Verification of image processing based visibility models

Methods are presented for testing visibility models that simulated photographs to display results of model . An experimental protocol is developed and to obtain input data including standard photographs chosen scenes on a clear day and during a smog event Pasadena, CA. With the clear day photograph as a , pollutant properties measured on the smoggy are introduced into the visibility model, and results the model calculations are displayed as a synthetic of the expected appearance of the smog event. comparisons are made between the predicted actual appearance of the smog event. Diagnostic developed are applied to the visibility modeling proposed by Malm et al. That model is shown reproduce the contrast reduction characteristic of urban pollution but produces synthetic photographs with sky that differ substantially from a real photograph the actual smog event.