Fusion of Heterogenous Sensor Data in Border Surveillance.

Area of Science:

• Surveillance technology
• Sensor fusion
• Computer vision

Background:

• Wide area surveillance is crucial for border control, especially in challenging forested terrains.
• Single sensors like radar often produce false detections in dense vegetation.

Purpose of the Study:

• To develop an automatic detection system for forbidden areas in forested land borders.
• To overcome limitations of single-sensor systems in dense vegetation.

Main Methods:

• A multi-sensor fusion and tracking system was developed.
• Passive infrared detectors were combined with automatic person detection from thermal and visual video cameras.
• A weighted map approach with a rule engine was used for data association.

Main Results:

• The system effectively eliminates single-sensor false detections.
• Surveillance accuracy was enhanced by up to 50% in real-world tests.
• The approach was validated using data from the EU FOLDOUT project.

Conclusions:

• Multi-sensor fusion significantly improves the reliability of wide area surveillance in forested border regions.
• The proposed system offers a robust solution for automatic activity detection, reducing false positives and increasing accuracy.