Systems and methods for high throughput anomaly detection with 3-d millimeter wave imaging

Abstract

Systems and methods for detecting anomalies in an envelope including scanning an envelope moving along a conveyor with multiple millimeter wave imaging sensor modules and collecting reflection measurements from the sensor modules. The systems and methods can further include constructing a one or more voxel thick horizontal 3D image of the envelope and then creating an image of the envelope comprising the magnitude of the mean of the one or more voxel layers. The image of the envelope can be fed into a binary logistic regression classifier to provide a probability that the envelope contains an anomaly from the classifier. The systems and methods can include causing the envelope to be diverted from the conveyor for further inspection when the probability is greater than a selected threshold.

Dependencies are inferred from the claim text.