Key Techniques For Generating Vectorized Road Network From Large-Scale Vehicle Trajectory

Rapid development of location services based on data is powered by the emerging mobile computing environment such as internet of vehicles, mobile social networks.There are a lot of challenges in location understanding ability with the increasing capacity to provide location data.And now ubiquitous mapping and location-based social awareness has been widely attention from both academic and industrial circles. Location data contains the geographic, trajectory, space media will become an important data resource.It can provide location services such as city operation, the life of individuals, enterprises economic aspects. Vehicle trajectory data which can reflect the user habit preference, became an important component element. This paper focuses on the research of related technology about generating vectorized road network from large-scale vehicle trajectory. So as to get rid of the dependence on the map service, constitutes a basic part of location data analytics.This paper tracks traffic data directly into a binary image, by such as morphology, gauss fillter image processing algorithm, the road surface track position point data discrete components, fitted to only a few data points (2/1000) composed of polygonal regions of the set of vertices. It saves the memory and store space, greatly improves the calculation performance analysis of relevant network.The key technologies of this paper as following:1). Vehicle trajectory data extraction and segmentation of data set.2).Select the best appropriate parameters.3).Data manipulation algorithms such as morphology, gauss filter, achieve denoising, wipe thorn, hole filling, smoothing etc..The contribution of this paper is mainly reflected in:1).The conversion of the problem domain.2).Local partitioned data set, using a statistical rule.3).The road network vector processing does not depend on the base map.