Method and apparatus for extracting lane line and computer readable storage medium
A method and an apparatus for extracting a lane line, a device, a computer readable-storage medium and a collection entity are provided. The method includes: obtaining a first group of lane lines of a road based on a first image generated from a point cloud collected by a laser radar; obtaining a second group of lane lines of the road based on a second image collected by a camera; and determining a lane line set of the road based on the first group of lane lines and the second group of lane lines.
1. A method for extracting a lane line, comprising:
obtaining a first group of lane lines of a road based on a first image generated from a point cloud collected by a laser radar, comprising:
obtaining a plurality of lane lines based on a plurality of images generated from the point cloud collected by the laser radar, in which, the plurality of images comprises the first image; and
determining the first group of lane lines based on an optimization for the plurality of lane lines by: organizing a plurality of reflection value images according to a world coordinate system of a physical world, connecting disconnected lane lines by using a relationship between lane lines in neighboring images in the plurality of images;
obtaining a second group of lane lines of the road based on a second image collected by a camera; and
determining a lane line set of the road based on the first group of lane lines and the second group of lane lines by:
generating a first group of two-dimensional lane lines and a second group of two-dimensional lane lines by projecting a first group of three-dimensional lane lines and a second group of three-dimensional lane lines to a plane where the road is located;
performing a geometric transformation on the second group of two-dimensional lane lines, in which, the geometric transformation comprises at least one of a rotation and a translation;
determining one or more lane lines present in the second group of two-dimensional lane lines subjected to the geometric transformation and absent from the first group of two-dimensional lane lines; and
determining the lane line set by combining the first group of two-dimensional lane lines with the one or more lane lines.
2. The method according to claim 1 , wherein, obtaining the first group of lane lines comprises:
extracting the first group of lane lines by using a lane line extracting model based on the first image, in which, the lane line extracting model is generated by performing a training on an image set and a lane line marked in the image set.
3. The method according to claim 1 , wherein, obtaining the first group of lane lines comprises:
transforming the first group of lane lines to a first group of three-dimensional lane lines based on height information of each pixel in the first image.
4. The method according to claim 1 , wherein, obtaining the second group of lane lines comprises:
transforming the second group of lane lines to a second group of three-dimensional lane lines based on a calibration for the camera or height information of each pixel in the first image.
5. The method according to claim 1 , further comprising:
determining at least one lane line set of at least one road related to the road;
transforming the lane line set and the at least one lane line set to a plurality of three-dimensional lane line sets;
determining a confidence of a lane line in the plurality of three-dimensional lane line sets; and
optimizing the plurality of three-dimensional lane line sets based on the confidence.
6. An apparatus for extracting a lane line, comprising:
one or more processors; and
a memory, configured to store one or more computer programs; wherein when the one or more computer programs are executed, the one or more processors are caused to:
obtain a first group of lane lines of a road based on a first image generated from a point cloud collected by a laser radar by performing: obtaining a plurality of lane lines based on a plurality of images generated from the point cloud collected by the laser radar, in which, the plurality of images comprises the first image; determining the first group of lane lines based on an optimization for the plurality of lane lines by: organizing a plurality of reflection value images according to a world coordinate system of a physical world, connecting disconnected lane lines by using a relationship between lane lines in neighboring images in the plurality of images;
obtain a second group of lane lines of the road based on a second image collected by a camera; and
determine a lane line set of the road based on the first group of lane lines and the second group of lane lines by causing the one or more processors to:
generate a first group of two-dimensional lane lines and a second group of two-dimensional lane lines by projecting a first group of three-dimensional lane lines and a second group of three-dimensional lane lines to a plane where the road is located;
perform a geometric transformation on the second group of two-dimensional lane lines, in which, the geometric transformation comprises at least one of a rotation and a translation;
determine one or more lane lines present in the second group of two-dimensional lane lines subjected to the geometric transformation and absent from the first group of two-dimensional lane lines; and
determine the lane line set by combining the first group of two-dimensional lane lines with the one or more lane lines.
7. The apparatus according to claim 6 , wherein, the one or more processors obtain the first group of lane lines by performing an act of:
extracting the first group of lane lines by using a lane line extracting model based on the first image, in which, the lane line extracting model is generated by performing a training on an image set and a lane line marked in the image set.
8. The apparatus according to claim 6 , wherein, the one or more processors obtain the first group of lane lines by performing an act of:
transforming the first group of lane lines to a first group of three-dimensional lane lines based on height information of each pixel in the first image.
9. The apparatus according to claim 6 , wherein, the one or more processors obtain the second group of lane lines by performing an act of:
transforming the second group of lane lines to a second group of three-dimensional lane lines based on a calibration for the camera or height information of each pixel in the first image.
10. The apparatus according to claim 6 , wherein the one or more processors are configured to:
determine at least one lane line set of at least one road related to the road;
transform the lane line set and the at least one lane line set to a plurality of three-dimensional lane line sets;
determine a confidence of a lane line in the plurality of three-dimensional lane line sets; and
optimize the plurality of three-dimensional lane line sets based on the confidence.
11. A non-transitory computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the processor is caused to perform acts of:
obtaining a first group of lane lines of a road based on a first image generated from a point cloud collected by a laser radar, comprising: obtaining a plurality of lane lines based on a plurality of images generated from the point cloud collected by the laser radar, in which, the plurality of images comprises the first image; determining the first group of lane lines based on an optimization for the plurality of lane lines by: organizing a plurality of reflection value images according to a world coordinate system of a physical world, connecting disconnected lane lines by using a relationship between lane lines in neighboring images in the plurality of images;
obtaining a second group of lane lines of the road based on a second image collected by a camera; and
determining a lane line set of the road based on the first group of lane lines and the second group of lane lines by:
generating a first group of two-dimensional lane lines and a second group of two-dimensional lane lines by projecting a first group of three-dimensional lane lines and a second group of three-dimensional lane lines to a plane where the road is located;
performing a geometric transformation on the second group of two-dimensional lane lines, in which, the geometric transformation comprises at least one of a rotation and a translation;
determining one or more lane lines present in the second group of two-dimensional lane lines subjected to the geometric transformation and absent from the first group of two-dimensional lane lines; and
determining the lane line set by combining the first group of two-dimensional lane lines with the one or more lane lines.