Neighbour Pixel Color Correlativity On Image Steganalysis

Jabbar, Alaa Abdulhussein (2015) Neighbour Pixel Color Correlativity On Image Steganalysis. Doctoral thesis, Universiti Teknikal Malaysia Melaka.

Text (24 Pages) Neighbour Pixel Color Correlativity On Image Steganalysis.pdf - Submitted Version Download (617kB)

Abstract

Steganalysis is science of discovering presence of secretively embedded data within the potential carriers. Since images are one of the most commonly used digital media on networks, they grab less attention than other digital media types. As a result, images are very appropriate cover for concealing presence of secret data transmission. Insertion of external data degrades the natural color correlativity among the pixels. Statistical steganalysis is an effective approach for discovering the generated anamolies through embedding process in content of images. Various statistical steganalysis techniques have been designed to discover presence of embedding artifacts and anomalies. However, the current practices are not efficient enough for detection of steganography. To overcome the problem, this research proposes a novel statistical steganalysis technique which efficienctly detects data embedding. Since this method statistically examines the suspicious carriers then it has more capability in distinguishing effects of new steganographic techniques. Color correlativity among image pixels varies based on image content and data embedding degrades this natural correlativity. This research classifies images in four image themes and then studies their behaviors after data embedding in different embedding ratios. The proposed method examines color correlativity of pixels of the given image with the neighbor pixels and then compares the results with the extracted color correlativity behavioral patterns. The most similar color correlativity behavioral pattern determines image theme of the analyzed image and its embedding ratio. Efficiency evaluation of the designed color correlativity statistical steganalysis method shows outstanding efficiency ehancement in comparison with the current practices. The designed technique efficiently determines both image theme and embedding ratio. The efficiency dramatically increases in detecting either of image theme or estimating message length.

Item Type:	Thesis (Doctoral)
Uncontrolled Keywords:	Image processing Digital techniques, Image processing, Neighbour Pixel Color Correlativity, Image Steganalysis
Subjects:	T Technology > T Technology (General) T Technology > TK Electrical engineering. Electronics Nuclear engineering
Divisions:	Library > Tesis > FTMK
Depositing User:	Mohd Hannif Jamaludin
Date Deposited:	05 Aug 2016 02:11
Last Modified:	11 Nov 2020 09:04
URI:	http://eprints.utem.edu.my/id/eprint/16873
Statistic Details:	View Download Statistic

Actions (login required)

View Item