Lane line processing method and device

Abstract

Embodiments of the present invention provide a lane line processing method and device. The method includes: obtaining the distance between the lane line points included in the first image; using the distance between each lane line point to determine the direction density of each lane line point; using each of the lane line points The direction density of the points determines the lane line points included in the group corresponding to each lane line; using the lane line points included in each group to obtain the representation information of the lane line corresponding to each group. In the embodiment of the present invention, the lane line points in the image are grouped by using the directional density of each lane line point, and the obtained grouping is more accurate, so the fitted lane line is accurate, is not affected by the image quality, and has high robustness. Stickiness.

Description

technical field [0001] The invention relates to the technical field of traffic, in particular to a lane line processing method and device. Background technique [0002] Lane lines play an important role in automatic driving and can provide important information output for modules such as positioning and decision-making control. On the two-dimensional (2D, two dimensional) images captured by the camera, the traditional solution is to use image processing methods to detect the edges of lane lines, and then use clustering or other point grouping schemes to obtain lane lines. [0003] Using traditional image processing, deep learning and other methods for lane line detection or segmentation requires cumbersome post-processing solutions to output lane lines, and the effect is easily affected by image quality and other conditions. For example, traditional post-processing schemes can use Hough transform to post-process lane lines, which is easily affected by noise points and is no...

AI Technical Summary

Problems solved by technology

[0003] Using traditional image processing, deep learning and other methods for lane line detection or segmentation requires cumbersome post-processing solutions to output lane lines, and the effect is easily affected by image quality and other conditions.

For example, the traditional post-processing scheme can use the Hough transform to post-process the lane lines. This method is easily affected by noise points and is not robust.

Images