Research On Geometric Distortion-Resilient Image Integrity Authentication Method

With the rapid development and wide application of the multimedia technology and network technology, digital multimedia information (images, video, audio, etc.) is quickly and easily transferred with various forms on the network, and information exchange have development from simple text messages to more kinds of multimedia information. However, with convenience brought by digital media to people, on the same time, some potential risks have been introduced, such as infringement (for example copy) becomes easier, and content tampered will be more convenient. Copyright protection and integrity (authenticity) certification issue of digital media content need to be solved urgently.In this paper, the method for image authentication——image digital signature and digital watermarking technology have been researched and analyzed. The main works are as follows:The characters of the digital signature for authentication, specially, geometric distortion-resilient image digital signature are studied. By improving the signature generated method in RASH, an image signature algorithm is proposed based on Radon transform and wavelet transform. This algorithm possesses of the authentication function, and maintains a good robustness for the geometric distortion (such as rotating and scaling) of the image, and is able to locate and identify the tamper region. It achieves the function of locate malicious tamper after large-angle rotation. Experimental results show that this scheme is robust to geometric distortion, and sensitive to malicious attack.Zero-watermarking technology, specially, DCT-based Zero-watermarking technology is studied. Then we proposed a new geometric distortion-resilient Zero-watermarking algorithm. Proposed method is based on geometric invariance of Radon transform and uses statistics feature to constructed watermark and registered it in the database. In watermark detection, matching the extracted watermark information to the watermark in the database, one can certify the image integrality. Simulation experiments are carried out on Matlab7 platform, the experimental results show that our method is robust to the general geometric transformation (rotation, scaling, and translation) operation.