Open AccessAnti-forensic Approach to Remove Stego Content from Images and Videos
Author(s) -
P. P. Amritha,
M. Sethumadhavan,
R. Krishnan,
Saibal K. Pal
Publication year - 2018
Publication title -
journal of cyber security and mobility
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.198
H-Index - 9
eISSN - 2245-4578
pISSN - 2245-1439
DOI - 10.13052/2245-1439.831
Subject(s) - steganography , steganalysis , computer science , covert , information hiding , artificial intelligence , cover (algebra) , steganography tools , computer vision , image (mathematics) , pattern recognition (psychology) , engineering , mechanical engineering , linguistics , philosophy
Covert transmission of information hidden in different media to either a general or targeted audience constitutes steganography. However, this technique can be misused to transmit undesirable information. Traditionally the removal of such content necessitated the knowledge of the steganographic algorithm used. However,we address the scenario where such stego is removed using generic image processing operations along with an anti forensic method without assuming any knowledge of the steganographic algorithm used. The application of generic image processing operations also causes degradation of cover image, which can also be restored using this anti forensic method. Our procedure has been tested on a variety of steganographic algorithms including HUGO-BD,WOW, Synch and J-UNIWARD. By applying universal steganalysis we found that all images which have been subjected to our procedure have become stego free. However, a direct evaluation of the stego content assuming knowledge of the stego content and its location showed that 80 percentage of the stego is removed without significantly impacting the visual image quality. Video stream containing isolated static images have been addressed in this paper. The peak signal-to-noise ratio and structural similarity metric values of cleaned images and videos are found to be in the range 30dB–40dB and 0.81–0.99 respectively.