Augmenting Synthetic Aperture Radar Data with Simulated Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a need for high-quality and sufficient Synthetic Aperture Radar (SAR) data to develop and test algorithms and train machine learning models, but real SAR images may not always be readily available in the required quantity and quality.
Innovation Solution
A computer-implemented method for augmenting real SAR data by combining it with defocussed additional SAR data, which may be simulated, to generate augmented SAR data that includes artificial features such as ambiguities and radio frequency interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If real SAR data is used for training machine learning models, then data authenticity is improved, but data quantity and variety are insufficient
Solution Approach 1:
The patent creates synthetic SAR data by copying and simulating the characteristics of real SAR data. A simulation system generates artificial SAR images that replicate the statistical properties, noise patterns, and feature distributions of authentic SAR data, providing abundant training data while maintaining realistic characteristics for machine learning model development
Solution Approach 2:
The patent employs parameter changes by systematically varying simulation parameters such as signal-to-clutter ratio, noise levels, target characteristics, and environmental conditions. This allows generation of diverse SAR data scenarios from a unified simulation framework, expanding data variety while preserving the fundamental statistical properties of real SAR data
2Manufacturing precision
If SAR data is processed to remove ambiguities and artifacts, then image quality is improved, but processing complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-processing SAR data through defocusing operations before training algorithms. This preliminary defocusing reduces the impact of ambiguities and artifacts in the training data, allowing algorithms to learn from cleaner representations while the full processing complexity is avoided during deployment
Solution Approach 2:
The patent introduces an intermediary processing step using defocused SAR data as a intermediate representation. This intermediary form serves as a bridge between raw SAR data with artifacts and fully processed images, enabling algorithm training on simplified data while maintaining the ability to handle complex real-world scenarios
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided here is a computer-implemented method, data, and a computing system for augmenting real synthetic aperture data, the method comprising obtaining defocussed real SAR data comprising data from imaging an area on the surface of the Earth, obtaining defocussed additional SAR data of one or more features which may be artificial or real, and combining the defocussed real SAR data with the defocussed additional SAR data of an artificial feature to obtain augmented defocussed SAR data.