Research On Automatic Extraction Method Of Road Traffic Markings Information Based On Close-Range Photogrammetry

With the rapid development of new industries such as "Internet Plus",intelligent transportation and autonomous driving system,people have higher requirements for vehicle navigation.On one hand,the High Precision Vehicle Navigation can reduce the pressure of drivers and optimize the driving path.On the other hand,it can improve traffic efficiency,reduce traffic accidents and promote urban management.The road navigation map data is the core component of the vehicle navigation.Its quality is directly related to the whole navigation application.Nowadays,traditional navigation maps are gradually unable to meet the demand in terms of accuracy and contents.To overcome the drawbacks of traditional maps,high precision road navigation maps have been proposed.Compared to traditional maps,they have higher accuracy,can provide richer and more detailed contents,and can reflect road specific information more clearly.Meanwhile,they can provide data support for high-level applications,such as vehicle position and path plan of autonomous driving.Road traffic markings are basic and essential data of road navigation map.How to obtain its information is an important focus.However,relying on the collection of aerospace images,the update cycle is too long and the details of markings are not obvious.Relying on the traditional artificial collection,it is time-consuming and laborious,and easy to be affected by other aspectsTherefore,we research on the road traffic markings,and put forward a new method to automatically extract information of road traffic marking based on vehicle-borne close photogrammetry technology in views of its efficient and accurate collection.Firstly,referring to?GB5768-2009 Road traffic signs and markings?,we analyze the characteristic of the road traffic markings,and build a set of rules to extract the information of the traffic markings,combining the characteristic with the knowledge of azimuth calculation and coordinates calculation.After that,we research on the detection and positioning of the seed point of the road traffic marking in the images.Then,the real geographical coordinates of the feature points can be calculated through vehicle stereo images.Finally,the extraction system of traffic line information is developed.Based on Visual C#development language,this paper builds a traffic line information extraction system with Arc Engine secondary development component.Based on the experimental data of vehicle-borne sequence images in Nanjing,we compare the human-computer method with the method proposed in this paper in terms of acquisition efficiency,accuracy and geometric similarity.The results show that the proposed method improves the collection efficiency and accuracy,and can well deal with complex environment such as occlusion and strong light,meanwhile the road traffic markings collected have better geometric characteristics and more regular shapes.