Year

2020

Degree Name

Doctor of Philosophy

Department

School of Electrical, Computer and Telecommunications Engineering

Abstract

Steganography is the art and science of hiding information in plain sight with the help of multimedia communications. Steganography is beneficial in situations where sending encrypted messages may raise suspicion leading to communication breakdown between the sender and the receiver. Noticeable encrypted messages, no matter how unbreakable they are, arouse interest and may be considered illegal in several security protocols. Therefore, one of the main advantages of steganography over cryptography alone is that the communicated information does not attract attention to itself as an object of scrutiny. While the primary purpose of steganography is information security, at present, various steganographic systems can also be applied in a variety of scientific fields, commercial, social media, and for defence purposes. In image steganography, the confidential data is embedded over digital images and transmitted inconspicuous over unsecured communication channels. In contrast to information hiding, detecting and extracting hidden data from embedded media are collectively known as steganalysis and is currently an attractive area of research. Steganalysis performance can also be utilised in evaluating the effectiveness of the steganographic system against intruder attacks by using intelligent analysis.

This dissertation investigates state-of-the-art image steganography and steganalysis approaches and focuses on developing advanced image steganography and universal steganalysis methods to address the research gaps. The objectives of this research can be briefly described as an investigation on different embedding methodologies in cover region selection-based and coverless steganography in the transform domain, develop cover image selection from database-based steganography and develop universal steganalysis approach. This research has used dual-tree complex wavelet transform (DT-CWT) for developing four different embedding models based on steganography using innovative optimisation techniques and has used convolution neural network (CNN) for developing both cover image selection and a universal steganalysis model.

This thesis is unavailable until Thursday, June 01, 2023

Share

COinS
 

Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.