Data Fusion for Space Object Classification using Human-Quality Artificial Intelligence

Stottler Henke developed a sensor data fusion system using low-level pixel processing techniques to process actual infrared and radar images for better correlation, classification, and lethality discrimination for ballistic missile defense targets. The system applies artificial intelligence techniques to extract features from raw sensor data and reason about those features and their associations with hypothetical objects.

Specifically, this project implements human-quality reasoning on sparse, physics-based object features extracted from sensor data. The employment of human-quality reasoning methods results in superior performance relative to other systems utilizing the same sensor data. The techniques and software were proven with actual X-band and IR sensor data for relevant IR/radar targets and achieved excellent performance results on this real data.

​