Method and apparatus for detecting lane line, and medium

Patent metadata

Field	Value
Publication number	US-10846543-B2
Application number	US-201816230998-A
Country	US
Kind code	B2
Filing date	Dec 21, 2018
Priority date	Dec 29, 2017
Publication date	Nov 24, 2020
Grant date	Nov 24, 2020

Abstract

According to the exemplary embodiments of the present disclosure, a method and apparatus for detecting a lane line, and a medium are provided. A method for generating a lane line detection model includes: detecting a lane line in an original image to generate a first image associated with the detected lane line; acquiring a second image generated based on the original image and associated with a marked lane line; generating at least one tag indicating whether the detected lane line is accurate, based on the first image and the second image; and training a classifier model for automatically identifying the lane line, based on the first image and the at least one tag. In such case, the lane line detection may be achieved in a simple and effective way.

First claim

What is claimed is:

1. A method for generating a lane line detection model, comprising:
detecting a lane line in an original image to generate a first image associated with a detected lane line;
acquiring a second image generated based on the original image and associated with marked lane line;
generating at least one tag indicating whether the detected lane line is accurate, based on the first image and the second image; and
training a classifier model for automatically identifying the lane line, based on the first image and the at least one tag.

2. The method according to claim 1 , wherein the generating a first image comprises:
performing an inverse perspective transformation on the original image; and
detecting the lane line in an inverse-perspective transformed original image, to generate the first image.

3. The method according to claim 1 , wherein the generating a first image comprises:
performing gray processing on the original image to generate a grayed original image;
binarizing the grayed original image, to generate a binary image; and
detecting the lane line in the binary image to generate the first image.

4. The method according to claim 1 , wherein the generating a first image comprises:
denoising the original image to generate a denoised image; and
detecting the lane line in the denoised image to generate the first image.

5. The method according to claim 1 , wherein the generating a first image comprises:
applying a contour detection on the original image to generate a contour of the lane line; and
generating the first image based on the contour.

6. The method according to claim 5 , wherein the generating the first image based on the contour comprises:
performing curve fitting on the contour to generate a curve representing the lane line; and
generating the first image by mapping the curve to the original image.

7. The method according to claim 1 , wherein the generating at least one tag comprises:
dividing the first image into a first set of image blocks, wherein each of the image blocks includes a portion of the detected lane line;
dividing the second image into a second set of image blocks, wherein each of the image blocks includes a portion of the marked lane line; and
generating a plurality of tags for a plurality of portions of the detected lane line by comparing corresponding image blocks in the first and second set of image blocks, wherein each of the tags indicates whether a corresponding portion of the detected lane line is accurate.

8. A method for detecting a lane line, comprising:
detecting a lane line in an original image to generate a first image associated with the detected lane line; and
inputting the first image into the classifier model according to one of claims 1 - 7 , to automatically identify the lane line.

9. An apparatus for generating a lane line detection model, comprising:
at least one processor; and
a memory storing instructions, the instructions when executed by the at least one processor, cause the at least one processor to perform operations, the operations comprising:
detecting a lane line in an original image to generate a first image associated with a detected lane line;
acquiring a second image generated based on the original image and associated with marked lane line;
generating at least one tag indicating whether the detected lane line is accurate, based on the first image and the second image; and
training a classifier model for automatically identifying the lane line, based on the first image and the at least one tag.

10. The apparatus according to claim 9 , wherein the generating a first image comprises:
performing an inverse perspective transformation on the original image; and
detecting the lane line in an inverse-perspective transformed original image, to generate the first image.

11. The apparatus according to claim 9 , wherein the generating a first image comprises:
performing gray processing on the original image to generate a grayed original image;
binarizing the grayed original image, to generate a binary image; and
detecting the lane line in the binary image to generate the first image.

12. The apparatus according to claim 1 , wherein the generating a first image comprises:
denoising the original image to generate a denoised image; and
detecting the lane line in the denoised image to generate the first image.

13. The apparatus according to claim 9 , wherein the generating a first image comprises:
applying a contour detection on the original image to generate a contour of the lane line; and
generating the first image based on the contour.

14. The apparatus according to claim 13 , wherein the generating the first image based on the contour comprises:
performing curve fitting on the contour to generate a curve representing the lane line; and
generating the first image by mapping the curve to the original image.

15. The apparatus according to claim 9 , wherein the generating at least one tag comprises:
dividing the first image into a first set of image blocks, wherein each of the image blocks includes a portion of the detected lane line;
dividing the second image into a second set of image blocks, wherein each of the image blocks includes a portion of the marked lane line; and
generating a plurality of tags for a plurality of portions of the detected lane line by comparing corresponding image blocks in the first and second set of image blocks, wherein each of the tags indicates whether a corresponding portion of the detected lane line is accurate.

16. An apparatus for detecting a lane line, comprising:
at least one processor; and
a memory storing instructions, the instructions when executed by the at least one processor, cause the at least one processor to perform the operations according to claim 8 .

17. A non-transitory computer readable storage medium storing a computer program, wherein the program, when executed by a processor, cause the processor to perform the operations according to claim 1 .

18. A non-transitory computer readable storage medium storing a computer program, wherein the program, when executed by a processor, cause the processor to perform the operations according to claim 8 .

Assignees

• Baidu online network technology beijing co ltd

Inventors

• Duan Xiong
• Lang Xianpeng
• Zhou Wang
• Yan Miao
• Zhan Yifei
• Ma Changjie
• Jin Yonggang

Classifications

• G06V10/774

  Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06V10/82

  using neural networks · CPC title

• G06V20/588 Primary

  Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road · CPC title

• G06V10/28 Primary

  Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns · CPC title

• G06F18/2414

  Smoothing the distance, e.g. radial basis function networks [RBFN] · CPC title