A

Synopsis
on
<TITLE OF PROJECT/SYNOPSIS>

Submitted for Partial Fulfillment of award of

the degree

BACHELOR OF TECHNOLOGY
in
COMPUTER SCIENCE & ENGINEERING
by

XYZ (190)
XYZ (190)
XYZ (190)
XYX (190)

Under the Guidance of

Mr. XYZ
(Assistant Professor, Department of CSE)

Department of Computer Science & Engineering

INSTITUTE OF TECHNOLOGY & MANAGEMENT
GORAKHPUR
(AFFILLIATED TO DR. A.P.J. ABDUL KALAM TECHNICAL UNIVERSITY, LUCKNOW)
Session 2023-24
 INTRODUCTION

A technique used for hiding secret data within a standard, public file or message to avoid
detection is known as Steganography. The secret data is then extracted once it arrives at its
destination. Steganography, in conjunction with encryption, can be used to further conceal or
safeguard data.

Steganography is the practice of concealing a secret message within another, ordinary, message.
Commonly, Steganography is used to unobtrusively hide a small message within the noisy
regions of a larger image. In this study, we attempt to place a full size color image within another
image of the same size. Deep neural networks are simultaneously trained to create the hiding and
revealing processes and are designed to specifically work as a pair. The system is trained on
images drawn randomly from the ImageNet database, and works well on natural images from a
wide variety of sources. Beyond demonstrating the successful application of deep learning to
hiding images, we carefully examine how the result is achieved and explore extensions. Unlike
many popular steganographic methods that encode the secret message within the least significant
bits of the carrier image, our approach compresses and distributes the secret image's
representation across all of the available bits.
Technology has blitz scaled over the past years leading to a wide usage of multimedia for
transferring data, especially Internet of Things. Usually, the transfer happens over insecure
network channels. In particular, the internet has gained accelerated popularity for exchanging
digital media and individuals, private companies, institutions, governments use these multimedia
data transfer methods for exchanging data. Though there are numerous advantages attached with
it, one prominent disadvantage is the privacy and security of the data. The availability of
numerous readily available tools capable of exploiting the privacy, data integrity and security of
the data being transmitted has made the possibility of malicious threats, eavesdropping and other
subversive activities. The prominent solution is data encryption where the data is converted into
a cipher text domain using encryption key. At the receiving end, the cipher text is converted into
plain text using a decryption key. Using data encryption, the original data is not visible, however,
cipher text is visible in a scrambled form to human eyes leading to suspicion and further
scrutiny. A new research topic, steganography, has gained acceptance in this context to hide the
data that is not perceptible to human eyes.

Information hiding techniques have been available for a long time but their importance has been
increasing recently. The main reason is the increase in the data traffic through the internet and
social media networks. Though the objectives of cryptography and steganography are similar,
there is a subtle difference. Cryptography makes the data unbreakable and unreadable but the
cipher text is visible to human eyes. Steganography, which is used to hide the information in
plain sight, allows the use of wide variety of the secret information forms like image, text, audio,
video and files. Digital watermarking is another method where confidential information is
embedded to claim ownership. Cryptography is the popular method used for information hiding,
but, steganography is gaining popularity in recent times.

Steganography can be defined as the process of hiding a secret small multimedia data inside
another but much larger multimedia data such as image, text, file or video. Image steganography
is a technique to hide an image inside another image. In image steganography, the cover image is
manipulated in such a way that the hidden data is not visible thus making it not suspicious as in
the case of cryptography. Inversely, Steganalysis is used to detect the presence of any secret
message covered in the image and to extract the hidden data. Steganalysis helps in classifying if
the image is either a stego image or a normal image. Apart from classifying the image, further
investigation is carried out to detect the location and the content of the secret image inside the
cover image.

With the availability of massive amounts of data, deep learning (DL) has become the trend and is
extensively used for many applications. Deep learning is a useful tool in various applications like
image classification, automatic speech recognition, image recognition, natural language
processing, recommendation systems, processing of medical images. Though research on
steganography is quite recent, it has benefited from DL methods including convolutional Neural
Networks (CNNs) Generative Adversarial Networks (GANs) based methods and their
deployment in both steganography and Steganalysis.
 OBJECTIVE

 To recognize suspected data streams, determine whether or not they have hidden
messages encoded into them, and, if applicable, recover the hidden data.
 To identify suspected packages, determine whether or not they have a payload encoded
into them, and, if possible, recover that payload.
METHODOLOGY/PLANNING OF WORK

After reviewing all the frameworks available, the methodologies are primarily grouped into three
categories, namely, traditional image steganography methods, CNN-based image steganography
methods and GAN-based image steganography methods. Traditional methods are frameworks
which use methods that are not related to machine learning or deep learning algorithms. Many
traditional methods are based on the LSB technique. CNN-based methods are based on deep
convolutional neural networks for embedding and extracting the secret messages and GAN-
based methods use some of the GAN variants.
The CNN model is composed of three parts: The Preparation Network, Hiding Network
(Encoder) and the Reveal Network. Its goal is to be able to encode information about the secret
image S into the cover image C, generating C_prime that closely resembles C, while still being
able to decode information from C_prime to generate the decoded secret image S_prime, which
should resemble S as closely as possible. The Preparation Network has the responsibility of
preparing data from the secret image to be concatenated with the cover image and fed to the
Hiding Network. The Hiding Network then transforms that input into the encoded cover image
C_prime. Finally, the Reveal Network decry the Reveal Network, as suggested by the paper.
Although the author of the paper didn't originally specify the architecture of the three networks,
we discovered aggregated layers showed good results. For both the Hiding and Reveal networks,
we use 5 layers of 65 filters (50 3x3 filters, 10 4x4 filters and 5 5x5 filters). For the Preparation
Network, decodes the secret image S_prime from C_prime. For stability, we add noise before we
use only 2 layers with the same structure. Note that the loss function for the Reveal Network is
different from the loss function for the Preparation and Hiding Networks. In order to correctly
implement the updates for the weights in the networks, we create stacked Keras models, one for
the Preparation and Hiding Network (which share the same loss function) and one for the Reveal
Network. To make sure weights are updated only once, we freeze the weights on the layers of the
Reveal Network before adding it to the full model. Experimentation Basic Experimentation has
been done by changing the activation function of the architecture to examine and learn the
performance of the model. All other parameters such as ConvNet architecture, no. of epochs and
the optimizer function are kept constant.