Coupled variational image decomposition and restoration model for blurred cartoon-plus-texture images with missing...

Area of Science:

• Computer Vision
• Image Processing
• Numerical Analysis

Background:

• Image restoration is crucial for applications like surveillance and medical imaging.
• Existing methods often struggle with images containing both sharp features (cartoon) and fine details (texture), especially when pixels are missing.

Purpose of the Study:

• To develop a robust image decomposition model for restoring blurred images with missing pixels.
• To separate images into cartoon and texture components for improved analysis and further processing.

Main Methods:

• A decomposition model is proposed, assuming images are a superposition of cartoon and texture.
• Total variation norm and its dual norm are used for regularization of cartoon and texture components, respectively.
• An efficient numerical algorithm based on splitting augmented Lagrangian methods is employed.

Main Results:

• The model successfully restores blurred images with missing pixels.
• The algorithm provides a simultaneous decomposition into cartoon and texture parts.
• Theoretical guarantees for the existence of a minimizer and algorithm convergence are established.

Conclusions:

• The proposed decomposition model offers a significant advancement in image restoration and component separation.
• The extracted cartoon and texture components can be effectively utilized in subsequent image segmentation and inpainting tasks.
• The numerical algorithm is efficient and theoretically sound, outperforming existing state-of-the-art methods in comparative studies.