Abstract:A finite element-based image segmentation strategy enhanced by an anisotropic mesh adaptation procedure is presented. The methodology relies on a split Bregman algorithm for the minimisation of a region-based energy functional and on an anisotropic recovery-based error estimate to drive mesh adaptation. More precisely, a Bayesian energy functional is considered to account for image spatial information, ensuring that the methodology is able to identify inhomogeneous spatial patterns in complex images. In addition, the anisotropic mesh adaptation guarantees a sharp detection of the interface between background and foreground of the image, with a reduced number of degrees of freedom. The resulting split-adapt Bregman algorithm is tested on a set of real images showing the accuracy and robustness of the method, even in the presence of Gaussian, salt and pepper and speckle noise.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is: **How to improve the accuracy and robustness of image segmentation when dealing with image inhomogeneity and noise?** Specifically, the author proposes a method that combines anisotropic mesh adaptation and a region - based Bayesian segmentation model to better capture the complex spatial information in the image and reduce the number of degrees of freedom required to describe the contour.
### Specific Problems and Methods in the Paper
1. **Challenges in Image Segmentation**:
- Images usually contain inhomogeneous spatial patterns, which pose challenges to traditional segmentation methods.
- Image noise (such as Gaussian noise, salt - and - pepper noise, and speckle noise) will further affect the accuracy of the segmentation results.
2. **Proposed Solutions**:
- **Bayesian Energy Functional**: A Bayesian - based energy functional is introduced, which takes into account the spatial information of the image and ensures that inhomogeneous spatial patterns in complex images can be identified.
- **Anisotropic Mesh Adaptation**: Mesh adaptation is driven by anisotropic recovery error estimation, which ensures clear detection of the interface between the background and the foreground and reduces the number of degrees of freedom required.
- **Split Bregman Algorithm**: It is used to minimize the energy functional. This algorithm effectively solves the optimization problem through a splitting strategy and performs particularly well when dealing with regularization functionals.
### Formula Summary
- **Bayesian Energy Functional**:
\[
F_{\text{BAY}}(\varphi, \tilde{p}_I, \tilde{p}_E) := -\int_\Omega [\log(\tilde{p}_I(U(x))) + \log(\tilde{p}_E(U(x)))] \, dx + \nu \int_\Omega g(U(x)) |\nabla \varphi(x)| \, dx
\]
where $\tilde{p}_i(U(x))$ is the probability density function of pixel intensity, $\nu$ is the regularization parameter, and $g(U(x))$ is the edge detection function.
- **Anisotropic Recovery Error Estimation**:
\[
\eta^2 = \sum_{K \in T_h} \eta_K^2
\]
\[
\eta_K^2 := \frac{1}{\lambda_{1,K} \lambda_{2,K}} \left[ \lambda_{1,K}^2 r_{1,K}^T G_{\Delta K} r_{1,K} + \lambda_{2,K}^2 r_{2,K}^T G_{\Delta K} r_{2,K} \right]
\]
- **Optimal Eigenvalue Calculation**:
\[
s^\star_K = \left( \frac{\theta_1}{\theta_2} \right)^{1/2}, \quad r^\star_{1,K} = t_2, \quad r^\star_{2,K} = t_1
\]
\[
\lambda^\star_{1,K} = \theta^{-1/2}_2 \left( \tau^\star \frac{\|\nabla \varphi_h\|_2^2}{2 |\hat{K}|} \right)^{1/2}
\]
\[
\lambda^\star_{2,K} = \theta^{-1/2}_1 \left( \tau^\star \frac{\|\nabla \varphi_h\|_2^2}{2 |\hat{K}|} \right)^{1/2}
\]
### Conclusion
By combining anisotropic mesh adaptation and the Bayesian segmentation model, this method can provide more accurate and robust image segmentation results when dealing with image inhomogeneity and noise. The experimental results show that this method...