Feature selection method for color image steganalysis based on fuzzy neighborhood conditional entropy

The color image steganalysis method creats many redundant features during feature extraction, which reduces the classification accuracy. To reduce the dimensionality of color image steganalysis features and improve classification accuracy, this paper proposes the C-FNCES method. First, we use the Fisher score to evaluate the importance of each feature, providing the basis for selecting the features of color image steganalysis. Second, the fuzzy neighborhood decision information system is introduced into the color image steganalysis feature since it can effectively process continuous data. The decision information system of color image steganalysis based on a fuzzy neighborhood is constructed. Then, we propose the fuzzy neighborhood conditional entropy model. The model is used to evaluate the role of features, providing a theoretical basis for feature selection in color image steganalysis. Finally, according to the Fisher score and fuzzy neighborhood condition entropy model, a steganalysis feature selection algorithm is designed. Our experiment showed that the C-FNCES method can not only effectively reduce the feature dimension but also improve the classification accuracy, which is better than the Steganalysis-α and CGSM methods.