THE RECOGNITION AND TRACKING OF SEVERE CONVECTIVE CLOUD FROM IR IMAGES OF GMS

摘要

It is very important to locate and track weather systems which cause severe calamity,such as severeconvective clouds (SCC),in nowcasting. In this paper the recognition and tracking of SCC is studied withGMS IR images using computer image techniques. As an IR image preprocessing, a SCC futerlng algorithm is put forward that combines a segment smoothing filtering and a removal procedure by thresholds. To the filtered SCCs the T algorithm and IP algorithm of contour coding method are applied to extract the contour line and its initial point. The description of SCCs includes four characteristic quanti-ties, i. e. center of gravity, cloud size, moment invariant M and R-shaped descriptor. Pattern recosnitionand pattern matching techniques are used to track the SCCs. Two procedures of rough and fine matchingare given. The former procedure include the setting of searching area and recognition of area and the latter is composed by the matching of shape descriptor R and moment invariant M and the analysis of correlative brightness temperature analysis.