Lane-level guide line extraction method and system
An extraction method and a guide line technology, applied in the field of high-precision map production, can solve problems such as vehicle damage, inability to express vehicle lane change behavior well, inappropriateness, etc.
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0025] A lane-level guiding line extraction method, see figure 1 , the extraction method consists of:
[0026] S1. Obtain the crowdsourcing trajectory collected by the crowdsourcing collection vehicle, and filter some trajectories that obviously deviate from the road surface.
[0027] Specifically, considering that the lane guidelines are extracted from the trajectories of crowdsourcing vehicles, in order to avoid some accidental driving behaviors from interfering with the lane guideline extraction, it is necessary to filter some trajectories that deviate significantly from the road surface. The embodiment of the invention combines the linear characteristics of the trajectory and the driving direction characteristics of the road, calculates the similarity between the trajectories through FrechetDistance, and realizes the clustering processing of the trajectories through the Mean Shift clustering algorithm, and filters some isolated trajectories that obviously deviate from the ...
Embodiment 2
[0045] see figure 2 , shows the overall flow chart of a lane-level guiding line extraction method. First, the crowdsourcing trajectory is preprocessed, and the trajectory that obviously deviates from the road surface is eliminated; 100m, which is almost the length to complete a lane change) to interrupt the grouping, to avoid restricting the lane change behavior in the middle part of the trajectory when the trajectory is too long, and at the same time extend the segmented trajectory forward and backward for a certain distance K (about 30m); The lane guidelines are aggregated based on the segmented trajectories. At this time, there will be an overlapping area of length K between the guidelines before and after the segmentation. The straight lane guidelines basically overlap and can be directly fused. The lane guidelines with lane changes will be It presents an obvious feature of one change to many or multiple changes to one. The change feature point is also the lane change p...
Embodiment 3
[0047] A structure diagram of a lane-level guiding line extraction system, such as image 3 As shown, the lane-level guiding line extraction system includes an acquisition module filtering module 301, a segmentation module 302, a clustering and grouping module 303, an extraction module 304 and a fusion module 305, wherein:
[0048] The filtering module 301 is used to obtain the crowdsourcing trajectory collected by the crowdsourcing collection vehicle, and to filter some trajectories that obviously deviate from the road surface; the segmentation module 302 is used to filter the filtered trajectory according to the number of lane sidelines or lane change conditions. Segmentation is carried out to obtain a plurality of segmented trajectories; the clustering and grouping module 303 is used to cluster and group a plurality of segmented trajectories, and obtains the trajectories belonging to each lane; the extraction module 304 is used to divide and belong to the same The segmented...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More