Color image watermark embedding and detection method based on quaternion PHT synchronous correction

Abstract

The invention discloses a color image watermark embedding and detection method based on quaternion PHT synchronous correction, comprising: converting color RGB images into a YUV space, and respectively selecting appropriate embedding intensities to embed watermarks; defining the quaternion PHT of the color images; training an LS-SVR model, constructing a training sample set, performing circumcircle quaternion PHT decomposition on each training image, extracting a specific moment value as the feature vector of the training image, and using a relevant information coefficient to describe each training sample; utilizing circumcircle quaternion PHT to decompose an images to be detected, and decomposing an obtained moment value as a sample feature vector; utilizing a training model to train the feature vector of the image to be detected, solving a geometric transformation parameter, and utilizing the geometric transformation parameter to perform synchronous correction on the image be detected to obtain a corrected image; and extracting the watermarks in a corresponding mode of the embedded watermarks.

Description

technical field [0001] The invention relates to a color image watermarking method, in particular to a color image watermark embedding and detection method based on quaternion PHT synchronous correction capable of resisting conventional attacks and geometric attacks. Background technique [0002] Due to the characteristics of easy duplication, modification and transmission of digital media, there are serious piracy, infringement and leaking phenomena in the use of digital assets, which greatly damages the relevant rights and interests of digital content providers and digital asset owners. It also has a great impact on information security. In order to solve the problem of copyright protection of digital media, digital watermarking technology has been introduced. [0003] In recent years, the research on digital image watermarking technology has made great progress, and a series of digital image watermarking methods have been proposed successively. But unfortunately, most of...

AI Technical Summary

Problems solved by technology

But unfortunately, most of the existing digital image watermarking methods are based on grayscale images, and at the same time, they focus on the research against conventional signal processing (such as lossy compression, low-pass filtering, noise interference, etc.), and such as rotation , scaling, translation, row and column removal, shearing, and other geometric attacks have poor resistance

Images