Statistical Modeling & Training Algorithms
The Air Force required statistically robust methods to compare the response of substances for security screening equipment.
ILW developed a statistical model and training algorithms to validate explosives and identify image features in security screening equipment.
- Identified key image features which influence explosive detection (i.e. density, atomic number, texture, patterns)
- Created robust experimental design processes
- Repeatability and efficiently identifying objects and signatures in images across modalities
- Defined statistically sound hypothesis tests for multiple features and system noise
- Improved detection performance
- Training algorithms
- X-ray, CT, and millimeter wave (MMW)
- Python, OpenCV, MATLAB and Octave