Robust adaptive estimator for filtering noise in images

Area of Science:

• Image processing
• Computer vision
• Signal processing

Background:

• Additive noise significantly degrades image quality, impacting visual perception and subsequent analysis.
• Existing image restoration methods often struggle to balance detail preservation with effective noise suppression.
• Outlier noise presents a particular challenge for traditional denoising algorithms.

Purpose of the Study:

• To develop and evaluate novel filtering techniques for restoring images corrupted by additive noise.
• To enhance the performance of image denoising by improving both detail preservation and noise reduction.
• To address the limitations of current methods in handling both Gaussian noise and outliers.

Main Methods:

• Adaptive Mean Median Filter: Designed to preserve image details during restoration from additive Gaussian noise.
• Minimum-Maximum Method: Specifically developed to address and remove outlier noise.
• Robust Adaptive Mean P-Median Filter: A hybrid approach combining the strengths of the adaptive mean median and minimum-maximum methods.

Main Results:

• The adaptive mean p-median filter demonstrates superior performance in both detail preservation and noise suppression compared to existing methods.
• The robust adaptive mean p-median filter shows good performance even when dealing with images containing significant outlier noise.
• Comparative analysis with other median-based algorithms validates the effectiveness of the proposed methods.

Conclusions:

• The proposed adaptive mean p-median filters offer significant advancements in image restoration, particularly for noisy images.
• These novel methods provide robust solutions for preserving image fidelity while effectively removing various types of noise.
• The findings suggest these filters are valuable tools for applications requiring high-quality image reconstruction.