Concealed Weapon Detection

In order to detect concealed weapons, a sensor collects data that is viewed by the human observer. We propose in this work to investigate how the weaknesses of different sensors can be reduced through the processing of the collected data before it is viewed in order to enhance desired targets (e.g., weapons) and reduce undesired noise and clutter. Among the technologies to be investigated are the millimeter wave and infrared. Disadvantages shared among all sensors are: clothing penetration, range, and time needed to display the data. Because the disadvantages are mutually exclusive, the only way to satisfy all three of them is to either improve the sensors' hardware in collecting data or introduce signal processing procedures to enhance the data before it is displayed. We suggest doing the latter and use the Automated Statistical Characterization and Partitioning of Environments (A'SCAPE) processore which uses two separate procedures to determine all homogeneous regions and sub-regions in the scene. The first procedure is used to separate contiguous homogeneous regions by segregating between their power levels whereasa the second procedure separates contiguous homogeneous regions by segregating between their data distributions. One or more of these homogeneous patches would represent weapons.