Curvature-based interface restoration algorithm using phase-field equations

Area of Science:

• Image Processing
• Computational Geometry
• Differential Geometry

Background:

• Object distortions pose challenges in image analysis and restoration.
• Existing methods may struggle with preserving intricate details during reconstruction.

Purpose of the Study:

• To develop an effective restoration algorithm for distorted objects.
• To leverage curvature-driven flow for accurate shape reconstruction.

Main Methods:

• Utilizing mean curvature flow to capture object contours.
• Employing Allen-Cahn and Cahn-Hilliard equations for image restoration.
• Applying Dirichlet and Neumann boundary conditions for feature preservation and mass conservation.

Main Results:

• Successfully restored distorted half-circle and parentheses shapes into a perfect circle.
• Demonstrated the algorithm's ability to preserve non-distorted regions.
• Validated the effectiveness of curvature derivative analysis in shape selection.

Conclusions:

• The proposed curvature-driven flow algorithm offers a robust solution for object restoration.
• The method accurately reconstructs shapes by preserving essential features and mass.
• This technique shows promise for various image processing and computer vision applications.