Three-dimensional mesh model blind watermarking method based on image discrete cosine transformation

Abstract

The invention relates to a three-dimensional mesh model blind watermarking method based on image discrete cosine transformation. The three-dimensional mesh model watermarking method involves selecting two stable geometrical characteristics to facilitate ordering to obtain a unique three-dimensional vertex sequence and adding watermark data to the texture coordinates of a vertex so that the vertex sequence with a watermark embedded therein can also be correctly obtained even after affine transformation, and the watermark can be correctly extracted accordingly. By using the method provided by the invention, original grids and pretreatment are unnecessary for watermark extraction. Multiple attacks can be resisted, including rotation, zooming, translation, noise and vertex reordering. At the same time, an algorithm also has a quite large watermarking capacity because when a carrier image is taken as the watermark for adding to a three-dimensional mode, the watermark transparency of the carrier image does not have to be considered, and thus the watermarking capacity is greatly improved.

Description

technical field [0001] The invention specifically relates to a three-dimensional mesh model blind watermarking method based on image discrete cosine transform. Background technique [0002] Digital watermarking is a technology that utilizes the redundancy commonly found in digital products to hide watermark information in it to achieve the purpose of copyright protection. With the development of technologies such as computer graphics, virtual reality, and computer-aided design, 3D models are used more and more, and they also need to be protected. However, most of the current digital watermarking technologies are aimed at media data types such as still images, video streams, and audio streams. There are relatively few researches on watermarking technology for 3D geometric model data. [0003] Compared with image watermarking, 3D models not only contain geometric information, but also topological information. The usual attacks on 3D models include rotation, translation, unif...

AI Technical Summary

Problems solved by technology

It is precisely because of these characteristics of the 3D model itself and the diversity of its attack methods that the embedding and extraction of watermarks are more complicated.

All of these make the traditional image watermarking method cannot be simply applied to the 3D geometric model

3D model data is composed of elements such as points, lines, and planes. These elements can be combined into various data expression methods. In addition, there is no fixed sorting standard for each element set of 3D model data, but for still images, it can be sorted according to The plane position of pixels is sorted, and the audio stream and video stream data can be sorted according to the time axis. For this irregular data type, various existing transform domain watermarking methods cannot be simply applied, and it is necessary to find an appropriate one that can reflect Parameters of 3D model data features for various transform domain watermarking methods

Audio and video data are arranged in time order, still images and video frames are arranged in scan line order, but 3D models do not have a fixed data order

[0006] (2). There is no clear concept of sampling rate, and the data in the three-dimensional surface model does not have convenient mathematical tools like images, audio, and video (such as cosine transform, Fourier transform, wavelet transform, etc.) that can be used

[0007] (3). The model data not only contains geometric information but also topological information, which makes the synchronization problem during watermark extraction more complicated.

[0008] (4). The representation method is not unique

The same 3D model can be represented by many different models, and it is easy to cause attribute loss in the process of converting between different models

Images