Weather Radar Image Filtering with Fuzzy Features Using Non Linear Technique

Abstract

A fuzzy feature of weather radar echo was introduced as the measurement of horizontal texture of echo image, and the local mean of the second lowest level was taken as the measurement of vertical echo scale. An algorithm of weather radar base reflectivity noise filtering was proposed based on Support Vector Machine (SVM). The results show that the echoes of non-precipitation such as ground clutter (GC), clear sky, abnormal propagation (AP) can be removed effectively, especially for the mixing echoes of precipitation and non-precipitation coexisted together, the algorithm works pretty well to identify the two kinds of echoes that mean different weathers.2008 Congress on Image and Signal Processing 978-0-7695-3119-9/08 $25.00