A Method Of Road Network Selection Considering Semantic Information

Map generalization refers to the operations of selecting,simplifying,summarizing,and coordinating the features in the map when generating a small-scale map from a large-scale map.Due to its complexity and diversity in expression,map generalization has always been a challenging research field in modern cartography.Road features often spread throughout the map and are one of the important geographic elements that cannot be ignored.Road generalization has always been a hot research topic in map generalization.Road selection is the basis for other operations in road generalization and is also a crucial step in the process.Previous road selection studies have considered the geometric and topological characteristics of roads,while in terms of semantic features,only information such as road rank and name has been considered,and such semantic information often exists in an incomplete or even completely missing state.With the development of spatiotemporal big data,there are more ways to obtain semantic information,and introducing relevant data into road network selection can provide important scientific basis and real-time reference.Therefore,this paper aims to address the lack of consideration of semantic features in road selection due to the incomplete and difficult-to-obtain road semantic data,fully utilizing the advantages of spatiotemporal big data,such as ease of access and large volume,and proposes a road network selection method that considers multiple features to improve the accuracy and rationality of road selection.This paper first processes and analyzes road traffic flow and related facility point inform ation under the support of taxi trajectory data and POI data,providing support for the measure ment of road semantic features.Based on this,traffic flow is introduced to construct road stro kes.Specifically,the road segment with the largest traffic flow in the untreated roads is select ed as the starting road segment,and then the roads within the range of the angle threshold are filtered out when multiple segments are associated with a node.Finally,the starting road seg ment is connected to the road that is most similar to the traffic flow,and the road segments ar e connected to form a stroke,to better maintain the overall coherence of the road.Finally,the importance of all strokes is evaluated and ranked based on multiple feature information,and r oad selection is based on the importance value of road network strokes.The main innovations of this paper are as follows:(1)Introducing taxi trajectory data to generate road network strokes.Existing studies mainly generate strokes based on the principle of good continuity,which takes into account the geometric features of roads but to some extent ignores their semantic information.Taxi trajectory data can reflect human activity patterns to some extent and have advantages such as good continuity,long operating time,and dynamic updates.The road traffic flow calculated from taxi trajectory data reflects the importance and actual connectivity of roads to some extent.Therefore,this paper supplements the road network semantic measurement indicators using taxi trajectory data to assist in constructing strokes.(2)This paper proposes a road selection and generation method that considers multiple features.Existing research has mainly focused on single semantic feature information,with insufficient consideration of the measurement factors of semantic information.Therefore,this paper introduces POI data while considering road geometric features,topological features,and traffic flow semantic information,enriching the semantic feature information.The Simpson index,a descriptor operator from the field of biology,is introduced as a road semantic feature,which intuitively reflects the vitality and importance of roads in people’s cognition,improving the rationality of road semantic feature measurement.This paper verifies and analyzes different types of road network data.The experimental results show that:(1)by introducing taxi trajectory data into the stroke generation process,the problem of selecting the starting road segment and geometric branching road segment that may arise in the existing stroke construction process is effectively solved,providing data support for subsequent road selection and ultimately obtaining determined selection results;(2)by introducing POI data,the probability of selecting important roads is greatly improved,and important roads can be preserved,making the selection results more reasonable.