SAR Geolocation Error Detection Using CNN Reference Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional radar systems require impractically long antennas for high-resolution imaging, and synthetic aperture radar (SAR) images can suffer from geolocation errors due to timing inaccuracies, satellite orbit uncertainties, and elevation model errors.
Innovation Solution
A method using an artificial neural network, specifically a convolutional neural network (CNN), to detect geolocation errors in SAR images by comparing them with reference images, generating difference images, and applying criteria for tile selection and scoring to accurately classify errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional radar is used to capture high resolution images, then image resolution is improved, but antenna length becomes impractically long
Solution Approach 1:
The patent applies segmentation by dividing the imaging process into multiple passes or segments. Instead of requiring a single long antenna, the system uses a shorter antenna that makes multiple measurements along the flight path, collecting data in segments that are later combined to achieve high resolution imaging.
Solution Approach 2:
The patent transitions from a single-dimension (antenna length) solution to a multi-dimensional approach by combining measurements taken over time and space. The synthetic aperture is created by integrating data along the flight path dimension, effectively trading antenna length for a combination of time and spatial sampling.
2Ease of manufacture
If SAR images are captured using a shorter antenna, then antenna practicality is improved, but geolocation errors increase due to timing inaccuracies and orbit uncertainties
Solution Approach 1:
The patent introduces an intermediary processing stage that acts as a mediator between the raw SAR data and the final geolocation output. This intermediary process involves comparing the SAR image with reference images and using difference image analysis to detect and correct geolocation errors, thereby compensating for the inherent inaccuracies introduced by using a shorter antenna.
Solution Approach 2:
The patent implements a feedback mechanism where the system continuously compares SAR images against reference images, detects geolocation errors through difference analysis, and uses this information to correct subsequent imaging operations. This closed-loop feedback system progressively improves geolocation accuracy despite the limitations of the shorter antenna.
3Productivity
If SAR images are processed without error detection, then processing speed is improved, but analysis accuracy deteriorates due to undetected geolocation errors
Solution Approach 1:
The patent applies preliminary action by performing geolocation error detection as an early step in the processing workflow, before conducting detailed analysis. By generating difference images and identifying errors upfront, the system prevents propagation of errors through subsequent analysis stages, thereby maintaining both efficiency and accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Methods, systems, and techniques for detecting geolocation error in a synthetic aperture radar (SAR) image. A SAR image purportedly depicting the geographical area is obtained. At least one reference image of the geographical area is also obtained. Data based on the SAR image and the at least one reference image are input into an artificial neural network trained as a classifier to determine that the SAR image and the reference image are of different areas, which results in a finding that the SAR image suffers from geolocation error.