A visual search model, VIDEM, has been formulated for predicting the detectability of a single, unknown target in an unstructured surround. The intended application is aircraft detection. The model consists of four components: a liminal contrast threshold, a frequency-of-seeing curve, a soft shell search representation, and discrete cumulation of single glimpse detection probabilities. The formulation was developed by registering five existing models against three controlled search experiments. The five models used represent all appropriate laboratory threshold data, including those of Blackwell, Lamar, Sloan, and Taylor. The search experiments included a large set of aircraft field tests, with precise photometric target measurements correlated to the detection events. The model registrations were done using nonlinear parameter estimation techniques and by comparing model predictions to actual event cumulatives with the Kolmogorov-Smirnov statistic. The resultant VIDEM model is a derivative of Sloan's data, cast into the popular visual lobe equations of Lamar.
Computing the cumulative distribution function of the Kolmogorov–Smirnov statistic
