Anti-attack robustness multiple zero watermark method

Abstract

The invention relates to an anti-attack robustness multiple zero watermark copyright authentication scheme. Harris corner point detection is adopted for extracting carrier image feature points, then, the feature points are expanded into feature domains with the radium being R. Zernike moment invariants of the feature domains are used for constructing feature vectors, and in addition, the feature vectors are subjected to binaryzation to form the multiple zero watermark. In the watermark detection period, the copyright attribute of detected patterns is verified through the comparison between the correctly detected watermark number and the threshold value. According to the scheme, various attacks including fuzzy, noise, median filtering, joint photographic experts group (JPEG) compression, rotation, zooming, sharing and Stirmark random bending attacks (RBAs) can be avoided. Compared with the similar scheme, the method provided by the invention has better performance.

Description

technical field [0001] The present invention designs an anti-attack robust multi-fold zero watermark scheme, which is used to solve the copyright authentication of images, especially to verify the copyright ownership of a type of image (remote sensing image and medical image) that does not allow slight modification. The technology belongs to the field of information security. Background technique [0002] At present, in order to avoid piracy, digital watermark technology is widely used in the copyright authentication of digital images. Most of the existing watermarking schemes modify the image content in the spatial domain or frequency domain, and these methods cannot be directly applied to protect the copyright of remote sensing images and medical images, because the content of such images is not allowed to be slightly modified. Recently, a lossless watermarking technique called zero watermarking has been proposed. This kind of zero watermarking will not modify the content...

AI Technical Summary

Problems solved by technology

[0007] These zero-watermark schemes proposed so far are robust to common signal processing attacks such as blurring, noise, filtering, and compression, as well as attacks such as scaling, but are very robust to attacks such as shearing, large-scale rotation, and Stirmark RBA (random bending attack). sensitive

Existing studies on anti-shearing, large-scale rotation and Stirmark RBA attacks mainly focus on embedded watermarks, and there are not many zero-watermarking researches in this area.

Images