JPEG隐写分析中的校准原理探究

The study of calibration in JPEG steganographic analysis

传统的隐写图经过剪切重压缩得到原始图像的近似图像,通过对该方法及其与检测特征之间的关系研究,建立了校准与校准特征的一维数学空间,该空间引入参考变换概念,定性地分析了校准方法和校准差分特征的相关性,从而得出校准技术提高检测性能的根本原因,同时得出校准图像不一定是原图相似图的结论,对校准特征的选择提供了理论依据。

Traditional steganographic figures got the approximate images of the original images after cropped and recompressed. through the study about the relationship between this method and the detection feature, the article showed a kind of mathematic space between calibration and calibration features in one dimensional. This space introduced a concept of reference transformation and had qualitative analysis of the correlation between calibration methods and calibration difference features, thus it concluded that the root reason that the calibration technology improves the detection performance, at the same time calibration images are not necessarily similar to the original images, this conclusion provides theory basis for the selection of calibration features.