Multispectral sensors provide valuable spectral information suitable for a variety of remote sensing applications, especially when cost-effectiveness, portability, and energy efficiency are essential, such as on small tactical drones. Since all materials have unique spectral signatures, this spectral information can be utilized to detect camouflaged targets in military reconnaissance scenarios. However, varying environmental conditions like vegetation and weather in real-world reconnaissance scenarios, and the computational complexity introduced by the large number of bands, make successful camouflage detections difficult. To address these issues, a novel sensor-managed anomaly detection method is proposed. This approach determines a subset of sensor bands of a multispectral imaging system based on current environmental conditions and their ability to expose camouflaged targets. Instead of processing all available bands, a spectral anomaly detector then processes only this subset of bands. To evaluate the robustness of the proposed approach, four different spectral anomaly detection methods are employed and compared: the RX detector, LRX detector, AED detector, and LPD detector. Experiments on an extensive multi-spectral dataset for camouflage detection show nearly equivalent detection performance when spectral anomaly detectors are applied to the managed subset compared to all sensor bands. This demonstrates the effectiveness of the proposed approach in reducing computational complexity while maintaining detection performance, which is particularly important for small tactical drones in military reconnaissance scenarios.
Sensor-Managed Anomaly Detection for Camouflage Detection in Airborne Multispectral Imagery
2024-03-02
3256663 byte
Conference paper
Electronic Resource
English
De-Shadowing of Satellite/Airborne Multispectral and Hyperspectral Imagery
British Library Conference Proceedings | 2005
|Synthetic imagery to simulate camouflage effectiveness
Tema Archive | 1989
|Multispectral Imagery Applications
NTRS | 2017
|Multispectral Imagery Applications
NTIS | 2017
|Multispectral Neural Network Camouflaged Vehicle Detection Using Flight Test Imagery
British Library Conference Proceedings | 1994
|