Sensor Fusion Algorithms and Performance Limits

The concealed weapons detection program involves auto-detection and recognition of weapons under clothing. Single sensor methods are insufficient for detection in many situations. This program develops methodologies for detecting weapons under clothing using multiple dissimilar imaging sensors. Several signal processing steps are considered starting with image acquisition and concluding with a binary or tertiary decision about the presence or absence of a weapon. Image registration, noise removal filtering, image fusion and shape recognition are all discussed. Individual attributes are used to register images which are then fused in the wavelet domain.