Dynamic Early Warning Method of Meteorological Disasters Based on Dynamic Region of Interest and Big Data
In order to reduce the loss of meteorological disasters,a dynamic early warning method of meteorological disasters based on dynamic regions of interest and big data is proposed.This paper builds a dynamic early warning framework for mete-orological disasters in the client/server mode.The system collects meteorological-related information and disaster subject infor-mation,and processes them by the data processing layer.The fuzzy C-means clustering algorithm is improved to extract the dy-namic area of interest of the satellite cloud image,determine the meteorological disaster scene and the possible affected area.Three improved deep shrinkage self-encoding networks are used to extract the meteorological factors,environmental factors,and disaster-affected subject factors in the area of interest respectively.feature,and then construct a Softmax classifier for scene recognition,determine the degree of correlation between the scene and various disaster-causing factors,and combine the support vector machine to determine the level and type of meteorological disasters to achieve dynamic early warning of meteoro-logical disasters.The experimental results show that the clustering effect of this method is outstanding,and it can identify the types and grades of meteorological disasters.
dynamic region of interestbig datameteorological disasterdynamic early warningdisaster causing factor
万方数据
维普
