A statistical approach for image difficulty estimation in x-ray screening using image measurements

Authors:
Adrian Schwaninger;Stefan Michel;Anton Bolfing
Affiliations:
University of Zurich, Switzerland;University of Zurich, Switzerland;University of Zurich, Switzerland
Venue:
Proceedings of the 4th symposium on Applied perception in graphics and visualization
Year:
2007

Citing 0
Cited 4

Improving search task performance using subtle gaze direction

Proceedings of the 5th symposium on Applied perception in graphics and visualization
How Image Based Factors and Human Factors Contribute to Threat Detection Performance in X-Ray Aviation Security Screening

USAB '08 Proceedings of the 4th Symposium of the Workgroup Human-Computer Interaction and Usability Engineering of the Austrian Computer Society on HCI and Usability for Education and Work
Search task performance using subtle gaze direction with the presence of distractions

ACM Transactions on Applied Perception (TAP)
An image retrieval approach to setup difficulty levels in training systems for endomicroscopy diagnosis

MICCAI'10 Proceedings of the 13th international conference on Medical image computing and computer-assisted intervention: Part II

Quantified Score

Hi-index	0.00

Visualization

Abstract

The relevance of aviation security has increased dramatically at the beginning of this century. One of the most important tasks is the visual inspection of passenger bags using x-ray machines. In this study, we investigated the role of image based factors on human detection of prohibited items in x-ray images. Schwaninger, Hardmeier, and Hofer (2004, 2005) have identified three image based factors: View Difficulty, Superposition and Bag Complexity. This article consists of 4 experiments which lead to the development of a statistical model that is able to predict image difficulty based on these image based factors. Experiment 1 is a replication of earlier findings confirming the relevance of image based factors as defined by Schwaninger et al. (2005) on x-ray detection performance. In Experiment 2, we found significant correlations between human ratings of image based factors and human detection performance. In Experiment 3, we introduced our image measurements and found significant correlations between them and human detection performance. Moreover, significant correlations were found between our image measurements and corresponding human ratings, indicating high perceptual plausibility. In Experiment 4, it was shown using multiple linear regression analysis that our image measurements can predict human performance as well as human ratings can. Applications of a computational model for threat image projection systems and for adaptive computer-based training are discussed.