Low-frequency GPS track road network matching method based on multi-dimensional data fusion analysis

Abstract

A low-frequency GPS track road network matching method based on multi-dimensional data fusion analysis comprises the steps that firstly, a candidate point set of each GPS track point is calculated, the matching degree of adjacent candidate points is considered from multiple dimensions such as space, time and environment, multi-dimensional data fusion analysis is conducted, and the final matching degree is formed; then, a global static voting matrix is constructed and compressed, low-correlation-degree abnormal track points are removed, meanwhile, factors such as signal timing are comprehensively considered, and a dynamically-optimized local weighted voting matrix is generated; and finally, path generation and candidate point voting are carried out, the candidate path with the highest voting value is selected as a final matching path, and road network matching of the GPS track is completed. The method is suitable for an urban road network, considers the space-time relationship and environmental influence of adjacent track points, carries out multi-dimensional data fusion analysis, eliminates low-correlation abnormal points, and improves the accuracy of road network matching. Meanwhile, dynamic factors such as signal timing and the like are considered, and the road network matching precision is further improved.

Description

technical field [0001] The invention relates to a GPS track road network matching method for intelligent transportation. The GPS track road network matching can not only be used for vehicle navigation, road planning, traffic flow prediction, etc., but also can be used for various location services and mobile social networks. Background technique [0002] In recent years, the development of cloud computing and the Internet of Things has driven the development of the urban Internet of Vehicles. The detection instruments of many networked vehicles generate and upload massive amounts of data in real time, triggering a global wave of research on "data cities". Currently, we are in a transitional stage from traditional information-based transportation to an intelligent and interconnected society. Limited by cost factors, while installing and deploying various advanced detector equipment, it is necessary to be compatible with early data acquisition equipment with low quality, high ...

AI Technical Summary

Problems solved by technology

Compared with the first type of method, the second type of method takes into account the relationship between the road network topology and GPS track points, which greatly improves the matching accuracy, but still requires higher sampling accuracy of GPS track points, and cannot directly detect abnormalities. GPS track data

[0004] At present, the existing low-frequency GPS trajectory road network matching methods have the following main problems: 1) The first type of method does not comprehensively consider the spatial correlation, temporal correlation and environmental correlation between adjacent GPS trajectory points, resulting in low The existence of outliers in the degree of correlation reduces the matching accuracy of the road network and is not suitable for complex urban road networks; 2) Although the second type of method considers factors such as space and time, it does not consider factors such as signal timing of urban roads. It also does not eliminate the interference of abnormal trajectory points, which affects the final road network matching accuracy

Images