Study On A Semi-fragile Image Watermaking Based On Zernike Moment Invariants

In the past decade, digital multimedia has gradually penetrated into people’s daily lives.Among a variety of multimedia information, digital image plays a very important position. Butas the development of image processing technique, the image editing and modifying withoutperceptible becomes very easy. And it is very difficult to distinguish them with the naked eyeauthenticity. When the image content refers to political, military affairs, forensic evidence andother information, its credibility must be recognized. Therefore, the image authenticationtechnology is becoming an important topic in the area of information security research. Andsemi-fragile image watermark is usually achieved image authentication.In this paper, based on the summary of the existing semi-fragile digital image watermarking,this thesis focuses on semi-fragile watermarking for the semi-fragile of Zernike invariantmoments. A semi-fragile watermarking algorithm using Zernike moments is mainly put proposed;the major work of this paper is as follows:Firstly, the algorithm of extracting Zernike moments magnitudes as the image features isproposed. The algorithm first does integer wavelet transform three times on images, and thencalculate the Zernike moments of LL3and Zernike moments. ZMMs has semi-fragile natureunder the image application environment, it is proved to be a good feature vector. This paperproved its change caused by normal manipulations is less than that caused by malicioustampering. Under the application environment it also has semi-fragile characteristics, and couldbe used as feature vector in image authentication.Secondly, based on the semi-fragile characteristics of Zernike moments, the paper proposesan algorithm: these feature vectors as a watermark embed into the image. A large number ofexperiments show that the algorithm has good visibility.Thirdly, the Zernike moment combined with wavelet transform, watermark extraction anddetection algorithm is proposed. The semi-fragile characteristics of Zernike moments are different between normal operations and malicious attacks. And locate the tamper regions on thisbasis. The experiment results show that the algorithm has better robustness, for a normal attack,such as JPEG compression, rotation, it also has better stability; but for cutting, replacing andother operations, it has worse frangibility, which can be distinguished. In addition, the algorithmcan accurately locate the tampered areas, so it has good positioning capabilities.