The relevance of aviation security has increased dramatically in the last years. 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 the three image-based factors view difficulty, superposition and bag complexity on human detection of familiar prohibited items (knives) in x-ray images. In Experiment 1 we replicated earlier findings in order to provide converging evidence for the validity of these factors. In Experiment 2 we assessed the subjective perception of the same image-based factors. Twelve participants rated the x-ray images used in Experiment 1. Threat images were rated for view difficulty, superposition, clutter, transparency and general difficulty. Except for clutter ratings obtained in Experiment 2 were significantly correlated with detection performance in Experiment 1. We then developed statistical and imageprocessing algorithms to calculate the image-based factors automatically from x-ray images. In Experiment 3 it was revealed that most of our computer-generated estimates were well correlated with human ratings of image-based effects obtained in Experiment 2. This shows that our computer-based estimates of view difficulty, superposition, clutter and transparency are perceptually plausible. Using multiple regression analysis we could show in Experiment 4 that our computer estimates were able to predict human performance in Experiment 1 as well as the human ratings obtained in Experiment 2. Applications of such a computational model are discussed for threat image projection systems and for adaptive computer-based training.