Research On The Intelligent Selection Model Of OSM Road Network Based On Deep Learning

The road network is an element with a wide coverage in the map.Compared with other map element data,the road network is used more frequently,and has a more important significance in military and economic aspects.In the field of cartographic synthesis,the selection of road networks has always been one of the key points and difficulties of map scholars.The manual extraction of the road network is not only complicated,but also takes very large workload.Therefore,experts and scholars have been seeking breakthroughs in intelligent selection of road network in recent years.With the advent of the computer age,electronic maps have emerged,and the development of intelligent methods has brought new vitality to the intelligent selection of road networks.In today's world,road construction is developing rapidly,the number of electronic maps is increasing,and road information is rapidly updated,which brings new challenges to the selection of road networks.As the typical representative of electronic maps and VGI(Volunteered Geographic Information)data,OSM(Open Street Map)data is updated quickly at low cost,and with good quality.Generating the OSM road network data into standard map data is of great significance for the application of electronic map product updates,providing city maps,and increasing OSM data.Based on the OSM road network data,this paper establishes a road network selection parameter system,constructs a cyclic neural network of the deep learning framework,and finally intelligently selects the 1:250000 OSM road network in Shanghai.The main research contents of this paper are as follows:(1)Construct knowledge system.The knowledge system is a quantitative indicator of the importance of the road and plays a key role in the selection of the road network.According to the intelligent method and road network characteristics adopted in this paper,the parameter system of OSM road network intelligent selection is constructed.After the experiment,according to the experimental results,the parameter system was improved and the experimental results were improved.(2)Construct the intelligent selection model.Using machine learning to learn the rules of road network selection to realize the intelligent selection of road network,it is necessary to construct a good selection model.The deep learning method uses less samples to extract data features accurately and quickly,and learns the rules of experience to achieve road network selection.This paper combines the OSM data to construct a circular network method in the deep learning framework,to intelligently select the road network.(3)Select experimental data.The OSM road network has a high level of detail and a complex structure,which is an important way to acquire and update contemporary map data.Using OSM road network data as the data source can not only provide realtime high-quality data for this paper,but also provide new ideas for the application of OSM data.With the OSM road network data,the deep learning intelligent selection model is used to automatically select the road network.The experimental results show that the cyclic network in deep learning has a better selection effect for the automatic selection of OSM road network.The multi-parameter system is beneficial to improve the selection effect of the road network.The accuracy of the experimental results increased after adding three topological parameters.The RNN model solves the problems existing in the current road network selection,for example,the problem of large amount of road data,complicated extraction operation.At the same time,the RNN model describes the relationship between the roads within the road network in detail,and can achieve a higher selection effect.After the OSM data is selected by the intelligent method,it is more in line with the standard map data specification,and the application scope will be more extensive.