SAR Object Coherence Estimation with Geolocation-Aware Pixel Masking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for creating interferometric coherence data products for objects imaged by Synthetic Aperture Radar (SAR) face challenges with speckle noise, inaccurate coherence estimation for small and complex-shaped objects, and low absolute location accuracy, leading to unreliable data products.
Innovation Solution
A method and system that determine geocoded pixels within identifiable object geometries using known geolocation accuracy, compute backscatter intensity and interferogram components, and calculate coherence values and statistics for objects with small sizes and complex shapes, employing geocoded interferometric SAR images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If spatial averaging is done using a local neighborhood window (Boxcar averaging) to estimate coherence, then processing is simplified, but coherence estimation becomes distorted for small and complex-shaped objects because data outside objects are included
Solution Approach 1:
The patent applies local quality by defining object-specific spatial masks that selectively include only pixels corresponding to each individual object's geometry in the coherence estimation process. This allows different regions (inside vs. outside objects) to have different inclusion properties, thereby maintaining processing simplicity while improving coherence estimation accuracy for small and complex-shaped objects.
2Ease of operation
If existing techniques are used to produce coherence data products, then processing is straightforward, but reliability is poor for objects with low absolute location accuracy and low SAR resolution
Solution Approach 1:
The patent applies preliminary action by performing geolocation accuracy compensation and object geometry refinement before coherence estimation. The system pre-processes the SAR data by adjusting object boundaries according to known geolocation errors, ensuring that subsequent coherence calculations are performed on accurately defined object regions, thereby improving reliability without complicating the overall workflow.
3Productivity
If conventional coherence estimation methods are applied to small and complex-shaped objects, then processing time is reduced, but the produced coherence data products are meaningless and unreliable
Solution Approach 1:
The patent applies local quality by implementing object-specific spatial masks that precisely define which pixels contribute to coherence estimation for each object. This ensures that even for small and complex-shaped objects, only relevant pixels within the object boundaries are included in the calculation, maintaining processing efficiency while producing meaningful and reliable coherence data products.
Data Source
AI summary
A method for creating interferometric coherence data products for objects imaged by Synthetic Aperture Radar (SAR) having polarization(s). The method includes: for each identifiable object geometry for which there are acquired geocoded interferometric SAR images with flat-earth and topographic phase removed having following pixel values: backscatter intensity (V1) in polarization(s) for master image; backscatter intensity (V2) in polarization(s) for slave image; in-phase component (V3) of geocoded interferogram in polarization(s); and quadratic-phase component (V4) of geocoded interferogram polarization(s), determining which geocoded pixels are within identifiable object geometry considering known geolocation accuracy of identifiable object geometry and geocoded interferometric SAR images; and determining coherence values and statistics of coherence values for objects for polarization(s), based on pixel values V1, V2, V3, and V4 within identifiable object geometry.