| In recent years, aviation security and the detection of threat items has become increasingly important. Multiple technologies, methods and procedures, including x-ray screening of passenger bags, are utilized to help ensure threats do not enter the civil aviation system. Security inspection is a form of visual inspection that presents a dynamic selection of potential threats which are often concealed within the image. A lot of visual search and inspection literature exists analyzing and describing the general inspection process, however it has not been shown how the unique features of security inspection fit into these more general models. Drury (1975) and Spitz and Drury (1978) developed equations that divide the total inspection time into the functional components of search- and decision-time for a general inspection task. This provides a simple way to understand how search and decision components change in relation to outside variables (training, age, work experience, gender, etc.). The current research tested the applicability of this two-component inspection model to x-ray screening of bags, and utilized the model to determine how an individually adaptive computer-based training program affected the search and decision parameters within the x-ray screening task. Peformance data from 416 airport screeners evaluating 40 x-ray images of passenger bags was obtained and analyzed. R-square values averaging above 0.8 were obtained showing the applicability of the model to a security x-ray task and showing that this task has similar characteristics to other inspection tasks. Improvements in both search and decision time for threats present (hits) was found. No effects for false alarms or changes in stopping time were seen. This model allows the locus of inspection error to be determined, beyond just classification into misses and false alarms, and provides an interesting alternative to more labor intensive research methods, such as eye movement recordings. |
