WO2014129474A1

WO2014129474A1 - Device for recognizing demarcation line and method for same

Info

Publication number: WO2014129474A1
Application number: PCT/JP2014/053811
Authority: WO
Inventors: 俊輔鈴木; 俊也熊野
Original assignee: 株式会社デンソー; 株式会社日本自動車部品総合研究所
Priority date: 2013-02-25
Filing date: 2014-02-18
Publication date: 2014-08-28
Also published as: JP2014164492A

Abstract

A device for recognizing a demarcation line is configured so as to recognize a demar cation line on the basis of a grayscale image obtained by converting a road-surface image photographed by a camera, so that when a vehicle enters a two-lane expressway, etc. for which a yellow demarcation line (yellow line) is used (Yes in S105), the accuracy of recognizing the yellow line is increased (S115). Specifically, by focusing on the point at which the luminance of R incr eases in a yellow-line region of the road-surface image, the luminance of R is weighted while the road-surface image is converted into a grayscale image, and an edge threshold referenced when extracting an edge point is reduced, facilitating extraction of a yellow-line edge point. The accuracy of recognizing the yellow line is thereby increased.

Description

Apparatus and method for recognizing lane markings

The present invention relates to an apparatus and method for recognizing lane markings based on road surface images.

Conventionally, a device that recognizes a lane marking from a road surface image obtained by photographing a road surface in front of a vehicle and performs driving support based on the recognition result is known. As an example, the travel path recognition apparatus described in Patent Document 1 stores a composite line pattern composed of a plurality of lane markings in advance, and compares the pattern with the pattern to determine the type of the composite line. Recognize correctly.

Japanese Patent No. 4207935

By the way, white roads (white lane markings) are usually used on roads in many countries, but for example, on a provisional two-lane road (hereinafter simply referred to as a provisional two-lane road) where two-way traffic occurs, A composite line composed of yellow lines (yellow to orange partition lines) is used as the center line. Yellow lines are also used in carpool lanes in the United States and construction sections in Europe.

A provisional two-lane road is a road that was planned as a road with four or more lanes, but only two lanes were provisionally opened. A car pool lane is a vehicle with a certain number of people on board. In this case, the lane is allowed to run.

And since the yellow line has a lower contrast with the road surface than the white line, it is difficult to extract the edge from the road image, so the accuracy of detecting the yellow line and the type of the yellow line (whether it is a solid line or a broken line) ) Is reduced, and the yellow line cannot be accurately recognized.

This invention is made | formed in view of the said subject, and it aims at providing the lane marking recognition apparatus which can recognize the lane marking of the specific color different from the color normally used accurately.

An apparatus for recognizing a lane marking according to an aspect of the present invention, which has been made in view of the above problems, captures a road surface of a road in front of a vehicle and generates a road surface image that is a color image. Detecting a color change of the road surface image along the scanning line, and extracting the edge of the lane marking drawn on the road surface by comparing the degree of the change and a predetermined edge threshold; Is provided.

The lane marking recognition device is generally used as a lane marking color, with a recognition section for recognizing a lane marking based on the edge extracted by the extraction section, and a road surface area in which no lane marking is drawn on the road surface. Road determination means for determining whether or not the vehicle is traveling on a specific road using a specific color lane marking whose degree of color change relative to the road surface area is smaller than that of the normal color.

Then, when the road determination unit determines that the vehicle is traveling on the specific road, the extraction unit improves the extraction sensitivity, which is a sensitivity at the time of extracting the edge of the specific color marking line.
According to such a configuration, when the vehicle is traveling on a specific road, the edge of the specific color lane marking is easily extracted from the road surface image. The accuracy of determining the type of can be improved. For this reason, the partition line of a specific color can be recognized with high accuracy.

Also, as already mentioned, the white line is usually used for roads in many countries, but the yellow line is used for provisional two-lane roads in Japan, car pool lanes in the US, construction sections in Europe, and the like.

Therefore, for example, the normal color is white and the specific color is yellow to orange.
According to such a configuration, the yellow line can be accurately recognized when the vehicle is traveling on a provisional two-lane road or the like.

In particular, as described above, a composite line composed of a yellow line and a white line is used as the center line of the provisional two-lane road. However, since the yellow line can be accurately recognized by the above configuration, such a composite line is used. It becomes possible to recognize correctly.

In many cases, rubber poles and concrete blocks are installed on the center line of the provisional two-lane road. In such cases, the maintenance of the composite line is not sufficient, and the yellow line is faded. There may be. On the other hand, according to the above configuration, since the edge of the yellow line is easily extracted, even a yellow line in a faint state can be accurately recognized.

In the attached drawing:
It is a block diagram which shows the structure of a lane marking recognition apparatus. It is a flowchart of a lane marking recognition process.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. The embodiment of the present invention is not limited to the following embodiment, and can take various forms as long as they belong to the technical scope of the present invention.

[Description of configuration]
The lane marking recognition device 10 of this embodiment is configured to give a lane departure warning to a driver, and includes a camera 11, a buzzer 12, a communication unit 13, a control unit 14, and the like (see FIG. 1). The lane marking recognition device 10 includes a rain sensor 15 that detects the amount of rainfall, a vehicle speed sensor 16 that detects the vehicle speed of the vehicle VE, a yaw rate sensor 17 that detects the yaw rate of the vehicle, and the like. This lane marking recognition device 10 is mounted on a vehicle VE.

The control unit 14 includes a CPU, a ROM, a RAM, an I / O, and the like, and is a part that performs overall control of the lane marking recognition apparatus 10 according to a program stored in the ROM or the like.
The camera 11 is configured as a CCD camera, an infrared camera, or the like, and captures a road surface around the vehicle or in front of the vehicle, and generates a road surface image that is a color image.

The buzzer 12 is a part that emits various warning sounds in response to instructions from the control unit 14.
Moreover, the communication part 13 is a site | part which communicates with other ECU via in-vehicle LAN which is not shown in figure.

[Description of operation]
(1) Outline The lane marking recognition device 10 of the present embodiment performs edge extraction from the road surface image generated by the camera 11, detects the lane marking of the road that is running based on the extraction result, and The division line is recognized by determining the type (whether it is a solid line or a broken line).

In addition, it is good also as a structure which recognizes a division line only by detecting a division line. In addition, the lane marking recognition device 10 further recognizes the type of the composite line based on the recognition result of the lane line when the traveling road is provided with a composite line composed of a plurality of lane markings. Also good.

Further, the lane marking recognition device 10 predicts the traveling locus of the vehicle based on the rainfall amount detected by the rain sensor 15, the vehicle speed detected by the vehicle speed sensor 16, the yaw rate detected by the yaw rate sensor 17, and the like.

Then, the lane marking recognition device 10 grasps the lane that the vehicle should travel from the recognition result of the lane marking or the composite line, and when it is determined that there is a risk that the vehicle departs from the lane, the buzzer 12 sounds a warning sound. A lane departure warning is issued. In addition, when there exists a danger that a vehicle will deviate from a lane, it is good also as a structure which corrects the course of a vehicle by performing steering assistance and avoids the departure of a lane.

More specifically, the control unit 14 of the lane marking recognition device 10 converts the road surface image into a grayscale image. At this time, the control unit 14 calculates the average value of the luminance of each color component of R, G, B for each pixel of the road surface image, and in accordance with the average value, the luminance of each pixel in the grayscale image (gradation level) ) May be determined.

Thereafter, the control unit 14 sets a plurality of horizontal scanning lines in the grayscale image, compares the luminance of each pixel adjacent along the scanning line, and determines a difference between the luminances in advance. A point that has reached the threshold is extracted as an edge point of the lane marking.

It should be noted that the direction of the scanning line is not limited to the horizontal direction. Further, the control unit 14 may compare not only the adjacent pixels but also, for example, the luminance of pixels arranged at a predetermined interval along the scanning line.

Then, the control unit 14 detects the straight line by performing the Hough transform on the edge points, and selects a straight line candidate having a large number of votes for the Hough transform from these straight lines, and further narrows down the lane line candidates. Thus, a lane line candidate to be detected as a lane line is determined.

Specifically, the control unit 14 may narrow down the lane line candidates based on the contrast or brightness difference between the lane line candidates and an area (road surface area) where no lane line is drawn on the road surface. It is also possible to narrow down in consideration of various characteristics such as line thickness and total extension distance. Moreover, the control part 14 may narrow down in consideration of the distance of the horizontal direction from the center of a vehicle to a lane marking candidate.

Also, the lane marking recognition device 10 determines the type of the lane marking based on the number of edge points extracted for the detected lane marking. Specifically, when the number of edge points has reached the determination threshold, the type of lane marking is determined to be a solid line, and when the number of edges has not reached the determination threshold, the lane marking Is determined to be a broken line.

Here, in Japan and the like, a white lane line (white line) is usually used. However, for example, when there is some circumstance such as prohibition of an overtaking lane for overtaking, yellow or Orange demarcation lines (yellow lines) are used.

When this yellow line is photographed, the yellow line has a higher R brightness than the white line, but the G and B brightness is lower, and the B brightness is lower than the road surface area. For this reason, the yellow line has a lower contrast to the road surface region than the white line, and it is difficult to extract the edge, and as a result, the accuracy of recognizing the yellow line is lowered.

Therefore, the lane marking recognition device 10 according to the present embodiment improves the extraction sensitivity of yellow line edge points while driving on roads using yellow lines (yellow line roads) (yellow line edge points are extracted). Make it easier).

Also, since it becomes difficult to extract the edge point of the yellow line, there is a high possibility that the yellow line that is a solid line is erroneously determined as a broken line when determining the type of the yellow line. For this reason, the lane marking recognition device 10 of the present embodiment determines the type of yellow line by making it easier to determine that the lane line is a solid line (improves the determination sensitivity of the solid line) while traveling on the yellow line road. Improve accuracy.

As a result, the accuracy of recognizing the yellow line is improved, and as a result, the type of the composite line including the yellow line can be accurately recognized, and a more appropriate lane departure warning can be performed.
(2) About lane marking recognition processing Next, lane marking recognition processing for recognizing lane markings drawn on the road surface in front of the vehicle will be described with reference to the flowchart shown in FIG. This process is executed at regular intervals, that is, periodically by the control unit 14 of the lane marking recognition device 10. In the following description of the flowchart, the symbol S indicates a procedure (step) executed by the control unit 14.

In S100, the control unit 14 determines whether the yellow line edge point extraction sensitivity is improved, and if an affirmative determination is obtained (S100: Yes), the process proceeds to S125. When a negative determination is obtained (S100: No), the process proceeds to S105.

In S105, the control unit 14 determines whether or not the vehicle has entered the yellow road, and when an affirmative determination is obtained (S105: Yes), the process proceeds to S110 and a negative determination is obtained. If yes (S105: No), the process proceeds to S145.

Specifically, for example, the control unit 14 analyzes a color road surface image photographed by the camera 11, and on the road surface in front of the vehicle, a yellow line (or yellow line of a certain length along the traveling direction of the vehicle). It may be determined whether or not a composite line including a line is drawn. And the control part 14 may determine with the vehicle having started driving | running | working on the yellow line road, when such a yellow line is drawn.

In addition, it is known that the yellow line is used on provisional two-lane roads in Japan, car pool lanes in the United States, construction sections in Europe, and the like.
Therefore, the control unit 14 may communicate with an ECU such as a navigation device (not shown) via the communication unit 13 to acquire the type information of the road on which the vehicle is traveling, and determine the type of the road. . The type information may be stored, for example, in map data included in the navigation device, or may be information acquired from the outside via a radio beacon or the like.

Then, the control unit 14 may determine that the vehicle has started traveling on the yellow road when the road on which the vehicle is traveling is the above-described provisional two-lane road or the like.
In S110, the control unit 14 shifts to the yellow line mode and improves the extraction sensitivity of the yellow line edge point (S115).

Specifically, the control unit 14 may set the edge threshold for extracting the edge of the lane marking from the road surface image as a yellow line mode threshold smaller than the normal mode threshold. This makes it easier to extract the edges of all lane markings regardless of the color, so that the sensitivity of extracting yellow line edge points is improved, and consequently the accuracy of recognizing yellow lines is improved. .

Also, as described above, the yellow line on the road surface image has a higher luminance of R than the white line. Focusing on this point, the control unit 14 weights the R luminance of each pixel during the yellow line mode (for example, a state in which the R luminance of each pixel is multiplied by a coefficient larger than 1). Thus, the road surface image may be converted into a gray scale image.

By doing so, the contrast of the yellow line with respect to the road surface area on the gray scale image is increased, so that the sensitivity of extracting the edge point of the yellow line is improved, and as a result, the accuracy of recognizing the yellow line is improved.

In addition, it is good also as a structure which performs only one of the fall of an edge threshold value, and the weighting of the brightness | luminance of R. Even in such a case, the extraction sensitivity of the yellow line edge point can be improved.

In subsequent S120, the control unit 14 improves the determination sensitivity of the solid line by setting the determination threshold used for determining the type of the lane marking to a yellow line mode threshold that is smaller than the normal mode threshold (the lane marking). Is easily determined as a solid line), and the process proceeds to S145. Thereby, it can prevent that the yellow line which is a solid line is misjudged as a broken line, and can improve the precision which recognizes a yellow line.

Note that, when the mode is shifted to the yellow line mode, only the yellow line edge point extraction sensitivity may be improved without improving the solid line determination sensitivity.
On the other hand, in S125, the control unit 14 determines whether or not traveling on the yellow road has ended, and if an affirmative determination is obtained (S125: Yes), the process proceeds to S130 and a negative determination is made. Is obtained (S125: No), the process proceeds to S145.

Specifically, the control unit 14 analyzes the road surface image photographed by the camera 11 in the same manner as in S105, and a yellow line having a certain length is not drawn along the traveling direction of the vehicle. It may be determined that traveling on the yellow road has ended.

In addition, the control unit 14 determines the type of road on which the vehicle is traveling from the type information acquired from the navigation device or the like in the same manner as S105. For example, when the road on which the vehicle is traveling is not a provisional two-lane road or the like, Alternatively, it may be determined that traveling on the yellow road has ended.

In S130, the control unit 14 shifts to the normal mode, returns the edge threshold value to the normal mode threshold value, and cancels the weighting of the luminance of R at the time of conversion to the grayscale image. The extraction sensitivity is restored (S135).

In subsequent S140, the control unit 14 restores the determination sensitivity of the solid line by setting the determination threshold used for determining the type of the lane marking as the normal mode threshold, and the process proceeds to S145.
In S145, the control unit 14 converts the road surface image to generate a grayscale image while reflecting whether the mode is the yellow line mode or the normal mode, and extracts the edge of the lane marking from the grayscale image. And recognize the lane markings.

In subsequent S150, the control unit 14 determines the type of the lane marking based on the number of edge points extracted for the detected lane marking in a state reflecting whether the mode is the yellow line mode or the normal mode, This process ends.

[Other Embodiments]
(1) The lane marking recognition apparatus 10 according to the present embodiment reduces the edge threshold during the yellow line mode, and converts the road surface image into a grayscale image in a state where the luminance of R of each pixel is weighted. Thus, the yellow line edge point extraction sensitivity is improved. However, the yellow line edge point extraction sensitivity may be improved as follows.

That is, in S145, the control unit 14 of the lane marking recognition device 10 does not recognize the lane marking based on the luminance of the grayscale image but recognizes the lane marking based on the luminance R of the road surface image in the yellow line mode. You may do it.

Specifically, the control unit 14 sets a plurality of horizontal scanning lines in the road surface image, compares the R luminance of each pixel adjacent along the scanning line, and the difference in luminance is an edge threshold value. You may extract the point which reached | attained as an edge point of a lane marking. Then, based on the extracted edge point, the lane marking may be detected in the same manner.

Note that it is desirable to set the edge threshold so that the white line and the yellow line edge can be extracted, considering that the white line has a lower R luminance than the yellow line. At this time, a threshold smaller than the threshold used during the normal mode may be used as the edge threshold.

Even in such a case, the extraction sensitivity of the yellow line edge point can be improved, and as a result, the accuracy of recognizing the yellow line can be improved.
(2) The lane marking recognition device 10 of the present embodiment is normally running on a yellow line road on the assumption that a white line is used as a road lane marking, but a yellow line is used in some circumstances. In this configuration, the accuracy of recognizing yellow lines is improved. However, depending on the country or region, there may be a possibility that a lane marking of another color is used instead of the white line, or a lane marking of another color (specific color) may be used instead of the yellow line.

In such a case, when the vehicle is traveling on a road on which a specific color lane marking is used, the lane marking recognition device 10 performs edge points of the specific color lane marking in the same manner as in this embodiment. It is conceivable to improve the extraction sensitivity.

Specifically, the extraction sensitivity is improved by converting the road surface image into a grayscale image with the luminance of the color component corresponding to the specific color in each pixel being weighted, or by reducing the edge threshold value. Also good. In addition, as in (1), an edge point may be extracted from the road surface image based on the luminance of the color component corresponding to the specific color, and a partition line may be detected based on the edge point.

By doing so, it is possible to improve the accuracy of recognizing the specific line marking.

In the above embodiment, the camera 11 corresponds to the photographing means. Further, S105 of the lane marking recognition process corresponds to a road determination means, S115 corresponds to an extraction means, S120 corresponds to a recognition means, S145 corresponds to an extraction means and a recognition means, and S150 corresponds to a recognition means.

DESCRIPTION OF SYMBOLS 10 ... Marking line recognition apparatus, 11 ... Camera, 12 ... Buzzer, 13 ... Communication part, 14 ... Control part, 15 ... Rain sensor, 16 ... Vehicle speed sensor, 17 ... Yaw rate sensor.

Claims

Photographing means (11) for photographing a road surface of a road in front of the vehicle and generating a road image that is a color image;
By detecting a change in the color of the road surface image along a predetermined scanning line and comparing the degree of the change with a predetermined edge threshold value, the edge of the division line drawn on the road surface is extracted. Extracting means (S115, S145),
Recognition means (S120, S145, S150) for recognizing the lane marking based on the edge extracted by the extraction means;
An area in which the lane marking is not drawn on the road surface is defined as a road surface area, and the lane marking of a specific color whose degree of color change with respect to the road area is smaller than a normal color that is normally used as the color of the lane marking. Road determination means (S105) for determining whether or not the vehicle is traveling on the specified road;
With
The extraction means improves an extraction sensitivity, which is a sensitivity when extracting the edge of the lane marking of the specific color when the road determination means determines that the vehicle is traveling on the specific road. thing,
A device for recognizing lane markings.
The apparatus of claim 1.
The normal color is white, and the specific color is yellow to orange,
A device characterized by.
The apparatus of claim 1 or claim 2,
The extraction means converts the road surface image into a gray scale image, detects a change in shade on the gray scale image as a color change on the road surface image, and the road determination means causes the vehicle to perform the identification. When it is determined that the vehicle is traveling on a road, the extraction sensitivity is improved by converting the road surface image into the grayscale image in a state where the color component corresponding to the specific color is weighted. ,
A device characterized by.
The apparatus of claim 1 or claim 2,
When the road determination unit determines that the vehicle is traveling on the specific road, the extraction unit displays a change in luminance of a color component corresponding to the specific color on the road image on the road image. Improving the extraction sensitivity by detecting as a color change of
A device characterized by.
The apparatus according to any one of claims 1 to 4,
The extraction means improves the extraction sensitivity by lowering the edge threshold when the road determination means determines that the vehicle is traveling on the specific road;
A device characterized by.
The device according to any one of claims 1 to 5,
The recognition means is
The lane marking is detected based on the edge extracted by the extraction means, and the lane marking is determined to be a solid line when the number of edges related to the lane marking reaches a predetermined threshold for determination. At the same time, if the number of edges does not reach the threshold for determination, it is determined that the partition line is a broken line,
Lowering the threshold for determination when the road determination means determines that the vehicle is traveling on the specific road;
A device characterized by.
Photograph the road surface of the road in front of the vehicle, generate a road image that is a color image,
Detecting a change in the color of the road surface image along a predetermined scanning line;
By comparing the degree of change and a predetermined edge threshold, the edge of the lane marking drawn on the road surface is extracted, and the lane marking is recognized based on the extracted edge,
An area in which the lane marking is not drawn on the road surface is defined as a road surface area, and the lane marking of a specific color whose degree of color change with respect to the road surface area is smaller than a normal color normally used as the color of the lane marking. Whether the vehicle is traveling on the specified road,
The extraction improves an extraction sensitivity, which is a sensitivity when extracting the edge of the lane marking of the specific color, when the vehicle is determined to be traveling on the specific road;
A method for recognizing lane markings characterized by