Research On Road Network Model Based On Road Extraction Information From Remote Sensing Images

Road is one of the most important elements in geographic information.According to statistics,80% of human actions are related to location information and traffic elements.At the same time,road is also one of the geographical elements that change rapidly in the real world.With the rapid development of remote sensing technology and earth observation technology,remote sensing image has become the most direct and effective means to obtain geographic information.However,how to accurately and effectively obtain road and other key geographic information from remote sensing image and quickly map has become a bottleneck in the practical application of remote sensing technology.There are two key problems involved,one is fast extraction,the other is fast mapping.The focus of this paper is how to realize the rapid road network mapping.The rapid road network mapping here refers to the rapid construction of vectorized road network with complete topological relationship,and the analysis and application of road vector data such as path planning and multi-scale expression on the constructed road network map.In this context,this article has launched a series of studies on the basis of studying domestic and foreign related documents and research,the main contents are as follows:(1)Remote sensing image road extraction binary image vectorization processing.The vectorization process proposed in this paper is mainly aimed at the road binary image extracted from remote sensing image,which is used for fast and automatic vector data extraction and data optimization.Aiming at the problem of a large number of redundant data points in extracting vector data,this paper improves the traditional Douglas puck algorithm.Because of different road shapes,it is obviously unreasonable to set a unified threshold for all roads.In this paper,different roads are set a threshold in line with their characteristics,and the error of self intersection is eliminated in the process of simplification.At the same time,aiming at the problem of zigzag fluctuation of road shape,this paper proposes a method based on B-spline curve for smooth fitting of road data.(2)Extracting and constructing a relatively complete and independent road network usually involves multiple high-resolution remote sensing images.In this paper,a road network vector data model is proposed,which can quickly and parallelly network roads.Using this model,the road network extracted from multiple images can be quickly combined into a complete large regional road network.Due to the limitation of plane enhancement,the road vector data extracted from remote sensing image is composed of nodes and arcs.Due to the limitation of node weight and steering,the road network based on nodes cannot be analyzed.Aiming at this characteristic,the road network is represented as an adjacency table based on arcs by using arcs and extended forward associated edges,The weight is added to the data structure of arc segment to support path analysis.Aiming at the problem that the road network vector data can not be expressed in multi-scale,the road network vector data model proposed in this paper improves the road storage structure,adopts the multi-scale storage structure to store road information,and supports the multi-scale expression of the road.(3)Based on the above algorithm and model,the road network application system based on remote sensing image is designed.The system has basic GIS function,and integrates the algorithm and model proposed in this paper,and realizes the automatic mapping process from remote sensing image to road network.