Urban waterlogging monitoring and early warning based on video images

Urban flood disaster causes serious loss to urban residents. Timely access to urban waterlogging conditions has great significance for disaster prevention and disaster relief. Owing to the time resolution limitation of data, the traditional monitoring of urban flood disasters using remote sensing imagery cannot realise real-time automatic monitoring and continuous monitoring of key disaster areas. This paper selects road monitoring video, uses image difference operation and support vector machine (SVM) algorithm to identify the waterlogging area, and uses the region growing method to extract the waterlogging area range. The research results show that this method can be used for continuous monitoring and early warning of urban waterlogging in real time.