Feature object extraction applied to activity-based intelligence threat understanding

Abstract

Activity-based intelligence, the assessment of both the situation and potential threats, relies on several sources of information. Some of these sources provide information, such as time and position, that is easily represented mathematically. Other pieces of information may come in the form of patterns or imagery. The intelligence problem eventually requires a human analyst. However, the ever-increasing amount of data that must be investigated can overwhelm the analyst and, thus, necessitates the development of automated tools to process the information. An intelligent evidence accrual technique is proposed as a means to combine information in order to identify potential threats. The technique is able to provide both a measure of evidence and a level of uncertainty for the state. The approach is based upon a fuzzy Kalman filter that fuses several types of observations to monitor activities over a duration of time and provide an assessment of the situation. A data-to-image mapping and correlation that allows the combination of disparate elements is incorporated.