Current Proceedings on Technology

Yazarlar: Sajad Mirjavadi, A. M. S., M. Shariat Panahi, S. Maleki Jebeli, M. Mousavi

Konular:-

Anahtar Kelimeler:Machine learning,Combined Methods,Singular value based methods,RS method,SVM

Özet: Up to now, numerous approaches have been adopted for steganalysis of embedded data that is stored in gray-scale images through LSB method, each having some advantages and drawbacks. Recently, combined methods have received increasing level of popularity among the researchers in this field. Most of techniques for steganalysis are based on SVM which uses a series of features for difference. While, in combined methods, a set of features are taken from multiple methods. In this paper, a new approach for LSB steganalysis which inherits the merits of both Singular Value and well-known RS method, is proposed and its efficiency to produce acceptable results are shown. To reduce the complexity of the proposed approach, before providing SVM engine with features, a reduction framework like PCA is applied.

---