Euler's elastica and curvature based model for image restoration.

Area of Science:

• Computer Vision
• Image Processing
• Mathematical Modeling

Background:

• Euler's elastica energy is crucial for various computer vision and image processing tasks.
• Image restoration often faces challenges with multiplicative noise, leading to artifacts like blockiness.

Purpose of the Study:

• To propose a novel variational model for multiplicative noise reduction in images.
• To combine Euler's elastica curvature with Weberized total variation (TV) regularization for enhanced image restoration.

Main Methods:

• Developed a novel minimization functional integrating Euler's elastica energy and Weberized TV regularization.
• Employed an implicit gradient descent scheme for efficient minimization.
• Applied the model to images corrupted with multiplicative noise.

Main Results:

• The proposed model effectively preserves image edges and reduces blocky artifacts in smooth areas.
• Experimental results show significant visual improvement and increased peak signal-to-noise ratio (PSNR).
• Outperformed existing partial differential equation (PDE)-based methods in restoration quality.

Conclusions:

• The combined Euler's elastica and Weberized TV variational model offers superior performance for multiplicative noise removal.
• This approach enhances image quality by balancing edge preservation and artifact reduction.
• The method provides a robust and efficient solution for image restoration applications.