Threat Recognition from Gait Analysis

Recent threats to public safety, such as terror attacks, mass shootings, and suicide bombings have made it very clear that law enforcement needs better tools for the protection of innocent people. The ability to identify and neutralize a target before it becomes a threat is a widely sought-after tool by law enforcement and military agencies. This paper proposes using a combination of existing technologies to identify threats prior to an incident, giving law enforcement time to deescalate or remove a threat before harm is done. Threat detection is accomplished by imaging the target area and using gait analysis to determine the emotional state of all people in the area. The gait analysis is enhanced using gait energy images in order to complete data acquisition from subjects with items that normally interfere with data acquisition. Data complexity can be reduced using techniques such as PCA, MDA, ACDA, or another reduction method. Each target is assigned a membership or classification rating in a given emotional class from collected data. The emotional classification values are then input to a neural network to determine whether the current subject matches the emotional characteristics of a perpetrator preparing to attack. Subjects who match sufficiently are flagged for further review by a human operator. This tool will help law enforcement or military agencies do more to protect the public without large additions in training or manpower.