基于YCbCr高阶统计量的图像隐写分析

论文提出了一种应用小波变换基于YCbCr图层高阶统计量的JPEG图像通用隐写分析算法，该算法不局限于某种特定的隐匿算法。由于JPEG图片使用的是YCbCr颜色模型，因而若仍在RGB色彩层上构造高阶统计量，必然会淡化隐写后图像的特征，所以本文提出在YCbCr色彩层上处理高阶统计量，消除了RGB色彩模型同YCbCr色彩模型之间转化所产生的误差；统计量的构造通过原图与切割图像的差异，可减少SVM训练的样本数。

Steganalysis Based on Higher-order Statistics of YCbCr Panels for JPEG Images

This paper presents a new steganalysis method for detection of steganography in image＇s frequency domain, which is based on higher-order statistics of YCbCr panels for JPEG images. The detecting method is universal and not restrained to some special steganography methods. While the JPEG images are presented in YCbCr color channels, so if the higher-order statistics are still extracted on RGB color channels, the characteristics of steganographic images are inevitably be weakened. In this paper, the statistics are extracted in YCbCr color channels, so the differences between RGB channels and YCbCr channels have been eliminated. Those statistics are based on the differences between initial image and cut-image, which can decrease the amount of SVM＇s training samples, thus reduce the processing time and get better detection results.