Abstract
Abstract We propose the application of a phase-field framework for three-dimensional volume reconstruction using slice data. The proposed method is based on the Allen-Cahn and Cahn-Hilliard equations, and the algorithm consists of two steps. First, we perform image segmentation on the given raw data using a modified Allen-Cahn equation. Second, we reconstruct a three-dimensional volume using a modified Cahn-Hilliard equation. In the modified Cahn-Hilliard equation, a fidelity term is introduced to keep the solution close to the slice data. The numerical methods use a hybrid method and an unconditionally stable nonlinear splitting scheme. The resulting discrete equations are solved using a multigrid method. The experiments on synthetic and real medical images are performed to demonstrate the accuracy and efficiency of the proposed method.
Original language | English |
---|---|
Article number | 2215 |
Pages (from-to) | 115-124 |
Number of pages | 10 |
Journal | Computer Vision and Image Understanding |
Volume | 137 |
DOIs | |
Publication status | Published - 2015 Jun 2 |
Keywords
- Allen-Cahn equation
- Cahn-Hilliard equation
- Image inpainting
- Image segmentation
- Multigrid method
- Volume reconstruction
ASJC Scopus subject areas
- Software
- Signal Processing
- Computer Vision and Pattern Recognition