CRYPTO-STEG: A Hybrid Cryptology - Steganography Approach for Improved Data Security

• Main Authors: Atif Bin Mansoor, Zohaib Khan, Shoab AhmedKhan
• Format: Article
• Language: English
• Published: Mehran University of Engineering and Technology 2012-04-01
• Series: Mehran University Research Journal of Engineering and Technology
• Subjects: Cryptology; Steganography; Steganalysis; Classification
• Online Access: http://publications.muet.edu.pk/research_papers/pdf/pdf625.pdf
• ISSN: 0254-7821; 2413-7219

Internet is a widely used medium for transfer of information due to its reach and ease of availability. However, internet is an insecure medium and any information might be easily intercepted and viewed during its transfer. Different mechanisms like cryptology and steganography are adopted to secure the data communication over an inherently insecure medium like internet. Cryptology scrambles the information in a manner that an unintended recipient cannot easily extract the information, while steganography hides the information in a cover object so that it is transferred unnoticed in the cover. Encrypted data may not be extracted easily but causes a direct suspicion to any observer, while data hidden using steganographic techniques go inconspicuous. Cryptanalysis is the process of attacking the encrypted text to extract the information, while steganalysis is the process of detecting the disguised messages. In literature, both cryptology and steganography are treated separately. In this paper, we present our research on an improved data security paradigm, where data is first encrypted using AES (Advanced Encryption Standard) and DES (Data Encryption Standard) cryptology algorithms. Both plain and encrypted data is hidden in the images using Model Based and F5 steganographic techniques. Features are extracted in DWT (Discrete Wavelet Transform) and DCT (Discrete Cosine Transform) domains using higher order statistics for steganalysis, and subsequently used to train a FLD (Fisher Linear Discriminant) classifier which is employed to categorize a separate set of images as clean or stego (containing hidden messages). Experimental results demonstrate improved data security using proposed CRYPTO-STEG approach compared to plain text steganography. Results also demonstrate that the Model Based steganography is more secure than the F5 steganography.