Page 1 Measurement Formulae for Image-Based Factors in X-ray Imagery Bolfing, A. & Schwaninger, A. University of Zurich Visual Cognition Research Group (VICOREG) Technical Report November 26, 2007 Point of Contact: Dr. Adrian Schwaninger, a.schwaninger@psychologie.uzh.ch Image Based Factors Schwaninger (2003b) and Schwaninger, Hardmeier, and Hofer (2005) have identified three image based factors which affect threat detection in x-ray screening significantly: View Difficulty, Superposition and Bag Complexity. For the development of a statistical model for automatic image difficulty estimation these concepts were implemented mathematically by image measurement formulae (Schwaninger, Michel, & Bolfing, 2007). Such a model is very valuable for x-ray image difficulty estimation to analyze detection performance of computer-based training (CBT) and threat image projection (TIP) data. Automated image measurements of view difficulty, superposition and bag complexity are also essential for the development of effective and efficient CBT algorithms (e.g. Schwaninger 2004b, 2005c). Moreover, such measurements are important for the development of reliable computer-based tests for measuring x-ray image interpretation competeny (e.g. Koller, Hardmeier, Michel, & Schwaninger, 2007, in press) and are very useful for the development of pre-employment assessment tests (e.g. Hardmeier, Hofer, & Schwaninger, 2006a).The following sections describe the theoretical concepts of the image based factors and updated formulae on how they can be measured automatically. FTI View Difficulty (FtiVD) Even with the aid of 3D volumetric models, it is not (yet) possible to satisfyingly determine the degree of a 3-dimensional rotation of a physical threat item automatically from its 2-dimensional x-ray image. Difficulties regarding image segmentation arise from the very heterogeneous backgrounds of x-ray images. Therefore, this image based factor is not (yet) being calculated directly by image measurements, but statistically from TIP detection performance. FTI View Difficulty as it is conceptualized and implemented here reflects image difficulty attributes unique to each exemplar view of an FTI and not solely the degree of rotation. The general FTI View Difficulty formula above describes a slight modification of the mean of the inverted detection performance value (DetPerf). Inverted means here, that the measured detection Bolfing, A., & Schwaninger, A. (2007). Measurement formulae for image-based factors in x- ray imagery. VICOREG Technical Report, November 26, 2007.