Image authentication and tamper localization based on coupling between adjacent pixels

Digital image information has the advantages of easy storage and communication, especially with the continuous emergence of powerful image processing software, editing and modifying digital images has become extremely convenient. Subsequently, issues such as low security and easy tampering of digital images have emerged, and the integrity and authenticity of images have been questioned. Some important applications, such as news images, court evidence, medical diagnoses, etc., are not allowed to have their content modified. Passive authentication methods are often only suitable for specific images or situations, dont have the ability to locate the tamper areas. Active methods based on fragile watermarks often embed external information, making it inconvenient to perform blind authentication on the receiving end and resist malicious attacks that aim at bypassing tamper detection. In this paper, we propose to combine the advantages of passive authentication and active authentication. Firstly, an image is first divided into non-overlayed blocks, then generate check code for each pair of strongly coupled pixels within the same block. Fragile watermark technology is exploited to embed the check codes randomly based on a private key in the pixels of the image itself to achieve blind authentication for the receiver. Finally, we conduct the experiment in which a large number of images have been simulated for tampering and detected for authentication. The results show that compared with other similar methods, this paper not only has high detection accuracy, but also has high accuracy in locating the tampering location. In addition, the method proposed in this paper has other advantages in terms of computational cost and security.