CN112149572A - Road edge detection method, device and storage medium - Google Patents

Abstract

The application relates to the technical field of point cloud processing, in particular to a road edge detection method, a road edge detection device and a storage medium. The method comprises the following steps: acquiring point cloud data containing roads, and extracting an interesting region in the point cloud data; determining a ground plane in the region of interest; rasterizing the region of interest; acquiring statistical information in each grid according to the ground plane, and identifying road edge points in the grids according to the statistical information; and determining the road edge in the road according to the identified road edge points. The problem of poor road edge accuracy determined in the prior art is solved; the effect that the road edge can be accurately identified and obtained when the road surface has the slope and the accuracy of the road edge identification is improved is achieved.

Description

Road edge detection method, device and storage medium

technical field

The present application relates to the technical field of point cloud processing, and in particular, to a road edge detection method, device and storage medium.

Background technique

Road edge recognition is an important research content in the environment perception link of autonomous driving. After identifying the road edge, the perceptual environment can be divided into road areas and non-road areas, providing more accurate road sections for subsequent obstacle detection and identification of drivable areas, thereby reducing the amount of calculation and improving the real-time performance of environmental perception.

At present, the common road edge detection methods are: collect images based on visual sensors, perform binarization processing on the collected images to obtain a binarized image, and perform edge detection on the binarized image to obtain the road edges in the image; based on lidar The point cloud data is collected by using the interval collinear point extraction algorithm to extract the roadside data points, and the roadside data points are clustered, and then the least squares method is used to linearly fit the actual roadside.

However, in the above solution, under the influence of ambient light, the quality of the collected image is unstable, and accordingly, the accuracy of the processed road edge is also unstable. When there are wrong road edge points, the linear fitting error using the least squares method is larger.

SUMMARY OF THE INVENTION

In view of this, embodiments of the present application provide a road edge detection method, device, and storage medium to solve the problems in the prior art.

According to a first aspect, an embodiment of the present application provides a road edge detection method, the method comprising:

Obtain point cloud data containing roads, and extract the region of interest in the point cloud data;

determining a ground plane in the region of interest;

rasterizing the region of interest;

Acquire statistical information in each grid according to the ground plane, and identify roadside points in the grid according to the statistical information;

The road edge in the road is determined according to each road edge point identified.

Optionally, the extracting the region of interest in the point cloud data includes:

Through filtering is performed on the data of the two channels in the point cloud data, respectively, to obtain filtered point cloud data.

Optionally, the determining the ground plane in the region of interest includes:

Randomly select 3 points from each point of the region of interest to obtain a candidate ground plane;

Calculate the distances between each of the other points except the selected 3 points and the candidate ground plane in the various roadside points;

Counting the cumulative number of points whose distance from the candidate ground plane is less than the first preset distance;

+1 the number of statistics;

When the updated statistical number of times does not reach the first preset number of times, the step of randomly selecting 3 points from each point in the region of interest to obtain a candidate ground plane is performed again;

When the updated statistical count reaches the first preset count, the corresponding candidate ground plane with the largest accumulated number is determined as the ground plane.

Optionally, the obtaining statistical information in each grid according to the ground plane, and identifying the roadside points in the grid according to the statistical information, includes:

For each grid, count the number of point clouds in the grid, the maximum height between the point cloud and the ground plane, and the minimum height between the point cloud and the ground plane;

When the number of the point clouds reaches the number threshold, the maximum height is within the height threshold range, and the height difference between the maximum height and the minimum height is within the height difference threshold range, the points in the grid are Clouds are identified as roadside points.

Optionally, the determining the curb in the road according to each identified curb point includes:

Determine the left curb and the right curb according to the coordinates of each identified curb point;

For each side curb, straight line fitting is performed on each curb point through a random sampling consistency algorithm to obtain the curb in the road.

Optionally, the random sampling consistency algorithm is used to perform straight line fitting on each road edge point to obtain the road edge in the road, including:

Randomly select 2 points from the various roadside points to obtain candidate straight lines;

Calculate the distances between each of the other points except the selected 2 points and the candidate straight line in the various roadside points;

Counting the cumulative number of points whose distance from the candidate straight line is less than the second preset distance;

+1 the number of statistics;

When the updated statistical number of times does not reach the second preset number of times, the step of randomly selecting 2 points from the various roadside points to obtain a candidate straight line is performed again;

When the updated statistical count reaches the second preset count, the corresponding candidate straight line with the largest accumulated number is determined as the road along the route.

Optionally, the method further includes:

The Kalman filter is used to track and predict the road edge fitted by the previous frame, and the predicted value is obtained. Combined with the detection result of the current frame, the road edge of the current frame is obtained.

Optionally, the obtaining point cloud data including roads includes:

Obtain the point cloud data obtained by the lidar.

In a second aspect, a road edge detection device is provided, the device includes a memory and a processor, the memory stores at least one program instruction, and the processor loads and executes the at least one program instruction to implement the first The method described in one aspect.

In a third aspect, a computer storage medium is provided, wherein the storage medium stores at least one program instruction, and the at least one program instruction is used to be loaded and executed by a processor to implement the method of the first aspect.

By acquiring point cloud data containing roads, extract the region of interest in the point cloud data; determine the ground plane in the region of interest; perform rasterization on the region of interest; Statistical information in each grid is obtained, and roadside points in the grid are identified according to the statistical information; the roadside in the road is determined according to the identified roadside points. The problem of poor accuracy of the road edge determined in the prior art is solved; the effect of accurately identifying the road edge and improving the road edge recognition accuracy even when the road surface has a slope is achieved.

In addition, by rasterizing the area of interest, and then extracting the road edge according to the point cloud in the grid, the problem of large amount of data in the prior art is solved point by point screening, and the effect of fast screening of road edge points is achieved. .

The above description is only an overview of the technical solution of the present invention. In order to understand the technical means of the present invention more clearly, and implement it according to the content of the description, the preferred embodiments of the present invention are described in detail below with the accompanying drawings.

Description of drawings

1 is a method flowchart of a road edge detection method provided by an embodiment of the present invention;

Fig. 2 is a method flowchart of a method for determining a ground plane provided by an embodiment of the present invention;

FIG. 3 is a flowchart of a method for determining a road edge provided by an embodiment of the present invention.

Detailed ways

The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. The indicated orientation or positional relationship is based on the orientation or positional relationship shown in the accompanying drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the indicated device or element must have a specific orientation or a specific orientation. construction and operation, and therefore should not be construed as limiting the invention. Furthermore, the terms "first", "second", and "third" are used for descriptive purposes only and should not be construed to indicate or imply relative importance.

In the description of the present invention, it should be noted that the terms "installed", "connected" and "connected" should be understood in a broad sense, unless otherwise expressly specified and limited, for example, it may be a fixed connection or a detachable connection Connection, or integral connection; can be mechanical connection, can also be electrical connection; can be directly connected, can also be indirectly connected through an intermediate medium, can be internal communication between two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood in specific situations.

In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

Please refer to FIG. 1, which shows a method flowchart of a road edge detection method provided by an embodiment of the present application. As shown in FIG. 1, the method includes:

Step 101, acquiring point cloud data containing roads, and extracting a region of interest in the point cloud data;

In actual implementation, the point cloud data obtained by the lidar can be obtained.

After the point cloud data is acquired, the data of the two channels in the point cloud data are respectively subjected to through-pass filtering to obtain filtered point cloud data. Through the above filtering, the interference information in the point cloud data can be filtered, and the point cloud data in the region of interest can be obtained.

Step 102, determining the ground plane in the region of interest;

In actual implementation, through filtering can be performed on the Z channel of the point cloud data, point clouds near the ground can be selected, and then a random sampling consistency algorithm is used to perform plane fitting on the point clouds in the region of interest to obtain the point cloud of interest. ground plane in the area.

Optionally, please refer to Figure 2, this step includes:

(1) Randomly select 3 points in each point of the region of interest to obtain a candidate ground plane;

Three points are randomly selected from each point in the region of interest to form a plane, which is used as a candidate ground plane. In actual implementation, if the three randomly selected points are on the same straight line, perform this step again.

(2), calculate the distance between each point other than the selected 3 points and the candidate ground plane in the various roadside points;

(3), count the cumulative number of points whose distance from the candidate ground plane is less than the first preset distance;

The first preset distance is a system default distance, or a user-defined distance according to actual usage requirements.

When traversing each point, when the distance between the point and the candidate ground plane is less than the first preset distance, the point is regarded as the inner point, otherwise, if the distance is greater than the first preset distance, the point is regarded as the outer point. After the traversal is complete, calculate the cumulative number of interior points.

(4), increase the number of statistics by 1;

(5), when the number of statistics after the update does not reach the first preset number of times, perform the step of randomly selecting 3 points from each point of the region of interest to obtain a candidate ground plane again;

The first preset number of times is a default number of times or a user-defined number of times, and the number of times is the number of times of iteratively calculating the candidate ground plane. Moreover, in actual implementation, the first preset number of times can be determined according to the size of the region of interest. For example, the larger the region of interest is, the larger the first preset number of times is, and vice versa, the smaller the region of interest is, the first The preset times are also smaller.

The first preset number of times is an integer greater than or equal to 2, and the greater the first preset number of times, the greater the number of candidate ground planes obtained. For the accuracy deviation, in actual implementation, the value of the first preset number of times may be set according to actual requirements, which is not limited in this embodiment.

(6) When the updated statistical count reaches the first preset count, determine the candidate ground plane with the largest corresponding accumulated number as the ground plane.

When the number of updated statistics reaches the first preset number of times, it means that the first preset number of candidate ground planes have been obtained at this time, and at this time, the candidate ground plane with the largest corresponding cumulative number is selected from the multiple candidate ground planes as the final ground plane.

Step 103, rasterizing the region of interest;

After the region of interest is obtained, the region of interest can be rasterized. The size of each grid can be the same or different, which is not limited, and the region of interest is divided into M* in this embodiment. A grid of the same size of N is used as an example.

Step 104, obtaining statistical information in each grid according to the ground plane, and identifying roadside points in the grid according to the statistical information;

This step can include:

First, for each grid, count the number of point clouds in the grid, the maximum height between the point cloud and the ground plane, and the minimum height between the point cloud and the ground plane;

After the ground plane is determined in step 102, the height between each point cloud in the grid and the ground plane can be calculated to obtain the minimum height and the maximum height.

second,

When the number of the point clouds reaches the number threshold, the maximum height is within the height threshold range, and the height difference between the maximum height and the minimum height is within the height difference threshold range, the points in the grid are Clouds are identified as roadside points.

In actual implementation, by selecting a point whose height difference is less than the height difference threshold as the roadside point, the height difference of the selected roadside point is within the error range of the actual roadside height, and the accuracy of the determined roadside point is improved.

In addition, in actual implementation, the difference between the maximum height and the preset height can also be detected, and if the difference is greater than the preset threshold, it is ignored.

Step 105: Determine the road edge in the road according to the identified road edge points.

Since there will be curbs on both sides of the road, this step can include:

First, determine the left and right curbs according to the identified coordinates of each curb point;

Second, for each side curb, straight-line fitting is performed on each curb point through a random sampling consistency algorithm to obtain the curb in the road.

Optionally, please refer to Figure 3, this step may include:

(1), randomly select 2 points in each of the road edge points to obtain candidate straight lines;

(2), calculate the distance between each other point and the candidate straight line except the selected 2 points in each of the roadside points;

(3), count the cumulative number of points whose distance from the candidate straight line is less than the second preset distance;

(4), increase the number of statistics by 1;

(5), when the number of statistics after the update has not reached the second preset number of times, perform the step of randomly selecting 2 points in each of the roadside points to obtain a candidate straight line again;

(6) When the updated statistical count reaches a second preset count, determine the corresponding candidate straight line with the largest cumulative number as the road along the route.

This step is similar to the above-mentioned step 102, and is not repeated here.

In addition, after the road edge is obtained, Kalman filtering can be used to track and predict the road edge obtained by fitting the previous frame to obtain a predicted value, and combined with the detection result of the current frame, the road edge of the current frame can be obtained. The problem of frame loss or excessive error in the collected point cloud data is alleviated, and the effect of smoother road edge detection is achieved.

To sum up, by acquiring point cloud data containing roads, the region of interest in the point cloud data is extracted; the ground plane in the region of interest is determined; and the region of interest is rasterized; Statistical information in each grid is obtained according to the ground plane, and roadside points in the grid are identified according to the statistical information; the roadside in the road is determined according to each roadside point obtained by identification. The problem of poor accuracy of the road edge determined in the prior art is solved; the effect of accurately identifying the road edge and improving the road edge recognition accuracy is achieved even when the road surface has a slope.

In addition, by rasterizing the area of interest, and then extracting the road edge according to the point cloud in the grid, the problem of large amount of data in the prior art is solved point by point screening, and the effect of fast screening of road edge points is achieved. .

This embodiment also provides a road edge detection device, the device includes a memory and a processor, the memory stores at least one program instruction, and the processor loads and executes the at least one program instruction to implement the above-mentioned methods described in the examples.

This embodiment also provides a computer storage medium, where at least one program instruction is stored in the storage medium, and the at least one program instruction is used to be loaded and executed by a processor to implement the method described in the foregoing embodiment.

The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity, all possible combinations of the technical features in the above-described embodiments are not described. However, as long as there is no contradiction between the combinations of these technical features, All should be regarded as the scope described in this specification.

The above-mentioned embodiments only represent several embodiments of the present invention, and the descriptions thereof are specific and detailed, but should not be construed as a limitation on the scope of the invention patent. It should be pointed out that for those skilled in the art, without departing from the concept of the present invention, several modifications and improvements can be made, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the appended claims.

Claims (10)

1. A method of road edge detection, the method comprising:

acquiring point cloud data containing roads, and extracting an interesting region in the point cloud data;

determining a ground plane in the region of interest;

rasterizing the region of interest;

acquiring statistical information in each grid according to the ground plane, and identifying road edge points in the grids according to the statistical information;

and determining the road edge in the road according to the identified road edge points.

2. The method of claim 1, wherein the extracting regions of interest in the point cloud data comprises:

and performing direct filtering on the data of the two channels in the point cloud data respectively to obtain filtered point cloud data.

3. The method of claim 1, wherein the determining the ground plane in the region of interest comprises:

randomly selecting 3 points from all the points of the region of interest to obtain a candidate ground plane;

calculating the distance between each point of the road edge points except the selected 3 points and the candidate ground plane;

counting the accumulated number of points with the distance to the candidate ground plane smaller than a first preset distance;

counting the times by + 1;

when the updated statistical frequency does not reach a first preset frequency, the step of randomly selecting 3 points from all the points in the region of interest is executed again to obtain a candidate ground plane;

and when the updated statistical frequency reaches the first preset frequency, determining the candidate ground plane with the maximum corresponding accumulative number as the ground plane.

4. The method of claim 1, wherein obtaining statistical information in each grid from the ground plane, and identifying road edge points in the grid from the statistical information comprises:

for each grid, counting the number of point clouds in the grid, the maximum height between the point clouds and the ground plane and the minimum height between the point clouds and the ground plane;

and when the number of the point clouds reaches a number threshold, the maximum height is within a height threshold range, and the height difference between the maximum height and the minimum height is within a height difference threshold range, determining the point clouds in the grid as road edge points.

5. The method of claim 1, wherein determining the road edge in the road from the identified road edge points comprises:

determining a left road edge and a right road edge according to the coordinates of the road edge points obtained by identification;

and for each road edge, performing straight line fitting on each road edge point through a random sampling consistency algorithm to obtain the road edge in the road.

6. The method of claim 5, wherein the step of fitting a straight line to each road edge point through a random sampling consistency algorithm to obtain the road edge in the road comprises:

randomly selecting 2 points from the road edge points to obtain a candidate straight line;

calculating the distance between each point except the selected 2 points in each road edge point and the candidate straight line;

counting the accumulated number of points with the distance to the candidate straight line smaller than a second preset distance;

counting the times by + 1;

when the updated statistical frequency does not reach a second preset frequency, the step of randomly selecting 2 points in each road edge point is executed again to obtain a candidate straight line;

and when the updated statistical frequency reaches a second preset frequency, determining the candidate straight line with the maximum corresponding accumulated number as the route line.

7. The method of any of claims 1 to 6, further comprising:

and tracking and predicting the road edge obtained by fitting the previous frame through Kalman filtering to obtain a predicted value, and combining the detection result of the current frame to obtain the road edge of the current frame.

8. The method of any one of claims 1 to 6, wherein the obtaining point cloud data including a road comprises:

and acquiring the point cloud data acquired by the laser radar.

9. A road edge detection device, comprising a memory having at least one program instruction stored therein and a processor that implements the method of any of claims 1 to 8 by loading and executing the at least one program instruction.

10. A computer storage medium having stored therein at least one program instruction for loading and execution by a processor to perform the method of any of claims 1 to 8.

Images