基于图像位平面相关性的隐写分析*

通过对原始图像和经过LSB隐写的隐蔽图像的分析，信息的再次嵌入对隐蔽图像与其位平面之间相关性的影响很小，而对原始图像与其位平面之间相关性的影响很大。以此为基础，提出了一种基于图像和其位平面之间的线性相关性的隐写分析方法，实现了对图像是否载密的检测。实验表明，对于图像的LSB隐写，可以进行有效的检测。该方法也可以对图像的其他位平面的嵌入进行分析。

Steganalysis Based on Bit Plane Correlation in Images

In this paper, we study the difference between the cover image and the stego image, and find that the linear correlation between the image and its bit plane wiU be changed only shightly when we re-embed a message in stego images, but will be changed significantly in cover images. From above, we propose a steganalysis approach based on the correlation between image and its bit plane, which can detect whether an object image has a secret message or not. The experimental results show the validity of our method for KSB steganography software. In addition, the approach also can detect the steganography in the other bit planes of an image.