An On-Demand Construction Method Of Disaster Scenes For Multitype Users

China is a disaster-prone country.There are more than one million casualties caused by sudden natural disasters each year,and the comprehensive economic losses have reached hundreds of billions of CNY,scientific and effective disaster prevention and mitigation is of great significance to the protection of people's lives and property.Disaster prevention should be transformed from a single type of disaster to comprehensive reduction,and from passive response to active defense,and the aim is to improve the comprehensive disaster prevention capabilities.As an important supporting technology of integrated disaster reduction,disaster virtual scenes in 3D not only can intuitively and efficiently represent and share disaster information knowledge,but are also very important for assisting disaster prediction,judgment,planning,and decision-making.However,the current research still have certain limitations.In terms of disaster scene construction,the definition of the relationship between scene objects is not clear,the construction process lacks a detailed semantic description and guidance mechanism,preventing full consideration of the diferent needs of multitype users involved in disaster management;regarding scene visualization,the representation emphasizes visual effects,resulting in semantic information scarcity and low efficiency of disaster scene recognition and disaster information transmission,which is not conducive to standardizing the process of scene construction and supporting efficient analysis.To address above problems,this paper proposes an on-demand construction method of disaster scenes for multitype users.Key technologies and methods including the creation of a knowledge graph of disasters for multitype users,calculation of semantic relevanc,fusion modeling of disaster scenes with spatial semantic constraint and suitability representation are discussed in detail,so as to achieve the rapid and dynamic construction of disaster scenes in3 D for different needs of multitype users,improving the efficiency of disaster scenes visualization and disaster information transmission.The following is an overview of this dissertation.(1)A disaster scene knowledge graph for multitype users was created to achieve the specific definition and clear description of multitype users and disaster scene objects,as well as the efficient management of complex semantic relationships between users and scene objects.First,we introduced the structure of the knowledge graph of disaster scenes,the characteristics and relationship attributes of the multitype users,disaster scene components and their relationships were analyzed and summarized.Based on the seven-step method,a construction method of disaster scene ontology was proposed,and the ontology was used as the schema layer of knowledge graph;Then,the method of extracting the feature relationship of disaster scene entities was described,and the construction and formal expression of disaster scene knowledge graph was realized.(2)An on-demand fusion modeling method of disaster scene guided by user needs was proposed,the demand relationship between users and scene objects was quantitatively described through semantic relevance calculation,and the fusion of the disaster scene objects oriented to user needs was realized.First,utilizing the circulation and orientation of knowledge graph,the personalized Page Rank algorithm was adopted to calculate the semantic relevance between users and scene objects based on the disaster scene knowledge graph,and form a recommendation set;Second,the fusion process of disaster scenes was introduced from spatial semantic constraint rules and fusion modeling operations;Finally,the constraint rules such as spatial lcoation,attribute category and spatial topology were abstracted by analyzing the characteristics of disaster data,which were used to guide the construction of disaster 3D scenes.(3)The suitability representation method of disaster scenes for user perception was discussed,through the visualization of self-explanatory symbols and photorealistic scene cooperation and the dynamic augmented representation of the whole disaster process,the efficiency of disaster recognition and disaster information transmission were improved.First,to comprehensively express the disaster information,a multitype augmented representation framework of disasters was proposed;second,a visualization method based on selfexplanatory symbols and photorealistic scene cooperation was proposed,it can reveal more disaster semantic information while ensuring a certain degree of realism;finally,the semantic augmentation of scene objects with various visual variables combination was adopted to obtain more semantic information of disasters,and using a “storytelling” method to realize the dynamic augmented representation of the whole disaster process.(4)By integrating the preceding research achievements,we developed a prototype system for disaster scene fusion and visualization,two typical debris flow disasters were selected as the case areas for experimental analysis,including the creation of disaster scene knowledge graph,calculation of semantic relevance,fusion visualization,and dynamic augmented representation of the whole disaster process.Experimental results show that the proposed method can normalize the semantic relationships between multitype users and scene objects in a formal way,the 3D scenes of disasters driven by the knowledge graph can reduce the complexity and difculty of the modeling process while satisfying the diverse needs of multitype users,the suitability representation can effectively improve disaster scene cognition and disaster information transmission efficiency.