Investigation on the possibilities of tracking dynamic visibility changes in images

This material describes a methodology for tracking the dynamics of weather phenomena in computer images. The investigation is based on parameters used in the currently available techniques for determination of fog, rain, snow and sun. The developed methodology relies on the, mean value, dispersion and correlated color temperature by processing the dark prior channel of colored images. Tracking is done for a set of data derived from restricted fields with different positioning in one image. The occurrence of an event is analysed on in sequences of images obtained in different weather phenomenon. The research is aimed at improving the processing of images at monitoring traffic.