数字图像LSB隐写分析统计模型初探

数字图像LS(BLeast Significant Bits)在统计上与二项分布相似,结构上与原宿主图像仍然具有一定的相关性,从而导致经典LSB替换算法产生直方图"阶梯效应"。本文主要从图像像素以及像素间关系的特征函数出发,观察其常见统计模型的变化,为设计有效的图像隐写算法及隐写分析算法提供参考依据。

On Statistical Models for LSB-based Image Steganalysis

The LSB（Least Significant Bits） of a digital image is statistically very similar to binomial distribution,but correlated with the host image structurally to some extent,leading to the ＂stair-step effect＂ of the histogram after the classical LSB replacement.In this paper,we address the issue of inspecting the changing rules of the typical statistical models by studying the characteristic function of the image pixels and/or between pixels,for the future reference of effective image Steganography and steganalysis.