Blind Steganalysis for JPEG Images using SVM and SVM-PSO Classifiers
Deepa D. Shankar1, Prabhat Kumar Upadhyay2
1Deepa D.Shankar, Research Scholar, Banasthali Vidyapith, Rajasthan, India.
2Prabhat Kumar Upadhyay, Department Of Electrical and Electronics Engineering, Birla Institute of Technology, Mesra, Ranchi, India.
Manuscript received on 15 September 2019 | Revised Manuscript received on 23 September 2019 | Manuscript Published on 11 October 2019 | PP: 1239-1246 | Volume-8 Issue-11S September 2019 | Retrieval Number: K125009811S19/2019©BEIESP | DOI: 10.35940/ijitee.K1250.09811S19
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Blind steganalysis or the universal steganalysis helps to identify hidden information without previous knowledge of the content or the embedding technique. The Support Vector Machine (SVM) and SVM- Particle Swarm Optimization (SVM-PSO) classifiers are adopted for the proposed blind steganalysis. The important features of the JPEG images are extracted using Discrete Cosine Transform (DCT). The kernel functions used for the classifiers in the proposed work are the linear, epanechnikov, multi-quadratic, radial, ANOVA and polynomial. The proposed work uses linear, shuffle, stratified and automatic sampling techniques. The proposed work employs four techniques for image embedding namely, Least Significant Bit (LSB) Matching, LSB replacement, Pixel Value Differencing (PVD) and F5 and applies 25% embedding. The data to the classifier is split as 80:20 for training and testing and 10-fold cross validation is carried out.
Keywords: Blind steganalysis, SVM, SVM-PSO, DCT, LSB, PVD, ANOVA1.0, Embedded Techniques, Cross validation.
Scope of the Article: Image Security