Automated Detection and Cueing

The single most effective method to thwart security threats posed by people with hostile intent is to detect their presence early. Unfortunately, large scale standoff detection of humans is difficult. Various sensors, including still and video cameras, have been designed and applied to human detection, however, to date a person is required to be in the detection loop since robust, automated detection of humans with visual sensors has not yet been achieved. This is particularly unacceptable in military scenarios where the only people available to interact with the surveillance system are often occupied with other critical tasks. As a result, there is a clear need for a truly automated, robust human detection system that only interacts with a person when a threat has been detected. Such a system will provide soldiers with more complete situational awareness and allow them to perform their other activities with increased safety. PercepTek proposes to develop a system that will exploit algorithms that are insensitive to the many and varied image and video properties that are present in many different types of imagers. In cases where the algorithms can be affected by particular properties of the imager, compensating capability will be incorporated into the system.