WO2018053834A1

WO2018053834A1 - Paired lane lines efficient detection method and device

Info

Publication number: WO2018053834A1
Application number: PCT/CN2016/100098
Authority: WO
Inventors: 黄凯明
Original assignee: 深圳市锐明技术股份有限公司
Priority date: 2016-09-26
Filing date: 2016-09-26
Publication date: 2018-03-29
Also published as: CN106415603B; CN106415603A

Abstract

An efficient detection method for paired lane lines comprises: obtaining two straight lines to be detected, and obtaining a distance between the sampling points and predetermined common points as a to-be-checked input vector according to sampling points respectively selected on the two straight lines by a preset interval; obtaining a output result of a support vector machine model corresponding to the to-be-checked input vector according to preset support vector machine models; and determining, according to the output result of the support vector machine model, whether the two lines are paired lane lines. The detection method can effectively ensure real-time judgment of paired lane lines and improve accuracy of judgment.

Description

Efficient detection method and device for paired lane lines

Technical field

The invention belongs to the field of automatic driving, and in particular relates to an efficient detection method and device for a pair of lane lines.

Background technique

The lane departure warning system is an auxiliary system for driving a car that causes a driver to reduce a car accident due to a lane departure by means of an alarm. When the vehicle deviates from the driving lane, an early warning reminder may be issued by the lane departure warning system, which may include an alarm sound, a steering wheel vibration, or an automatic change of steering.

In the lane departure warning system, in order to ensure the accuracy of the warning, the lane line needs to be correctly extracted and identified. The current paired lane line detection method generally needs to consume more system resources. When a higher accuracy is required, a certain calculation time is required, and real-time detection cannot be guaranteed; or, in order to improve the real-time detection, It may cause leak detection, resulting in an increase in false positive rate.

technical problem

It is an object of the present invention to provide an efficient detection method for paired lane lines to solve the problem that the prior art cannot effectively ensure accuracy and real-time performance in paired lane line detection.

Technical solution

In a first aspect, an embodiment of the present invention provides a method for efficiently detecting a pair of lane lines, where the method includes:

Obtaining two straight lines to be detected, respectively selecting a sample point respectively selected on the two straight lines according to a preset spacing, and obtaining a distance between the sample point and a predetermined common point as an input vector to be detected;

Acquiring, according to a preset support vector machine model, an output result of the support vector machine model corresponding to the input vector to be detected, wherein a vector dimension in the support vector machine model is the same as a dimension of the input vector to be detected;

Whether the two straight lines are paired lane lines is determined according to the output result of the support vector machine model.

With reference to the first aspect, in a first possible implementation manner of the first aspect, the sample selected on the two straight lines according to a preset spacing is obtained between the sample point and a predetermined common point The distance steps include:

Selecting sample points on the two straight lines according to a preset pitch;

Taking the center point of the image as a common point, the distance between the sample point and the common point is obtained as an input vector to be detected.

In conjunction with the first aspect, or the first possible implementation of the first aspect, in a second possible implementation of the first aspect, the sum of the distances between each of the straight lines and the predetermined common point is substituted Before the preset classifier step, the method further includes:

Collecting a large number of paired lane line samples and unpaired lane line samples, and selecting samples on the lane line samples according to the spacing;

Calculating a sample input vector x formed by the distance between the sample point on each line and the common point;

Calculating a vector w of the support vector machine model equation w ^T x+b=0 according to the sample input vector, and a constant b, where w ^T represents a transpose of the vector w, and the dimension of w is the same as the dimension of the input vector .

In conjunction with the second possible implementation of the first aspect, in a third possible implementation manner of the first aspect, the determining, according to the output result of the support vector machine model, whether the two lines are paired lane lines comprises:

Substituting the input vector to be detected into the support vector machine model w ^T x+b and calculating an output value of the support vector machine model. When the output value is greater than or equal to 1, the two straight lines are paired lane lines. When the output value is less than or equal to -1, the two straight lines are unpaired lane lines.

In conjunction with the first aspect, or the first possible implementation of the first aspect, or the third possible implementation of the first aspect, in a fourth possible implementation of the first aspect, the sample is included in each N samples selected on the lane line, the N being a natural number greater than or equal to 2.

In a second aspect, an embodiment of the present invention provides an efficient detection device for a pair of lane lines, the device comprising:

a lane line acquiring unit, configured to acquire two straight lines to be detected, and select a sample point respectively selected on the two straight lines according to a preset spacing, and obtain a distance between the sample point and a predetermined common point as a waiting Detecting an input vector;

a support vector machine operation unit, configured to acquire an output result of the support vector machine model corresponding to the input vector to be detected according to a preset support vector machine model, wherein the vector dimension in the support vector machine model is The dimensions of the input vector to be detected are the same;

And a determining unit, configured to determine, according to an output result of the support vector machine model, whether the two straight lines are paired lane lines.

With reference to the second aspect, in a first possible implementation manner of the second aspect, the lane line acquiring unit includes:

a sample selection subunit for respectively selecting a sample point on the two straight lines according to a preset spacing;

The common point acquisition subunit is configured to take a center point of the image as a common point, and obtain a distance between the sample point and the common point as an input vector to be detected.

With reference to the second aspect, or the first possible implementation manner of the second aspect, in a second possible implementation manner of the second aspect, the device further includes:

a sample collection unit, configured to collect a plurality of pairs of lane line samples and unpaired lane line samples, and select samples on the lane line samples according to the spacing;

a sample input vector calculation unit for calculating between a sample point on each straight line and the common point The sample input vector x of the distance;

a parameter training unit, configured to calculate, according to the sample input vector, a vector w of the support vector machine model equation w ^T x+b=0, and a constant b, where w ^T represents a transpose of the vector w, a dimension of w Same as the dimension of the input vector.

With reference to the second possible implementation of the second aspect, in a third possible implementation manner of the second aspect, the determining unit calculating unit is specifically configured to:

With reference to the second aspect, or the first possible implementation manner of the second aspect, or the third possible implementation manner of the second aspect, in the fourth possible implementation manner of the second aspect, the determining unit is specifically configured to:

The sample includes N samples selected on each lane line, the N being a natural number greater than or equal to two.

Beneficial effect

In the present invention, two lines to be detected are acquired, and samples are selected on the two straight lines according to a preset pitch to obtain a distance between the sample points and the common points, and the sample points and the common points are obtained. The input vector to be detected formed by the distance between the two is substituted into a preset support vector machine model, and the output result of the support vector machine model is obtained. According to the output result, it can be determined whether the two straight lines to be detected are formed. On the lane line. By adopting the method of the invention, it is only necessary to substitute the acquired input vector to be detected into a predetermined support vector machine model, thereby quickly determining whether it is a pair of lane lines, and effectively ensuring real-time determination of paired lane lines. Sex, can improve the accuracy of judgment.

DRAWINGS

1 is a flowchart of implementing an efficient detection method for a pair of lane lines according to an embodiment of the present invention;

2 is a schematic diagram of a training process of a support vector machine model according to an embodiment of the present invention;

3-4 are schematic diagrams of samples in which two straight lines are paired lane lines according to an embodiment of the present invention;

5-6 are schematic diagrams of samples of two straight lines that are unpaired lane lines according to an embodiment of the present invention;

7-8 are schematic diagrams of lane lines to be detected according to an embodiment of the present invention;

FIG. 9 is a schematic structural diagram of an apparatus for detecting high efficiency of a pair of lane lines according to an embodiment of the present invention.

Embodiment of the invention

The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

The method for detecting a pair of lane lines according to an embodiment of the present invention aims to overcome the prior art method for detecting a lane line. In order to improve the detection rate of a pair of lane lines, a relatively complex detection algorithm is often required, resulting in a detection algorithm. The detection calculation process takes a certain period of time. If the vehicle is running at a high speed, it will cause the detection result to lag, and the detection has low real-time defects. However, if a simple lane line judgment method is adopted, it is easy to cause an error in the detection result, which affects the user's judgment. The invention will now be further described with reference to the accompanying drawings.

FIG. 1 is a flowchart showing an implementation process of an efficient detection method for a pair of lane lines according to an embodiment of the present invention, which is described in detail as follows:

In step S101, two lines to be detected are acquired, and samples respectively selected on the two straight lines according to a preset pitch are used to obtain a distance between the sample point and a predetermined common point as an input to be detected. vector.

Specifically, the lane-forming line in the embodiment of the present invention refers to an auxiliary line for defining a lane in which the vehicle travels. Since other marking lines may be included in addition to the lane line during the running of the vehicle, as shown in FIG. 3, in addition to the lane line, an arrow mark is included, which is composed of an arrow line and a lane line. The logo should not be recognized as a pair of lane lines.

The two straight lines to be detected can be obtained by identifying the image. For example, the recognition of the line may be identified according to the color in the image, such as identifying a line in which the color is white, or a line with a yellow color.

The preset pitch can be set according to the size of the image. For example, according to the width of the image, the 1/3 screen width is set to be the length of the pitch. Of course, the size of the spacing may also be selected according to the number of samples required, and the length of the spacing may be set such that the selected sample includes the end position of the straight line.

The selection of the common point can be flexibly set according to the needs of the user. For example, the midpoint of the upper portion of the image may be set as the common point, or the midpoint of the lower portion of the image may be set as the common point, and the center point of the image may be set as the common point. Depending on how the common point is selected, the parameters of the support vector machine model will change accordingly. And the position of the common point selected in the training process of the weight vector is the same as the position of the common point corresponding to the two straight lines to be detected.

The number of values in the input vector to be detected is related to the manner in which the samples are selected. When the number of samples selected on two straight lines is larger, the number of input vectors to be detected is also larger.

In step S102, an output result of the support vector machine model corresponding to the input vector to be detected is acquired according to a preset support vector machine model, wherein the vector dimension in the support vector machine model and the input to be detected are The dimensions of the vector are the same.

The dimension of the hyperplane normal vector in the support vector machine model is the same as the dimension of the input vector to be detected. In order to flexibly match the input vectors to be detected of different dimensions, the support vector machine models of multiple dimensions may be preset. For example, a 4-dimensional support vector machine model, a 6-dimensional support vector machine model, and an 8-dimensional support vector machine model may be preset.

Wherein, when the support vector machine model is preset, the following steps as described in FIG. 2 may be included:

In step S201, a plurality of pairs of lane line samples and unpaired lane line samples are collected, and samples are selected on the lane line samples according to the spacing;

In step S202, a sample of the distance between the sample point on each straight line and the common point is calculated. The input vector x;

In step S203, a vector w of the support vector machine model equation w ^T x+b=0 is calculated according to the sample input vector, and a constant b, where w ^T represents a transpose of the vector w, and a dimension of w The dimensions of the input vector are the same.

For example, the dimension of the hyperplane normal vector is 4 dimensions, and the sample points collected in the paired lane line or the unpaired lane line sample are 4 (two are collected on each line), and the sample input vector is substituted. By the formula w ^T x+b=0, the values of the vector w and the constant b can be calculated.

3-6 are schematic diagrams of training samples provided by an embodiment of the present invention, wherein FIG. 3 to FIG. 4 are training samples of paired lane lines, and FIG. 5 and FIG. 6 are training samples of unpaired lane lines. And four sample points are selected as sample input vectors on each training sample. It can be understood that the more the number of training samples, the more accurate the calculation of the vector w and the constant b.

The training process of Figure 3-6 is introduced as follows:

The two pairs of lines in Figures 3 and 4 belong to the "lane line" and correspond to two sample input vectors:

The vector value of Fig. 3 is <5 (CM, upper left green line length), 14 (CM, lower left green line length), 11 (CM, upper right red line length), and 10 (CM, lower right red line length).

The vector value of Fig. 4 is <5 (CM, upper left green line length), 7 (CM, lower left green line length), 10 (CM, upper right red line length), and 16 (CM, lower right red line length).

The two pairs of lines in Figures 5 and 6 belong to the "unpaired lane line", corresponding to two vectors:

The vector value of Fig. 5 is <6 (CM, upper left yellow line length), 9 (CM, lower left yellow line length), 2 (CM, upper right purple line length), and 3 (CM, lower right purple line length).

The vector value of Fig. 6 is <3 (CM, upper left yellow line length), 5 (CM, lower left yellow line length), 8 (CM, upper right purple line length), and 11 (CM, lower right purple line length).

When x is the eigenvector of the pair of lane lines (Figures 3 and 4), solve W^T*x+b>=1; when x is the eigenvector of the unpaired lane line (Figure 5 and Figure 6) , solve W^T*x+b<=-1. Where b is a constant, W is a vector, and the dimension of W is determined by the dimension of X. In this case, the dimension of W is 4.

Determine the W and b processes, which is the solution of the support vector machine.

In this case, it can be solved: when W=<0.25, 0.21, -0.05, 0.24>, b=-5.04, W^T*x1+b=1, W^T*x2+b=1.02, W ^T*x3+b=-1.03, W^T*x4+b=-1, where x1, x2, x3, and x4 are the vectors of Figures 3, 4, 5, and 6, respectively.

The data points falling on two hyperplanes of W^T*x+b=1 or W^T*x+b=-1 become support vector points, and x1 and x4 belong to the support vector machine W^T*x+b= 0 support vector point.

In step S103, it is determined whether the two straight lines are paired lane lines according to the output result of the support vector machine model.

Specifically, the step of determining whether the two lines are paired lane lines according to the output result of the support vector machine model includes:

After determining the values of the parameters and vectors in the support vector machine model according to the preset training samples, the detected input vectors can be detected. Figure 7 and Figure 8 show the input vector to be detected. The specific detection process is as follows:

The input vector to be detected obtained in Fig. 7 is: x=<7, 11, 8, 12>. Substituting the input vector to be detected into the support vector machine model, and calculating: W^T*x+b=1.5>1, the determination that "two straight lines are paired lane lines" is obtained.

In the middle two lines of Figure 8, the corresponding input vector to be detected is obtained as x=<4,7,5,7>. Calculation: W^T*x+b=-1.14<-1, the judgment of "two straight unpaired lane lines" is obtained.

The invention selects the sample points on the two straight lines according to the preset spacing by obtaining two straight lines to be detected, and obtains the distance between the sample points and the common points, and between the sample points and the common points The input vector to be detected is substituted into a preset support vector machine model, and the output result of the support vector machine model is obtained. According to the output result, it can be determined whether the two straight lines to be detected are paired lanes. line. By adopting the method of the invention, it is only necessary to substitute the acquired input vector to be detected into a predetermined support vector machine model, thereby quickly determining whether it is a pair of lane lines, which can effectively ensure the pair. The real-time judgment of the paired lane lines can improve the accuracy of the judgment.

FIG. 9 is a schematic structural diagram of a high-efficiency detecting device for a pair of lane lines according to an embodiment of the present invention, which is described in detail as follows:

The high-efficiency detecting device for the pair of lane lines in the embodiment of the invention includes:

The lane line acquiring unit 901 is configured to acquire two straight lines to be detected, and select a sample point respectively selected on the two straight lines according to a preset spacing, and obtain a distance between the sample point and a predetermined common point as The input vector to be detected;

The support vector machine operation unit 902 is configured to obtain an output result of the support vector machine model corresponding to the input vector to be detected according to a preset support vector machine model, where the vector dimension and the vector in the support vector machine model The dimensions of the detected input vector are the same;

The determining unit 903 is configured to determine whether the two straight lines are paired lane lines according to the output result of the support vector machine model.

Preferably, the lane line acquiring unit includes:

Preferably, the device further comprises:

a sample input vector calculation unit, configured to calculate a sample input vector x formed by a distance between a sample on each straight line and the common point;

Preferably, the determining unit calculating unit is specifically configured to:

Preferably, the sample comprises N samples selected on each lane line, the N being a natural number greater than or equal to 2.

The high-efficiency detecting device for the pair of lane lines in the embodiment of the present invention corresponds to the above-mentioned method for efficiently detecting the pair of lane lines, and details are not described herein again.

In the several embodiments provided by the present invention, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the device embodiments described above are merely illustrative. For example, the division of the unit is only a logical function division. In actual implementation, there may be another division manner, for example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.

The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a standalone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention, which is essential or contributes to the prior art, or all or part of the technical solution, may be embodied in the form of a software product stored in a storage medium. A number of instructions are included to cause a computer device (which may be a personal computer, server, or network device, etc.) to perform all or part of the methods described in various embodiments of the present invention. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like. .

The above is only the preferred embodiment of the present invention, and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. Within the scope.

Claims

An efficient detection method for a pair of lane lines, the method comprising:

Obtaining two straight lines to be detected, respectively selecting a sample point respectively selected on the two straight lines according to a preset spacing, and obtaining a distance between the sample point and a predetermined common point as an input vector to be detected;

Acquiring, according to a preset support vector machine model, an output result of the support vector machine model corresponding to the input vector to be detected, wherein a vector dimension in the support vector machine model is the same as a dimension of the input vector to be detected;

Whether the two straight lines are paired lane lines is determined according to the output result of the support vector machine model.
The method according to claim 1, wherein the sample selected on the two straight lines according to a preset pitch acquires a distance between the sample point and a predetermined common point as an input to be detected Vector steps include:

Selecting sample points on the two straight lines according to a preset pitch;

Taking the center point of the image as a common point, the distance between the sample point and the common point is obtained as an input vector to be detected.
The method according to claim 1 or 2, wherein before the step of substituting the sum of the distances between each of the straight lines and the predetermined common point into a predetermined classifier step, the method further comprises:

Collecting a large number of paired lane line samples and unpaired lane line samples, and selecting samples on the lane line samples according to the spacing;

Calculating a sample input vector x formed by the distance between the sample point on each line and the common point;

Calculating a vector w of the support vector machine model equation w T x+b=0 according to the sample input vector, and a constant b, where w T represents a transpose of the vector w, and the dimension of w is the same as the dimension of the input vector .
The method according to claim 3, wherein the step of determining whether the two straight lines are paired lane lines according to an output result of the support vector machine model comprises:

Substituting the input vector to be detected into the support vector machine model w T x+b and calculating an output value of the support vector machine model. When the output value is greater than or equal to 1, the two straight lines are paired lane lines. When the output value is less than or equal to -1, the two straight lines are unpaired lane lines.
The method of claim 1, 2 or 4 wherein said sample comprises N samples selected on each lane line, said N being a natural number greater than or equal to two.
An efficient detection device for a pair of lane lines, characterized in that the device comprises:

a lane line acquiring unit, configured to acquire two straight lines to be detected, and select a sample point respectively selected on the two straight lines according to a preset spacing, and obtain a distance between the sample point and a predetermined common point as a waiting Detecting an input vector;

a support vector machine operation unit, configured to acquire an output result of the support vector machine model corresponding to the input vector to be detected according to a preset support vector machine model, wherein the vector dimension in the support vector machine model is The dimensions of the input vector to be detected are the same;

And a determining unit, configured to determine, according to an output result of the support vector machine model, whether the two straight lines are paired lane lines.
The device according to claim 6, wherein the lane line acquisition unit comprises:

a sample selection subunit for respectively selecting a sample point on the two straight lines according to a preset spacing;

The common point acquisition subunit is configured to take a center point of the image as a common point, and obtain a distance between the sample point and the common point as an input vector to be detected.
The device according to claim 6 or 7, wherein the device further comprises:

a sample collection unit, configured to collect a plurality of pairs of lane line samples and unpaired lane line samples, and select samples on the lane line samples according to the spacing;

a sample input vector calculation unit, configured to calculate a sample input vector x formed by a distance between a sample on each straight line and the common point;

a parameter training unit, configured to calculate, according to the sample input vector, a vector w of the support vector machine model equation w T x+b=0, and a constant b, where w T represents a transpose of the vector w, a dimension of w Same as the dimension of the input vector.
The device according to claim 8, wherein the determining unit calculating unit is specifically configured to:

Substituting the input vector to be detected into the support vector machine model w T x+b and calculating an output value of the support vector machine model. When the output value is greater than or equal to 1, the two straight lines are paired lane lines. When the output value is less than or equal to -1, the two straight lines are unpaired lane lines.
The apparatus according to claim 6, 7 or 9, wherein said sample comprises N sample points selected on each lane line, said N being a natural number greater than or equal to 2.