key must be long

    enough to resist brute-force attacks.

    A 40-bit key is suggested which produces a key space

    equivalent to 2^40.

    30

    Fig:

    (a) Plain-image,

    (b and c) encrypted images using

    user keys with 1-bit difference in

    each part,

    (d) the similarity between (b) and(c), (99.76% different.)

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    31/35

    KEY ANALYSIS CONT.

    31

    Fig: Percentage of difference between encrypted images versus

    iterations.

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    32/35

    Decryption

    The decryption of encrypted images is possible if the

    chaotic function logistic map and the numbers of the

    population (or the numbers of the two populations) from

    which the final image is composed are known.

    32

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    33/35

    Conclusions

    A new method has been suggested for encrypting images with a

    chaotic function logistic map and a genetic algorithm.

    In this method, the chaotic function is employed for initial

    encryption, and the genetic algorithm is used to improve the

    encryption process of the image. The main innovation in this paper is that this is the first time genetic

    algorithms are used in this manner to encrypt images.

    The results obtained for the correlation coefficients and the

    entropies of the images proved the high efficiency of this method.

    This algorithm can be used in the future with a new crossover

    operatoror with another fitness function to encrypt images.

    33

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    34/35

    My Idea for further enhancement

    Use of RGB image

    Separate the Red, green and blue component.

    Use the chaotic function logistic map to encrypt the each

    component. Do the crossover among each component.

    Use Intensity Profile as the fitness function.

    34

  • 7/29/2019 A Hybrid Genetic Algorithm and Chaotic Function Model for Image Encryption

    35/35

    35