Modigari Narendra,M. L. Valarmathi,L. Jani Anbarasi,Amir H. Gandomi
Abstract:Watermarking is one of the crucial techniques in the domain of information security, preventing the exploitation of 3D Mesh models in the era of Internet. In 3D Mesh watermark embedding, moderately perturbing the vertices is commonly required to retain them in certain pre-arranged relationship with their neighboring vertices. This paper proposes a novel watermarking authentication method, called Nearest Centroid Discrete Gaussian and Levenberg–Marquardt (NCDG–LV), for distortion detection and recovery using salient point detection. In this method, the salient points are selected using the Nearest Centroid and Discrete Gaussian Geometric (NC–DGG) salient point detection model. Map segmentation is applied to the 3D Mesh model to segment into distinct sub regions according to the selected salient points. Finally, the watermark is embedded by employing the Multi-function Barycenter into each spatially selected and segmented region. In the extraction process, the embedded 3D Mesh image is extracted from each re-segmented region by means of Levenberg–Marquardt Deep Neural Network Watermark Extraction. In the authentication stage, watermark bits are extracted by analyzing the geometry via Levenberg–Marquardt back-propagation. Based on a performance evaluation, the proposed method exhibits high imperceptibility and tolerance against attacks, such as smoothing, cropping, translation, and rotation. The experimental results further demonstrate that the proposed method is superior in terms of salient point detection time, distortion rate, true positive rate, peak signal to noise ratio, bit error rate, and root mean square error compared to the state-of-the-art methods.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to embed watermarks in 3D mesh models to protect their copyright and authenticity, while ensuring that the embedding process has minimal visual impact and can resist various attacks (such as smoothing, cropping, translation, and rotation). Specifically, the paper proposes a new deep - learning technique based on the nearest - centroid discrete Gaussian geometry and the Levenberg - Marquardt method for watermark authentication of 3D mesh models. This method selects salient points and uses the multi - function centroid for watermark embedding, thereby achieving efficient data embedding while maintaining the integrity of the model. In addition, this method also uses the Levenberg - Marquardt deep neural network to extract watermarks to ensure good imperceptibility and robustness during the extraction process.
### Main Contributions
1. **Salient Point Detection**: Use the Nearest - Centroid and Discrete Gaussian Geometry (NCDG) method to efficiently identify salient points in 3D models, and use Gaussian curvature to overcome the challenges brought by data discretization.
2. **Map Segmentation**: Achieve map segmentation through the planar partition objective function, minimizing the distortion in the watermark embedding process while segmenting the salient point areas.
3. **Watermark Embedding**: Use the multi - function centroid method for watermark embedding to ensure effective embedding without compromising the integrity of the model.
4. **Watermark Extraction**: Use the Levenberg - Marquardt deep neural network for watermark extraction, and use the nonlinear characteristics of the optimized kernel to ensure error stability and weight boundedness.
### Method Overview
1. **Salient Point Detection**:
- Use the nearest - centroid technique to identify the center points in the mesh as potential salient points.
- Apply the discrete Gaussian kernel function to calculate the local vertex density distribution, highlighting important areas.
- Quantify the saliency of each candidate point through the 3D salient point counting function, based on its distance from dense vertex clusters and its contribution to shape uniqueness.
2. **Map Segmentation**:
- After obtaining the salient points, use the planar partition map function to divide the 3D mesh model into different sub - regions.
- Each sub - region contains approximately \(\frac{1}{n}\) of the green partition and \(\frac{1}{n}\) of the blue partition.
3. **Watermark Embedding**:
- Use the multi - function centroid method to embed watermarks in each segmented region.
- Calculate the displacement factor \(l\) to adjust the centroid position in the mesh.
4. **Watermark Extraction**:
- Use the Levenberg - Marquardt deep neural network to extract the embedded 3D mesh image from the re - segmented regions.
- Analyze the geometric features through Levenberg - Marquardt back - propagation to extract watermark bits.
### Formulas
- **Right - angled Plane Estimation**:
\[
n^T\begin{bmatrix}p - p_v\\q - q_v\\r - r_v\end{bmatrix}=0
\]
where \((p, q, r)\) represents the coordinates of vertex \(v\), and \((p_v, q_v, r_v)\) represents the vertex normal vector.
- **Average Distance of the \(k\)-th Ring Neighborhood Vertices**:
\[
\text{Dis}_{ij}=n^T\begin{bmatrix}p_{ij}\\q_{ij}\\r_{ij}\end{bmatrix}-n^T\begin{bmatrix}p_v\\q_v\\r_v\end{bmatrix}
\]
where \((p_{ij}, q_{ij}, r_{ij})\) represents the \(j\)-th coordinate in \(V_i(k)\).
- **Discrete Gaussian Kernel Function**:
\[
M_\alpha(p, q, r)=M(p, q, r)*DG(p, q, r, \alpha)
\]
\[
DG(p, q, r, \alpha)=\sum_{n = -\infty}^{\infty}f(p - n)f(q)