Abstract: When the conventional algorithm decomposes the key information of the image, there is coding redundancy in the pixels, which leads to poor confidentiality after the information is hidden. This paper proposes an image key information hiding algorithm based on scrambling transformation, which removes image pixel redundancy through onedimensional linear prediction, obtains the prediction error, removes psychovisual redundancy through data set mapping, and converts layered images into binary images to eliminate image coding redundancy. The algorithm uses the Doppler effect to decompose the key information in the compressed image, and uses the wavelet transform to calculate the probability distribution density of the image, so as to realize the encryption of the key information of the image. The algorithm uses the scrambling transformation technology to transform the pixel position of the key information, so that they can be inserted into pixel pair and hidden in the phase component of pixel pair, so as to realize macroscopic scrambling and microscopic scrambling of image information. The experimental results show that, compared with the conventional algorithm, the fixed point ratio and the diagonal correlation coefficient of the encrypted image after hiding the key information are reduced, the difficulty of attacking the key information is improved, and the confidentiality of the key information is fully guaranteed.

Key words: scrambling transformation, information hiding, pixel compression, information decomposition, wavelet transformation, pixel position

摘要： 常规算法分解图像关键信息时，像素点中存在编码冗余，导致信息隐藏后保密性较差.提出基于置乱变换的图像关键信息隐藏算法.通过一维线性预测去除图像像素冗余，求出预测误差，经过数据集合映射去除心理视觉冗余，将分层图像转换为二进制图像，消除图像编码冗余；采用多普勒效应分解压缩图像中的关键信息，利用小波变换计算图像的概率分布密度，实现图像关键信息加密；利用置乱变换技术变换关键信息像素位置，使其插入像素对，隐藏在像素对的相位分量中，从而实现图像信息的宏观置乱和微观置乱.实验结果表明，相比常规算法，降低了隐藏关键信息后加密图像的不动点比和对角方向相关系数，提高了关键信息被攻击难度，充分保证了关键信息的保密性.

关键词: 置乱变换, 信息隐藏, 像素压缩, 信息分解, 小波变换, 像素位置