Related Experiment Video
Updated: Aug 10, 2026
06:03
AMEBaS: Automatic Midline Extraction and Background Subtraction of Ratiometric Fluorescence Time-Lapses of Polarized Single Cells
Published on: June 23, 2023
Kuan-modified speckle reducing anisotropic diffusion: A hybrid Bayesian/non-Bayesian method for despeckling SAR
Abhishek Tripathi1, Manoj Hudnurkar1
1Symbiosis Centre for Management and Human Resource Development (SCMHRD), Symbiosis International (Deemed University), Pune, Maharashtra, India.
Methodsx
|August 9, 2026
Summary
This study introduces a novel hybrid algorithm combining Kuan MMSE and anisotropic diffusion filters to reduce speckle noise in SAR images. The new method effectively balances noise reduction and edge preservation, outperforming traditional filters.
Area of Science:
- Remote Sensing
- Image Processing
- Signal Processing
Background:
- Speckle noise degrades Synthetic Aperture Radar (SAR) image quality, hindering radiometric and textural feature extraction.
- Traditional despeckling methods present a trade-off between noise reduction and edge preservation.
- Bayesian filters (e.g., Kuan) smooth homogeneous regions but blur edges, while PDE-based filters (e.g., AD) preserve edges but are computationally intensive and unsuitable for multiplicative noise.
Purpose of the Study:
- To develop and evaluate a hybrid algorithm for speckle noise reduction in SAR images.
- To combine the strengths of Kuan MMSE and anisotropic diffusion filters using a window-based classification approach.
- To provide a method that improves both noise reduction and edge preservation compared to existing filters.
Main Methods:
- A hybrid algorithm was developed by combining the Kuan Minimum Mean Square Error (MMSE) filter and anisotropic diffusion (AD).
- The combination was achieved via window-based classification guided by the conduction function of the Speckle Reducing Anisotropic Diffusion (SRAD) method.
- Explicit analytical expressions for filter parameters, including effective number of looks, image/speckle CoVs, weight coefficient, and edge threshold, were derived.
Main Results:
- The developed hybrid filter demonstrated improved performance in speckle reduction while preserving image edges.
- Comparative analysis using Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) showed superior results against Lee, Frost, Kuan (adaptive), and Perona-Malik filters.
- Performance was evaluated across a speckle variance range of 0-0.1 on two different SAR test images.
Conclusions:
- The proposed hybrid despeckling method offers an effective solution for speckle noise in SAR imagery.
- The algorithm successfully integrates Bayesian and PDE-based filtering approaches to overcome individual limitations.
- The findings suggest this hybrid approach provides a better balance between noise suppression and detail preservation for SAR image analysis.
