How Image Based Factors and Human Factors Contribute to Threat Detection Performance in X-Ray Aviation Security Screening

Abstract

The present study examines the relative importance of a series of known and expected factors that highly affect threat detection performance in aviation security X-ray screening. Examined image based factors were threat item view difficulty, threat item superposition, bag complexity and bag size. Further, also the two human/demographic factors X-ray image interpretation training and age were examined. Image measurements and performance statistics for factors estimation are introduced. Three statistical approaches were applied in order to examine the impact of the introduced factors on threat detection performance and revealed consistent results. Bivariate correlations between detection performance and predictors/factors were analysed to estimate the isolated impact of each single factor independently of any other. Multiple linear regression analyses were applied for estimating the overall impact of all image based factors and human/demographic factors respectively. And analyses of covariance were applied in order to check for possible interaction effects between all factors of the models. All analyses were applied separately for the four threat item categories guns, knives, improvised explosive devices and other.