SAR Image Clustering for Accurate Phase Statistics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The calculation of coherence matrices in synthetic aperture radar (SAR) image processing becomes inaccurate when pixels with different average and variance properties are mixed, leading to reduced reliability in displacement and elevation analysis.
Innovation Solution
An image processing device and method that specify phases of sample pixels, cluster them based on phase correlation, and calculate phase statistic data for each cluster, ensuring accurate coherence matrix generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If pixels with different average and variance properties are mixed in coherence matrix calculation, then the calculation can be performed using all pixels, but the accuracy of phase statistic data deteriorates
Solution Approach 1:
The patent segments pixels into multiple clusters based on their statistical properties (average and variance of phase). Instead of treating all pixels uniformly, the system divides the pixel population into distinct groups that share similar statistical characteristics, then calculates coherence matrices separately for each cluster. This segmentation resolves the contradiction by maintaining calculation efficiency within each homogeneous cluster while improving overall accuracy through property-matched pixel grouping.
Solution Approach 2:
The patent applies local quality by recognizing that different pixels have different statistical properties and treating them accordingly. Each pixel or pixel group is assigned specific weights or processing parameters based on its local statistical characteristics (average phase and variance). This allows the coherence matrix calculation to adapt to local conditions, improving accuracy without sacrificing the ability to process the entire image dataset.
2Quantity of substance
If all pixels are used for coherence matrix calculation, then the quantity of data increases, but the reliability of displacement and elevation analysis deteriorates
Solution Approach 1:
The patent segments pixels into clusters based on statistical homogeneity, then processes each cluster separately. This segmentation allows the system to maintain a large quantity of measurement points (all pixels are still used) while improving reliability by ensuring that pixels with similar statistical properties are grouped together for coherent analysis. The clustering approach prevents mixing of statistically incompatible pixels that would degrade analysis reliability.
Solution Approach 2:
The patent changes the parameter of pixel selection from uniform inclusion to selective inclusion based on statistical properties. By modifying how pixels are selected and weighted based on their average and variance characteristics, the system maintains comprehensive data coverage while improving the statistical validity of the coherence matrix, thereby enhancing analysis reliability.
Data Source
AI summary
The image processing device 10A includes phase specifying means 11 for specifying a phase of a sample pixel from a plurality of SAR images, clustering means 12 for generating a plurality of clusters by clustering the sample pixels based on correlation of phases of a pair of the sample pixels in the SAR image, and phase statistic data calculation means 13 for calculating phase statistic data capable of grasping a phase statistic regarding the pixel for each of the clusters.