Research On Perceptual Hashing Algorithm Based On Image Content Authentication

With advances in digital technology and the rapid development of Internet, storage, processing and distribution of digital multimedia products become more convenient and faster. However, this also leads to potential message security problems. Consequently, the authentication of digital multimedia content is becoming increasingly pressing. As an useful digital multimedia information, authenticating the authenticity and the integrity of digital image is especially important. The purpose of traditional image hash functions is to extract a short binary string as a digest from a large digital message, and their key feature is that they are extremely sensitive to the input message; changing even one bit of the input message will change the output dramatically. However, image data often undergoes various manipulations of preserving content such as noise pollution, filtering and lossy compression etc.So the image hash values should depend on image contents. Not only the algorithm is robust, but also the algorithm can distinguish malicious tampering. To solve the above problems, the main work of this paper is as follows:(1) A novel image Hashing algorithm based on wavelet lifting scheme is presented. To begin with, the image is transformed by lifting wavelet. And then, the features of image are extracted by the DCT and DWT transform. After that, intermediate image hash is obtained with the chaos system. At last, the final image hash is generated by two-valued quantification. Compared with the traditional wavelet transform, the second generation wavelet transform has several unique advantages of time-domain to achieve the optimal compression, no edge effects, lossless compression, programming easy and fast. The experimental results show that the proposed scheme is robust against rotation, filtering, shear and JPEG compression. And it is sensitive to the image of malicious tampering. Additionally, the chaos system obviously enhances the security.(2) An image Hashing algorithm based on lifting wavelet transform and BP neural network is presented. Firstly, the use of the image pixel matrix and a constructed function to train the BP neural network, and then the low-frequency components obtained by the lifting wavelet transform composes the matrix. At last, the image hash sequence is generated by the well-trained BP neural network. The experimental results show that the proposed scheme can resist the content-preserving modifications, and it is robust and fragile.