An efficient method to remove mixed Gaussian and random-valued impulse noise

Area of Science:

• Computer Vision
• Image Processing
• Signal Processing

Background:

• Mixed Gaussian and Random-valued impulse noise (RVIN) poses significant challenges in image denoising.
• Existing algorithms often suffer from inadequate denoising performance and high computational complexity.

Purpose of the Study:

• To propose an efficient and effective method for removing mixed Gaussian and RVIN from digital images.
• To address the limitations of current denoising algorithms in terms of performance and computational cost.

Main Methods:

• The proposed method employs an improved "detecting then filtering" strategy combined with inpainting techniques.
• It involves two phases: noise classification (using Adaptive center-weighted median filter, three-sigma rule, and extreme value processing) and a three-step noise removal process (preliminary RVIN removal, Gaussian noise removal via Block Matching and 3D filtering (BM3D), and final RVIN removal).

Main Results:

• The algorithm successfully removes mixed Gaussian and RVIN, achieving superior quantitative and visual results compared to state-of-the-art methods.
• Experimental results demonstrate a significant reduction in computation time.

Conclusions:

• The proposed method offers an efficient and effective solution for mixed Gaussian and RVIN removal.
• It outperforms existing techniques in both denoising quality and computational speed, making it a valuable contribution to image processing.