基于提升小波变换和BP神经网络的图像哈希算法*

针对数字图像作为一种常用的数字多媒体信息，对其真实性和完整性的认证显得尤其重要，提出了一种基于提升小波变化和BP神经网络的图像哈希算法。首先利用图像像素矩阵和构造的函数来训练BP神经网络；再将图像进行提升小波变换，利用低频分量组成矩阵；最后利用已经训练好的BP神经网络来产生哈希序列。实验结果表明，本算法不仅可以抵抗内容保持的修改操作，而且能够很好地区分恶意攻击，有一定的鲁棒性和脆弱性。该技术在图像认证、版权保护、安全和基于内容的图像检索等方面有应用价值。

Image hashing algorithm based on lifting wavelet transform and BP neural network

As a useful digital multimedia information, authenticating the authenticity and the integrity of digital image is especially important. This paper presented an image hashing algorithm based on lifting wavelet transform and BP neural network. Firstly, used the image pixel matrix and the constructed function to train the BP neural network, and then the low-frequency components obtained by the lifting wavelet transform composed the matrix. At last, generated the image hash sequence by the well-trained BP neural network. The experimental results show that the proposed scheme not only can resist the content-preserving modifications, but also it is sensitive to the image of malicious tampering, so it is robust and fragile. Therefore, the method could be used to image authentication, copyright protection, image security and content-based image retrieval and so on.