Report for ODE month-long Project

BTech (Information Technology and Mathematical

Innovations), 1st Semester

Chaos-based image encryption using the Lorenz system

Agriya Khetarpal, Raunak Singh, Naman Priyadarshi, Saransh Chopra

Cluster Innovation Centre

University of Delhi
 Certificate of Completion

This is to certify that the following persons: Agriya Khetarpal, Raunak

Singh, Naman Priyadarshi, Saransh Chopra have completed this ODE
project under my guidance and supervision as per the contentment of the
requirements of the first semester in the course BTech (Information
Technology and Mathematical Innovations) at the Cluster Innovation
Centre, University of Delhi.

Dr Harendra Pal Singh

Assistant Professor
Department of Mathematics
Cluster Innovation Centre
University of Delhi
 Acknowledgement

It is a great pleasure for us to present this project titled “Chaos-based

image encryption using the Lorenz system” as a part of the first semester’s
curriculum in the BTech (Information Technology and Mathematical
Innovations) course at the Cluster Innovation Centre, University of Delhi.

We would like to thank our teacher and mentor, Dr Harendra Pal Singh,
who helped render this project an adequate success.
 Abstract

Chaos theory is an interdisciplinary branch of mathematics and physics

that deals with the mathematical concept of chaos described by dynamical
systems – a function whose output in a three-dimensional geometrical
plane depends upon the time variable. Many dynamical systems exist in
nature, from the simple pendulum to the water flow in a pipe, irregularities
in human heartbeat, or even the weather in a particular region. However,
some dynamical systems exist whose motion is not governed by simple
laws and is defined by specific non-periodic properties and high sensitivity
to initial conditions. Unlike in non-chaotic systems, here, even
infinitesimally small changes to the initial parameters yield a significantly
different trajectory of the particle/entity in motion, limiting the system’s
predictability. The focus of our project is to encrypt and decrypt an image
using the Lorenz chaotic system.
 Index

Certificate of Completion 2
Acknowledgement 3
Abstract 4
Body 6
Results 15
Future Work 19
References 21
Appendix 23
 Body

In this month-long project, our aim is to achieve the goal of encrypting an

image using a chaotic system. The chaotic system used is the Lorenz set of
linear ordinary differential equations, which are given by

𝑑𝑥
𝑑𝑡
= σ(𝑦 − 𝑥),
𝑑𝑦
𝑑𝑡
= 𝑥(ρ − 𝑧) − 𝑦,
𝑑𝑧
𝑑𝑡
= 𝑥𝑦 − β𝑧,

which attains the shape of a two fixed point attractor for the values
8
ρ = 28, σ = 10, and β = 3

The Lorenz chaotic attractor was discovered by Edward Lorenz in 1963

when he was investigating a simplified model of atmospheric convection.
It is a nonlinear system of three differential equations. With the most
commonly used values of three parameters, there are two unstable critical
points. The solutions remain bounded but orbit chaotically around these
two points.
A physical model simulating the Lorenz equations has been attributed to
Willem Malkus and Lou Howard around 1970. It consists of leaking cups
on the rim of a larger wheel, as shown in the diagram on the right. Liquid
flows from the pipe at the top; each cup leaks from the bottom.

Under different input flow rates, you should convince yourself that the
wheel will spin one way under just the right flow rate and then the other
chaotically.

The Lorenz system is deterministic, which means that if you know your
variables’ exact starting values, then, in theory, you can determine their
future values as they change with time. Lorenz demonstrated that if you
begin this model by choosing some values for x, y, and z, and then do it
again with just slightly different values, you will quickly arrive at
fundamentally different results. You can never know the exact value of any
physical measurement in real life, although you can get close.

The Lorenz system is non-linear in nature with two existing

non-linearities, i.e., 𝑥𝑦 and 𝑥𝑧

Symmetry within the Lorenz system of differential equations also exists.

The equations are invariant under the definition (𝑥, 𝑦)→ (− 𝑥, − 𝑦).
Hence, if a point in the phase space ((𝑥(𝑡), 𝑦(𝑡), 𝑧(𝑡)) is a solution, then
it follows that the point (− 𝑥(𝑡), − 𝑦(𝑡), 𝑧(𝑡)) is also a solution.
The point (𝑥*, 𝑦*, 𝑧*)= (0, 0, 0)is a fixed point of the Lorenz system for
all values of the parameters σ, ρ, and β. For β > 1, there is also a pair of
points at 𝑥* = 𝑦* = (β)(ρ − 1) which coalesce with the origin point in a
pitchfork bifurcation – a particular type of bifurcation where a system of
differential equations transitions from having one fixed point to three fixed
points.
Stability analysis of the Lorenz system

Linearisation of the equations mentioned above yields

ẋ = σ(y – z)
ẏ = ρ(x – y)
ż = –βz

The trace τ= – σ – 1 < 0 and determinant △ = σ (1 – ρ)

For ρ > 1, origin is a saddle point since △ < 0

In differential equations, a saddle point or minimax point is a point on the

surface of the graph of a function where the slopes (derivatives) in
orthogonal directions are all zero (a critical point), but which is not a local
extremum of the function. An example of a saddle point is when there is a
critical point with a relative minimum along one axial direction (between
peaks) and at a relative maximum along the crossing axis. However, a
saddle point need not necessarily be in this form.

For example, in the above plotted graph f(z) = x2 - y2 the point (0,0) is the
saddle point.
For ρ < 1, origin is a sink since τ2 – 4△ = (σ + 1)2 – 4σ(1 – ρ) = (σ – 1)2 –
4σρ > 0

A sink is defined as a point where all the local sets of points in the traced
function seem to coalesce towards, it can be thought to be analogous to a
regular kitchen sink installed in our homes and how the fluid entering the
sink wishes to exit from one specific point.

Actually, for ρ < 1 it can be shown that every trajectory approaches the
origin as t → ∞ the origin is globally stable, hence there can be no limit
cycles or chaos for ρ < 1
The substitution method

We attempt to describe the process used by us to achieve the goal of our

project here. We first import all the essential libraries which will be
necessary, such as matplotlib and NumPy. We then ask the user to enter the
path of the image they wish to encrypt. For the sake of confirmation, we
display the input image to the user. We then store the size of the image in
the form of variables. We then generate a key for every pixel of the image
in terms of its location and name it as the Lorenz key. This key is defined
by the parameters we use in the Lorenz system, and hence, it can be easily
customised and varied. We then initialise an empty image containing the
new space for the encrypted form of the input image. Now we attempt to
explain the central part of this project, the process of encryption. It is
essential to note in this method; we are just trying to encrypt by
substituting pixels. In another approach that we have attempted to explain
after this, we shuffle and substitute the pixels to secure the image further.
We now XOR each pixel with a pseudo-random number obtained using the
pseudo-random number generator function created previously. We
multiply each pseudo-random number with an enormous value of the order
105 and access the remainder generated by dividing the result by 256 to
create random numbers between 0 and 255 (both inclusive). We then
create the encrypted image by replacing it with the final pixels we get after
subjecting the initial pixels to the math we described above. We then
display the encrypted image. We initialise another empty image to store
the decrypted image once we subject the encrypted one to the exact
reverse procedure we followed to encrypt it and then display the decrypted
image.
The shuffle and substitution method

We again start by importing most of the important libraries required to

perform the project we aim to achieve, such as matplotlib and NumPy. We
then accept the image entered by the user they wish to encrypt. Again, for
the sake of confirmation, we display the entered image to the user. We
again store the size of the image in the form of variables. We generate the
same key as last time again by naming it as the Lorenz key and creating
three lists. This key, again, is defined by the parameters we use in the
Lorenz system and hence can be easily customised and varied. We now
initialise empty index lists to store an index of the pixels of the image. We
now initialise an empty image again to store the encrypted image to be
generated. We now populate the generated lists with the image’s
dimensions, preparing them to undergo the encryption process. We
proceed with the first step of encryption by shuffling the X index values
using a simple conditional algorithm explained in the code. We do the
same with the Y index, which has increased security by a greater level
than before. We create an encrypted image using these shuffled indices and
show the user the encrypted image after this step. This is where the
shuffling part of our method is over. We now subject this encrypted image
to the same substitution method we used in the previous section.
Confusion and diffusion

In cryptosystems, two desirable properties to hinder fast-scale decryption

using statistical analysis or other methods are confusion and diffusion.
Confusion in encryption refers to the process of encrypting an image
wherein the data is changed drastically from the input to the output, i.e.,
making it challenging to find the key, increasing the ambiguity and
vagueness of the data as a result.

Diffusion refers to the process of every part of the input data affecting
every part of the output to render the analysis of the cryptosystem a more
demanding task. In encryption, as opposed to confusion, diffusion is done
to reduce the number of attempts to deduce the key in itself.

However, diffusion is not a hundred percent accurate and leaves some

underlying patterns every time. If the diffusion process is fair, these
patterns get scattered throughout the output. If multiple patterns exist, they
are scattered amongst each other. Good diffusion processes and practices
vastly increase the data required to analyse the cryptosystem and crack it.

These two properties have a significant role to play in designing

encryption methods and systems. A combination of good confusion and
diffusion helps increase the security of a system. For example, the AES
(Advanced Encryption Standard) specification has excellent properties.
 Results

Substitution on a black-and-white image

Shuffle and then substitution on a black-and-white image
Substitution on a coloured image
Shuffle and then substitution on a coloured image
 Future Work

Based on the progress we have made so far in our effort towards this work,
we have come up with certain ideas we aim to implement in the future
regarding the same:

1. We aim to extend this method to video encryption since videos,

essentially, are a collection of continuous frames, i.e. images. We
can encrypt different frames with different systems so that the
whole video can not be decrypted by one approach. The image
frames can be shuffled in themselves as well, using an appropriate
cipher to add an extra layer of security.

2. Dynamic encryption: the encrypted image can keep on changing

with time; that is, the key to the image which governs the
encryption process can be made dynamic (changing with time).
Therefore, we will need to implement a process through which the
repeatedly changing key can be stored for a particular instant to
ease the decryption process while increasing security.

3. We can encrypt an image based on certain sectors classified by

certain shapes and patterns inside it. For example, we can use
object classification and other computer vision techniques to
identify items such as a flower pot or a tiger and encrypt them
using different paraphernalia compared to the surroundings.
Otherwise, different parts of the image can be divided (into squares
and rectangles), and the same process can be done on them.

4. We can encrypt the key to the system using similar encryption

methods to ‘hide’ the Lorenz system’ and implement two or
possibly higher (n) factor authentication.

5. Another topic of interest is enhancing the security of these

encrypted systems in general by testing them against brute force
 attacks and against AI-based attacks (for example, convolutional
neural networks trained to solve jigsaw puzzles) and so on.

6. In the future, we can encrypt images using other chaotic and

hyperchaotic systems, for example, the double and the triple
pendulums, the Chua’s circuit, the Rossler attractor, or even cellular
neural networks.
 References

1. Chen, Jun-xin & Zhu, Zhi-liang & Fu, Chong & Yu, Hai & Zhang,
Li-bo. (2015). A fast chaos-based image encryption scheme with a
dynamic state variables selection mechanism. Communications in
Nonlinear Science and Numerical Simulation. 20. 846–860.
10.1016/j.cnsns.2014.06.032.

2. Somaya Al-Maadeed, Afnan Al-Ali, Turki Abdalla, "A New

Chaos-Based Image-Encryption and Compression Algorithm",
Journal of Electrical and Computer Engineering, vol. 2012, Article
ID 179693, 11 pages, 2012. https://doi.org/10.1155/2012/179693

3. Shenyong Xiao, ZhiJun Yu, YaShuang Deng, "Design and Analysis

of a Novel Chaos-Based Image Encryption Algorithm via Switch
Control Mechanism", Security and Communication Networks, vol.
2020, Article ID 7913061, 12 pages, 2020.
https://doi.org/10.1155/2020/7913061

4. Bhagat, A., Abhishek Surve, Sanuj Kalgutkar and Apeksha

Waghmare. “Chaos Based Image Encryption and Decryption.”
(2016).

5. https://arxiv.org/ftp/arxiv/papers/1211/1211.0090.pdf

6. Alvarez, Gonzalo & Li, Shujun. (2006). Some Basic Cryptographic

Requirements for Chaos-Based Cryptosystems.. I. J. Bifurcation and
Chaos. 16. 2129-2151. 10.1142/S0218127406015970.

7. Chaos-based Cryptography: Theory, Algorithms and Applications.

Germany: Springer Berlin Heidelberg, 2011.

8. Gao, Tiegang & Chen, Zengqiang. (2008). Image encryption based

on a new total shuffling algorithm. Chaos, Solitons & Fractals. 38.
 213-220. 10.1016/j.chaos.2006.11.009.

9. https://youtu.be/DOwy0SuGD7Q

10. X. Zhang, L. Wang, Z. Zhou and Y. Niu, "A Chaos-Based Image

Encryption Technique Utilizing Hilbert Curves and H-Fractals," in
IEEE Access, vol. 7, pp. 74734-74746, 2019, doi:
10.1109/ACCESS.2019.2921309.

11. Rao, Surya. (2015). A New Chaotic Algorithm For Image

Encryption And Decryption Of Digital Color Images. International
Journal of Applied Engineering Research. 10.

12. Zhang, Jiangang & Zhang, Li & An, Xinlei & Luo, Hongwei &
Yao, Kutorzi. (2016). Adaptive coupled synchronization among
three coupled chaos systems and its application to secure
communications. EURASIP Journal on Wireless Communications
and Networking. 2016. 10.1186/s13638-016-0630-4.

13. Bonilla, L.L., Alvaro, M. & Carretero, M. Chaos-based true

random number generators. J.Math.Industry 7, 1 (2016).
https://doi.org/10.1186/s13362-016-0026-4

14. E. Solak, "Partial identification of Lorenz system and its

application to key space reduction of chaotic cryptosystems," in
IEEE Transactions on Circuits and Systems II: Express Briefs, vol.
51, no. 10, pp. 557-560, Oct. 2004, doi:
10.1109/TCSII.2004.834534.

15. Communication theory of secrecy systems,” Bell Systems

Technical Journal 28 (1949), 656 – 715

16. Lect6.dvi (ox.ac.uk)

 Appendix
Lorenz key generator
Substitution algorithm
Shuffle + Substitution algorithm