Research on Steganalysis for Text Steganography Based on Font Format
Lingyun Xiang, Xingming Sun, Gang Luo, Can Gan · 2007
In the research area of text steganography, algorithms based on font format have advantages of great capacity, good imperceptibility and wide application range. However, little work on steganalysis for such algorithms has been reported in the literature. Based on the fact that the statistic features of font format will be changed after using font-format-based steganographic algorithms, we present a novel Support Vector Machine-based steganalysis algorithm to detect whether hidden information exists or not. This algorithm can not only effectively detect the existence of hidden information, but also estimate the hidden information length according to variations of font attribute value. As shown by experimental results, the detection accuracy of our algorithm reaches as high as 99.3% when the hidden information length is at least 16 bits.