Please use this identifier to cite or link to this item: http://hdl.handle.net/10603/8912
Title: Image steganalysis using artificial neural networks
Researcher: Sujatha P
Guide(s): Purushothaman S
Keywords: Computing Sciences
Artificial Neural Networks
Upload Date: 17-May-2013
University: Vels University
Completed Date: 19/04/2013
Abstract: This research work presents the steganalysis of steganographic images under predefined way of hiding information. Hiding information in the cover image is done by different methods. However, steganalysis algorithms proposed in this research work are unaware of the methods in which procedure the information was hidden in the cover image. The proposed algorithms are based on Artificial Neural Networks (ANN) for finding out the presence of any hidden information. Steganography is the process of embedding text information or imageinformation into cover image. The embedding is otherwise called hiding information in a cover image. The cover image appears with no change in the content when looked at the embedded image. At the same time, the presence of hidden information can be identified and if possible, the original hidden information can be recovered for better interpretation. This research identified group of cover images and information images that analyses and modifies existing algorithms and proposed their implementation for steganalysis. There are situations in which the presence of hidden information can be identified but cannot be reconstructed as interpretable information. Existing literature has used mostly statistical methods to identify and reconstruct the original information. Very few works have been carried out using ANN algorithms. The scope of this research work considers existing ANN algorithms like 1. Back propagation algorithm (BPA) 2. Functional update back propagation algorithm (FUBPA) 3. Radial basis function (RBF) Supervised BPA has been considered and learning of the generated data has been done. Training of the network with different learning factors has been tried and finally a value for learning factor with value 1 has been selected. This indicates a standard convergence. As a development of the vii BPA algorithm, conditional BPA called functional update back propagation ealgorithm (FUBPA) have been developed and implemented.
Pagination: 133p.
URI: http://hdl.handle.net/10603/8912
Appears in Departments:School of Computing Sciences

Files in This Item:
File Description SizeFormat 
01_title.pdf89.18 kBAdobe PDFView/Open
02_certificate.pdf33.06 kBAdobe PDFView/Open
03_abstract.pdf75.09 kBAdobe PDFView/Open
04_declarations.pdf33.46 kBAdobe PDFView/Open
05_acknowledgements.pdf33.29 kBAdobe PDFView/Open
06_contents.pdf94.48 kBAdobe PDFView/Open
07_list of tables.pdf31.55 kBAdobe PDFView/Open
08_list of figures.pdf42.57 kBAdobe PDFView/Open
09_abbreviations.pdf111.17 kBAdobe PDFView/Open
10_chapter 1.pdf502.04 kBAdobe PDFView/Open
11_chapter 2.pdf562.68 kBAdobe PDFView/Open
12_chapter 3.pdf5.2 MBAdobe PDFView/Open
13_chapter 4.pdf616.01 kBAdobe PDFView/Open
14_chapter 5.pdf337.34 kBAdobe PDFView/Open
15_chapter 6.pdf466.27 kBAdobe PDFView/Open
16_chapter 7.pdf203.79 kBAdobe PDFView/Open
17_conclusion.pdf47.25 kBAdobe PDFView/Open
18_bibliography.pdf219.15 kBAdobe PDFView/Open


Items in Shodhganga are protected by copyright, with all rights reserved, unless otherwise indicated.

Altmetric Badge: