一种利用相邻像素相关的隐写分析算法

Steganalysis Based on Neighbor Image Pixels Correlation

自然图像在一定邻域内具有相关性,且相关性的大小随着像素间距离的增大而减小。基于图像的这种统计特性,提出一种隐写分析方法。对图像LSB(the least significant bit)序列间隔1、间隔2进行按位异或运算,统计运算结果中“1”的数目,根据图像LSB序列和嵌入的秘密消息序列的统计差异,利用统计结果通过卡方显著性检验判断图像中是否嵌入有隐秘消息。实验结果表明,在有效控制虚警概率的前提下本算法对LSB匹配隐写有良好的检测性能。另外算法从图像像素的相关特性着手进行隐写分析,在检测过程中不针对具体隐写算法,具有较强通用性。

The carrier image has the correlation in a certain range, and the size of the correlation decreases with the increase of the distance between the image elements. Based on this character, we introduce a steganalysis. With the statistic number of interval 1, interval 2 model 2 addition to the LSB（the least significant bit） serials. Then get the distinct test of Х^2 to detect whether it has secret information. Experiment shows that this algorithm has well detecting performance to LSB matching. The proposed method is worked on the image pixels correlation. It doesn＇ t focus on a given steganography in detecting process. So it has a good performance.