Evaluating artifact effects in threat image projection for X-ray baggage screening and strategies for improvement

Abstract

Many airports worldwide employ threat image projection (TIP) that occasionally projects prerecorded threat items (e.g., bombs) into X-ray images of passenger baggage to enhance and monitor security officers’ detection performance. However, some TIP-generated images appear artificial, potentially undermining the effectiveness of TIP. We conducted a simulated screening task with security officers using TIP images from an international airport. We applied path modeling to analyze how TIP artifacts affect threat detection and interact with other image characteristics. We found that threats aligning poorly with other items increased threat detection. Furthermore, artifacts occurred more often when the threat items were projected onto less complex bags or loose items in a tray. TIP should still be effective, because artifacts only occurred in every sixth image and showed a small impact on threat detection compared to other factors. To increase TIP effectiveness, projection onto baggage images with moderate complexity can significantly reduce artifacts while producing more realistic threat images.