Anisotropic Diffusion for SAR Image Speckle Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Synthetic aperture radar (SAR) images suffer from speckle noise, leading to lower resolution compared to electro-optical (EO) images, and existing methods for despeckling and resolution enhancement often blur important features or require complex processing.
Innovation Solution
A computer-implemented method that adjusts an anisotropic diffusion algorithm based on noise thresholds determined through statistical analysis or Fourier windowing, applied to both real and imaginary components of SAR data, to enhance resolution while preserving scene content and edges.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If anisotropic diffusion algorithm is applied to remove speckle noise, then noise reduction is improved, but image resolution and edge sharpness deteriorate
Solution Approach 1:
The patent applies dynamic anisotropic diffusion where the diffusion coefficient is not fixed but adapts locally based on the statistical properties of the speckle noise. The algorithm dynamically adjusts the diffusion strength in different regions of the image, allowing aggressive noise removal in homogeneous areas while preserving edges and fine details where the diffusion coefficient is reduced.
Solution Approach 2:
The patent changes the parameters of the diffusion algorithm by incorporating noise threshold values derived from statistical analysis of the SAR data. By adjusting the diffusion coefficient as a function of local noise characteristics, the algorithm achieves optimal balance between noise removal and detail preservation for each specific region of the image.
2Object-affected harmful factors
If low pass filters such as Taylor weighting are applied, then noise is reduced, but scatterers become blurred together resulting in reduced resolution
Solution Approach 1:
The patent implements local quality by applying different processing strengths to different regions of the image based on local noise characteristics. The diffusion coefficient is calculated locally for each pixel neighborhood, allowing the algorithm to apply strong smoothing in noisy homogeneous regions while maintaining high resolution in regions containing scatterers or edges.
3Object-affected harmful factors
If apodization is applied to remove main and side lobes, then interference is reduced, but the detected image acquires a grainy appearance and binary look
Solution Approach 1:
The patent uses dynamic adaptation where the diffusion process responds to local image characteristics rather than applying a fixed apodization window. The algorithm dynamically determines the appropriate level of smoothing based on local noise statistics, avoiding the artificial binary appearance caused by fixed apodization while still reducing interference.
Data Source
AI summary
A computer system for processing interferometric synthetic aperture radar (SAR) images includes a database for storing SAR images to be processed, and a processor for processing interferometric SAR images from the database. The processing includes receiving first and second complex SAR data sets of a same scene, with the second complex SAR data set being offset in phase with respect to the first complex SAR data set. Each complex SAR data set includes a plurality of pixels. An interferogram is formed based on the first and second complex SAR data sets for providing a phase difference therebetween. A complex anisotropic diffusion algorithm is applied to the interferogram. The interferogram includes a real and an imaginary part for each pixel. A shock filter is applied to the interferogram. The processing further includes performing a two-dimensional variational phase unwrapping on the interferogram after application of the shock filter.