Edge-Preserving Denoising of Image Sequences

Area of Science:

• Remote Sensing
• Medical Imaging
• Image Processing

Background:

• Satellite imagery (e.g., NASA Landsat) and functional magnetic resonance imaging (fMRI) provide valuable time-series data.
• Image sequences are crucial for studying Earth's surface changes and brain function.
• Noise and contaminations in image sequences hinder reliable analysis.

Purpose of the Study:

• To address the underexplored area of image sequence denoising.
• To propose an effective edge-preserving denoising procedure for image sequences.

Main Methods:

• Developed an edge-preserving image denoising method.
• Employed a jump-preserving local smoothing procedure.
• Optimized bandwidth selection to account for spatio-temporal correlations.

Main Results:

• The proposed method successfully preserves edges in denoised images.
• Numerical studies validate the effectiveness of the denoising technique.
• Theoretical arguments support the method's performance.

Conclusions:

• The suggested image sequence denoising method is effective.
• The approach is suitable for various applications, including remote sensing and medical imaging.
• This work contributes a novel solution to image sequence denoising challenges.