Vehicle LiDAR system and object detection method thereof

1 . An object detection method of a vehicle lidar system, comprising:

receiving point data from a lidar sensor;

determining an object to be separated from the point data;

forming a contour of the object by connecting peak points among point data of the object;

determining a contour line in the contour of the object, a connecting section of which includes point data less than or equal to a reference; and

setting a separation reference line crossing the contour line for separating the contour of the object into two regions to recognize point data of the two regions as respective objects.

2 . The object detection method of claim 1 , further comprising:

generating the contour based on all point data included in the object to be separated;

selecting objects in which at least a part of the contour is included in a region of interest (ROI) of a host vehicle;

selecting objects having sizes equal to or greater than a reference size among the objects included in the region of interest (ROI) of the host vehicle; and

selecting an object having one or more points between a first point and an end point of the contour among the objects having sizes equal to or greater than the reference size as the object to be separated.

3 . The object detection method of claim 1 , wherein the determining of the contour line including the point data equal to or less than the reference comprises:

determining a break point included in the contour of the object to be separated;

determining, based on the break point, a first peak point and a second peak point connecting the contour line including the point data equal to or less than the reference; and

extracting a first point, an end point, and the first and second peak points of the contour of the object to be separated as four peak points.

4 . The object detection method of claim 3 , wherein the determining of a break point included in the contour of the object to be separated comprises:

determining whether a straight line connecting the first point and the end point of the contour of the object to be separated is a diagonal line;

checking whether points of the contour of the object to be separated are present in one direction with respect to the diagonal line;

selecting a point farthest from the first point and the end point as a first peak point candidate;

calculating a variance of distances between the straight line connecting the first point and the first peak point and points on the contour of the object to be separated present between the first point and the first peak point and calculating a variance of distances between the straight line connecting the first peak point and the end point and points on the contour of the object to be separated present between the first peak point and the end point; and

determining that the contour of the object to be separated has two break points if variance values of the straight line connecting the first point and the first peak point and the straight line connecting the first peak point and the end point are within a range of variance threshold values of reference lines and a range of variance threshold values of non-reference lines, respectively.

5 . The object detection method of claim 4 , wherein the determining of the first peak point and the second peak point connecting the contour line including the point data equal to or less than the reference based on the break point comprises determining whether the contour line including point data equal to or less than the reference is present for the contour of the object to be separated in a linear direction having a larger variance value based on the first peak point upon determining that the contour of the object to be separated has the two break points.

6 . The object detection method of claim 3 , wherein, if there is one break point included in the contour of the object to be separated, determining whether a shape of the contour of the object to be separated defined by the break point satisfies a concave shape with respect to a host vehicle comprises:

calculating a variance of distances between a straight line connecting the first point of the contour of the object to be separated and the break point and points on the contour of the object to be separated present between the first point and the break point, and calculating a variance of distances between a straight line connecting the break point and the end point and points on the contour of the object to be separated present between the break point and the end point, with respect to the object to be separated which satisfies the concave shape; and

determining whether the contour line including point data equal to or less than the reference for the contour of the object to be separated is in a linear direction having a smaller calculated variance value.

7 . The object detection method of claim 3 , wherein the determining of a first peak point and a second peak point connecting a contour line including point data less than or equal to the reference based on the break point comprises:

dividing the contour line formed by the two points into a plurality of regions in a direction parallel to the contour line based on the break point; and

determining the two points forming the corresponding contour line as the first peak point and the second peak point if a number of regions having no point data among the plurality of regions is equal to or greater than a reference value.

8 . The object detection method of claim 3 , wherein the setting of a separation reference line for separating the contour of the object to be separated into two regions while passing through the contour line including the point data equal to or less than the reference to recognize point data of the two regions as respective objects comprises:

setting points included in a contour connected from the first point of the contour of the object to be separated to the first peak point as a first point group;

setting points included in a contour connected from the second peak point to the end point of the contour of the object to be separated as a second point group; and

selecting a straight line passing through a midpoint of the first peak point and the second peak point and not passing through the points of the first point group and the second point group as the separation reference line.

9 . The object detection method of claim 8 , wherein the selecting of the separation reference line comprises:

setting at least two straight lines passing through midpoint of the first peak point and the second peak point and not passing through the points of the first point group and the second point group as separation reference line candidates; and

selecting, as the separation reference line, a straight line having an average value of slopes of a straight line having a largest slope and a straight line having a smallest slope among the separation reference line candidates.

10 . The object detection method of claim 8 , wherein recognizing of the point data of the two regions as respective objects comprises:

deleting point data within a reference distance from the separation reference line;

grouping the point data of a first region of the two regions and the point data of a second region of the two regions divided by the separation reference; and

recognizing a first object based on the point data of the first region and recognizing a second object based on the point data of the second region.

11 . The object detection method of claim 1 , wherein the determining of a contour line in which a connection section of a contour line connecting the peak points includes point data equal to or less than a reference comprises determining a contour line having a lowest density of point data among contour lines connecting the peak points.

12 . A non-transitory computer-readable recording medium having recorded thereon a program for executing an object detection method of a vehicle lidar system, the program implementing:

receiving point data from a lidar sensor;

determining an object to be separated from the point data,

forming a contour of the object by connecting peak points among point data of the object;

determining a contour line, in the contour of the object, a connecting section of which includes point data less than or equal to a reference; and

setting a separation reference line crossing the contour line for separating the contour of the object into two regions while to recognize point data of the two regions as respective objects.

13 . A vehicle lidar system comprising:

a lidar sensor; and

an object tracking device configured to:

recognize objects based on point data obtained through the lidar sensor;

upon determining an object to be separated among the objects, determine, a contour line, in a contour of the object formed by connecting peak points among point data of the object, a connecting section of which includes point data less than or equal to a reference; and

set a separation reference line crossing the contour line for separating the contour of the object into two regions to recognize point data of the two regions as respective objects.

14 . The vehicle lidar system of claim 13 , wherein the object tracking device is configured to generate a contour based on all point data included in the recognized objects, select objects in which at least a part of the contour is included in a region of interest (ROI) of a host vehicle, select objects having sizes equal to or greater than a reference size among the objects included in the region of interest (ROI) of the host vehicle, and select an object having one or more points between a first point and an end point of the contour among the objects having sizes equal to or greater than the reference size as the object to be separated.

15 . The vehicle lidar system of claim 13 , wherein the object tracking device is configured to determine a break point included in the contour of the object to be separated, determine, based on the break point, a first peak point and a second peak point connecting the contour line including the point data equal to or less than the reference, and extract a first point, an end point, and the first and second peak points of the contour of the object to be separated as four peak points.

16 . The vehicle lidar system of claim 15 , wherein the object tracking device is configured to divide the contour line formed by the two points into a plurality of regions in a direction parallel to the contour line based on the break point and determines the two points forming the corresponding contour line as the first peak point and the second peak point if a number of regions having no point data among the plurality of regions is equal to or greater than a reference value.

17 . The vehicle lidar system of claim 15 , wherein the object tracking device is configured to set points included in a contour connected from the first point of the contour of the object to be separated to the first peak point as a first point group, set points included in a contour connected from the second peak point to the end point of the contour of the object to be separated as a second point group, and select a straight line passing through a midpoint of the first peak point and the second peak point and not passing through the points of the first point group and the second point group as the separation reference line.

18 . The vehicle lidar system of claim 17 , wherein the object tracking device is configured to delete point data within a reference distance from the separation reference line, group the point data of a first region of the two regions and the point data of a second region of the two regions divided by the separation reference, recognize a first object based on the point data of the first region, and recognize a second object based on the point data of the second region.