On image content diversity for steganalysis systems
The influence of the diversity of image content is often ignored in current steganalysis systems, and there is very tittle research focusing on the quantitative measurement of image content for steganalysis.This paper presents a new image content measure for steganalysis to study how image content diversity influences its performance.We define the measure based on image texture considering its useful statistical information in steganalysis.Together with steganographic payload, our proposed measure can provide accurate prediction for steganalysis performance.Experiments confirm that steganalysis performance strongly depends on image content.More importantly, the proposed measure is effective on improving the detection accuracy of steganalysis systems by constructing proper training database based on image content diversity.
steganalysis steganography image content diversity texture
QIAN Yinlong DONG Jing WANG Wei TAN Tieniu
Department of Automation, University of Science and Technology of China, Hefei, 230026 Center for Research on Intelligent Perception and Computing, National Laboratory of Pattem Recogniti
国内会议
武汉
英文
29-38
2015-03-28（万方平台首次上网日期，不代表论文的发表时间）