Phase pattern denoising using a regularized cost function with complex-valued Markov random fields based on a...

Area of Science:

• Image processing
• Computational physics
• Applied mathematics

Background:

• Phase patterns are crucial in various scientific imaging techniques.
• Noise in phase patterns can obscure important details and hinder analysis.
• Existing denoising methods may struggle with preserving phase information or computational efficiency.

Purpose of the Study:

• To develop a simple and effective method for denoising phase patterns.
• To transform the image denoising problem into an energy diffusion problem.
• To preserve 2π phase jumps and minimize computational effort.

Main Methods:

• A discrete model is employed for phase pattern denoising.
• The method models image denoising as an energy diffusion problem in complex-valued fields.
• A cost function based on the discrete form of complex-valued Markov random fields is established.

Main Results:

• The proposed filtering method is easily implemented via an iterative approach.
• The method effectively preserves 2π phase jumps in the denoised patterns.
• The filtering process demonstrates low computational effort.

Conclusions:

• The presented discrete model offers a simple and effective solution for phase pattern denoising.
• The method's ability to preserve phase jumps and its computational efficiency make it valuable for scientific applications.
• The performance is validated using both simulated and experimental phase patterns.