Research On Semantic Information Enhancement Method Of Monomerization Model Of Oblique Photography

The technical mechanism of automatic tilt photography modeling determines that the output model is a continuous TIN(Triangulated Irregular Network)network,which cannot be selected or separated from the target individually and needs to be "monomerized".However,in the actual production,the rich semantic information of the building itself is mostly ignored in the "monomerization".However,due to the differences in the sources and semantic expressions of the geographic data,the same geographic entity may have different names,locations and classification systems.In addition,there are differences in the information richness and completeness of these data.Moreover,as the application fields and application groups of crowdsource geographic data continue to expand and the application levels continue to deepen,the uncertainty of the quality of crowdsource geographic data becomes more and more prominent.Therefore,matching the geographic data from many sources to obtain a semantically consistent dataset of geographic information from many sources can make the location information of geographic data from many sources more accurate and the attribute information more rich,and provide extensive and accurate basic data for the semantic information enhancement of building monolithic model.To summarize,this paper selects to take crowdsource geographic data as the research object,analyzes the temporal characteristics,spatial characteristics and semantic information characteristics of crowdsource geographic data,studies the matching method of crowdsource geographic data,constructs the matching model of crowdsource geographic data considering temporal characteristics,and obtains semantically consistent geographic information dataset by using the matching model.Then,we construct the semantic information classification table of monolithic buildings,design the semantic information based on City GML(City Geography Markup Language)detail level coupling scale and coupling rules,add multi-scale functional areas,and finally give semantic information to the monolithic model through the inclusion relationship between functional areas and buildings to realize the semantic information of tilt photography monolithic model.Enhancement.The main research contents of the paper are as follows:1)A multi-feature semantic similarity calculation model considering temporal features is constructed.To explore and analyze the matching method of crowdsource geographic data for the problem of matching crowdsource geographic data,a multi-similarity feature matching method is selected,and a multi-feature semantic similarity calculation model is constructed to accurately match crowdsource geographic data.Among them,the matching process of crowdsource geographic data depends on the selection of similarity features,and different combinations of similarity will lead to different matching results and accuracy.By selecting appropriate similarity features,the matching precision and accuracy can be improved.Therefore,the selection and combination of similarity features is a key part of the matching process of crowdsource geographic data.In this paper,we fully analyze the temporal features,spatial features and semantic information features of crowdsource geographic data,and select five similarity features: temporal similarity,spatial similarity,literal similarity,bag-of-words similarity and category similarity to participate in the matching calculation.Combination experiments are conducted with a multilayer perceptron to determine the weight of each similarity,avoiding the subjective factor of artificially assigned weights.Finally,a multi-feature semantic similarity calculation model that takes into account temporal features is constructed to achieve accurate matching of geographic data from many sources.2)Semantic information coupling rules based on City GML are designed to realize multiscale semantic information enhancement of monolithic models.Based on the semantically consistent geographic information dataset with many sources of geographic data,the multiscale semantic information enhancement method is studied.By analyzing and summarizing the five levels of details of City GML,five building semantic scales are divided according to the five levels of details of City GML,and the semantic information of single building is summarized and classified into variable quantitative semantic information,non-variable quantitative semantic information and qualitative semantic information.Then,the coupling scales and coupling rules of semantic information based on the five levels of details of City GML are designed,and finally,multi-scale functional areas are added,and the multi-scale semantic enhancement of the building monolithic model is finally realized by judging the relationship between buildings and functional areas.3)The algorithm implementation and platform construction are carried out based on the above research.In this paper,the algorithm implementation of the matching model of crowdsourced geographic data using Gaode and OSM data is carried out,and the query and analysis of model monolithic and multi-scale monolithic model semantic information is realized based on the digital campus platform.