Loading...

Towards genetic feature selection in image steganalysis

Ramezani, M ; Sharif University of Technology | 2010

735 Viewed
  1. Type of Document: Article
  2. DOI: 10.1109/CCNC.2010.5421805
  3. Publisher: 2010
  4. Abstract:
  5. In this study, a new feature-based steganalytic method is presented and four classification methods: Fisher Linear Discriminant, Gaussian naïve Bayes, Multilayer perceptron, and k nearest neighbor, are compared for steganalysis of suspicious images. The method exploits statistics of the histogram, wavelet statistics, amplitudes of local extrema from the 1D and 2D adjacency histograms, center of mass of the histogram characteristic function and co-occurrence matrices for feature extraction process. In order to reduce the proposed features dimension and select the best subset, genetic algorithm is used and the results are compared through principle component analysis and linear discriminant analysis. The results show that the proposed method achieves higher accuracy in discriminating between innocent and stego images, as compared to one of well-known image steganalysis schemes
  6. Keywords:
  7. Genetic algorithm ; Steganalysis ; Center of mass ; Classification ; Classification methods ; Co-occurrence-matrix ; Dimension reduction ; Feature selection ; Feature-based ; Fisher linear discriminants ; Gaussians ; Histogram characteristic functions ; K-nearest neighbors ; Linear discriminant analysis ; Local extremum ; Multi layer perceptron ; Principle component analysis ; Steganalysis ; Stego image ; Wavelet statistics ; Discriminant analysis ; Genetic algorithms ; Graphic methods ; Security of data ; Feature extraction
  8. Source: 2010 7th IEEE Consumer Communications and Networking Conference, CCNC 2010, 9 January 2010 through 12 January 2010, Las Vegas, NV ; 2010 ; 9781424451760 (ISBN)
  9. URL: http://ieeexplore.ieee.org/document/5421805/?reload=true