Fuzzy based steganalysis pattern discovery for high accuracy

Abstract

Steganalysis pattern is used to identify and detect the stego from cover images. Embedded Mechanism is used to combine cover image with stego pattern for hiding significant information. In order to classify stego from cover image a pattern need to be constructed for steganography algorithm. In the proposed work high detection accuracy is expected, using evolutionary algorithm where significant features are classified for stego image via fuzzy rules. These rules are based on linguistic knowledge and numerical data. The fuzzy classifier system will automatically generate if then rules from training patterns and fitness value. This multi pattern classification is compact and high performance for accuracy. Through the pattern discovered steganography method is predicted by involving different image patterns used for stego images. Based on the discovered knowledge suitable models are trained and saved in databases. Image databases are validated using the methods investigated using Steganalysis and steganography techniques. The steganography method classifies the image into multiple classes and each class has feature pattern for stego and cover images. These features extracted are stored in databases by stego pattern classification (SPD). The experimental results obtained define the stego pattern extracted from training samples. Steganalysis techniques used are based on the text, Image formats like JPEG, BMP, GIF and PNG and audio. Steganography algorithms are used to hide information in cover media to obtain stego media. Steganography techniques protects and safeguards personal information of individuals from criminals. The performance results are evaluated using Anova tool and statistical calculations are done through Matlab for improved results.