ResearchBib Share Your Research, Maximize Your Social Impacts
Sign for Notice Everyday Sign up >> Login

Secured High Performance Reversible Covert Communication Over An Encrypted Color Carrier Image Using KNN and SVM Classifier

Journal: IPASJ International Journal of Computer Science (IIJCS) (Vol.6, No. 1)

Publication Date:

Authors : ;

Page : 008-009

Keywords : Keywords: Chi-square attack; Entropy; Huffman coding; Histogram; KNN Classifier; Public key encryption; PSNR; Reversible Data Hiding; Support Vector Machine classifier; SPA; UACI.;

Source : Downloadexternal Find it from : Google Scholarexternal

Abstract

ABSTRACT This Paper proposes a Novel frame work for covert communication of sensitive information over an encrypted public digital medium in a Reversible fashion. The Reversibility makes such an image data hiding approach more attractive in critical scenarios like medical image sharing , military and remote sensing. Reversibility assures perfect reconstruction of carrier medium information upon the extraction of secret image content embedded in it, while protecting payload information content's confidentiality. In the proposed algorithm color image is used as the carrier, Due to this, the amount of payload information that could be embedded is much more in comparison with gray scale image, since the information could be hidden in three different planes R, G and B. Encrypted color image is the carrier for sensitive information. Greater embedding capacity is further achieved through public key modulation technique. At the Receiver end, two of the powerful image classification techniques, K Nearest Neighbor and Support vector machine algorithms are used to separate encrypted and unencrypted image blocks. Training features used for classification are ? Pixel variation in all four directions ? Entropy ? Standard deviation ? Histogram Performance evaluation of image classifiers is made in terms of its accuracy i.e. their ability to correctly classify the image blocks. The proposed technique comes with a unique feature of jointly extracting both cover and payload information from a marked image in a lossless fashion. Effectiveness of the proposed algorithm is evaluated in comparison with existing techniques considering various parameters like embedding capacity, PSNR and image classifier accuracy. The proposed algorithm is proven more secured considering various cryptographic and steganographic attacks. Parameters considered for security analysis are chi-square, SPA (sample pair analysis), NPCR (number of changing pixel rate), UACI (unified averaged changed intensity), entropy and correlation.

Last modified: 2018-02-13 20:31:26