Lightning Recognition on the Filtered Videos in the Dynamic Clouds

Color models' opportunities were explored for lightning flashes extraction in significant dynamic cloudiness. It was shown that the greatest efficiently in recognition is provided by combining Lab and LCH model-based features. The ranges of color channels are defined for lightning aureoles. Linear and quadratic filters were developed for aureoles detection at the series frames of the video from meteorological satellites. An analysis of their effectiveness was done. The lightning detection filter is based on the processing a current frame of the video and the filtered one containing lightning aureoles. The method and software for lightning extraction were developed based on the image filtering.