The ability to accurately characterize either fixed objects or moving targets is a critical need for security, defence and intelligence purposes. Spaceborne Synthetic Aperture Radar (SAR) systems are revolutionizing this domain, and advancements in Inverse SAR (ISAR) techniques offer unprecedented improvements in moving target identification and analysis.
Current EO systems can provide very high-resolution (VHR) imagery multiple times a day, offering detailed insights into security-sensitive locations. These systems play a vital role in detecting and monitoring activities and potential threats in strategic and tactical operations. However, identifying and characterizing moving targets remains a challenge due to SAR imaging constraints, necessitating innovative solutions for enhanced motion tracking and classification.
The EO4Security ISAR project has pioneered the use of Inverse SAR (ISAR) techniques for spaceborne SAR data to refine moving target analysis. Traditional SAR imaging relies on platform motion to focus the image, under the assumption of a static observation scenario. Moving objects appear defocused and misplaced on the image. ISAR, in contrast, leverages object motion to refocus images and extract key motion parameters, enabling more accurate results.
Advancements in ISAR processing
Modern SAR systems providing higher resolutions and enabling longer integration times reveal limitations in traditional ISAR methods. Innovative techniques were explored in the project, and tested for various use cases to overcome such limitations:
- The application of Back Projection (BP) as an alternative to conventional Image Contrast Based Algorithm (ICBA).
- The integration of autofocus algorithms, particularly the Ash autofocus method.
- The implementation of polarimetric target decomposition.
Marine targets uncovered
The combination of BP and the Ash autofocus method proved to yield the best performance, particularly for manoeuvring targets and/or when the target is observed for a longer time. For a quantitative reference, using COSMO-SkyMed Second Generation data, the image contrast improvement with respect to simply applying the ICBA method amounts to 34.4% on maritime targets.
The polarimetric target decomposition greatly contributed to suppressing the main sea clutter power, thereby enhancing the Signal-to-Noise Clutter Ratio (SNCR).
The enhanced vessel detection and monitoring capabilities for both still and moving targets in coastal and offshore marine areas, even in challenging conditions, achieved thanks to these innovative techniques, was proved very robust by an extensive validation.
The processing chain was in fact tested across diverse maritime scenarios, including varying sea states and vessel movements.
Land targets challenges
While ISAR has shown strong results in maritime scenarios, terrestrial environments pose additional challenges due to complex background clutter and interference from static objects. The technology has demonstrated potential in refocusing land-based moving targets, such as trains, but further research is needed to generalize the methodology for vehicles, trucks, and other dynamic land-based entities. Surveillance potentialities are dramatically more limited on land when compared to marine scenarios.
Integrating Artificial Intelligence in ISAR for motion parameters extraction
AI integration within ISAR has opened new possibilities for automated motion parameter extraction. A convolutional neural network (CNN) was developed to analyse complex ISAR data and retrieve absolute motion values in a geodetic reference system, centred on the target and aligned with the geographic north. Although tested on a controlled scenario (i.e. single imaging collection mode, single acquisition geometry) with limited target types (i.e. vessel moving in a port area), the AI model exhibited promising accuracy and rapid processing times, enabling near real-time decision-making capabilities.
Future directions and impact
To enhance AI-based ISAR solutions, future efforts will focus on:
- Expanding the training dataset to include diverse target types and acquisition geometries.
- Testing AI models on real-world operational data to validate their robustness.
- Integrating ISAR-AI capabilities within additional maritime domain awareness scenarios for enhanced security monitoring.
The successful development and deployment of ISAR techniques, bolstered by AI advancements, have the potential to transform EO-based intelligence, improving the detection, classification, and monitoring of moving targets in both maritime and terrestrial environments. These innovations will empower defence, security, and intelligence agencies with faster and more precise situational awareness.
Featured image : COSMO-SkyMed Second Generation image over Rotterdam. Photo credit: © Italian Space Agency. Processed and distributed by e-GEOS