Second-order TGV model for Poisson noise image restoration

Area of Science:

• Image processing
• Computer vision
• Applied mathematics

Background:

• Poisson noise is a common challenge in image restoration.
• Total variation (TV) models are widely used but have limitations.
• Second-order regularization methods offer potential improvements.

Purpose of the Study:

• To propose a novel image restoration model based on second-order total generalized variation (TGV) regularization.
• To enhance the performance of existing regularization techniques for Poisson noise.
• To improve both visual quality and objective evaluation indexes in image restoration.

Main Methods:

• Developed a new image restoration model incorporating quadratic regularization into the TGV framework.
• Employed the split Bregman iteration algorithm for efficient model solving.
• Conducted experiments to evaluate the model's effectiveness.

Main Results:

• The proposed model effectively addresses Poisson image restoration challenges.
• Significant improvements were observed in visual restoration effects.
• Objective evaluation indexes demonstrated superior performance compared to existing methods.

Conclusions:

• The novel second-order TGV regularization model provides a robust solution for Poisson image restoration.
• The split Bregman iteration algorithm is suitable for solving the proposed model.
• This approach offers a promising direction for advanced image restoration techniques.