JP2013003634A - Partition line recognition device, partition line recognition method, program and medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a partition line recognition device capable of appropriately promoting a driver to maintain driving his/her own vehicle in a center of the own lane by issuing alarm more properly.SOLUTION: A partition line recognition device 1 includes: imaging means 2 for imaging an area including a road surface ahead of the own vehicle; edge point extraction part 3a for extracting edge points in the road surface from image information output by the imaging means 2; a line segment extraction means 3b for extracting a line segment group having a possibility of corresponding to a partition line from the edge points; and detection means 3c for detecting a line segment corresponding to the partition line from the line segment group. The device 1 also includes settlement means 3d for settling a detection when the detection means 3c successively detects a line segment the number of times more than a prescribed re-detection frequency M after the detection means 3c did not detect a line segment the number of times more than a prescribed non-detection frequency N.

Description

  The present invention performs lane maintenance support control for assisting the host vehicle to travel while maintaining the vicinity of the center of the travel lane in a vehicle such as a passenger car, a truck, a bus, and the like. The present invention relates to a lane marking recognition device, a lane marking recognition method, a program, and a medium, which are mainly used for a driving support device capable of securing

  Conventionally, Patent Document 1 discloses a lane marking recognition device used for a driving assistance device that promotes the vehicle to travel safely and stably while maintaining the left and right positions of the vehicle in the vicinity of the center in the vehicle lane. There is something like that.

  In such a lane marking recognition device, an imaging unit such as a camera whose optical axis is directed in front of the host vehicle is installed on the lower surface of the roof or the front panel of the host vehicle in the vicinity of the windshield, for example. The image ahead of the subject vehicle is subjected to processing such as binarization processing, differentiation processing, and filter processing to detect and recognize the lane markings on both sides of the subject lane where the subject vehicle is located.

  When it is determined that the vehicle may deviate to some extent from the center of the detected lane markings on both sides, a warning by vibrating a speaker or a steering wheel in the driving assistance device, or A warning is generated by generating a steering force, that is, an alarm torque so that the steering apparatus of the own vehicle is directed in a direction opposite to the direction in which the own vehicle deviates.

  Based on this warning, the driver of the host vehicle recognizes that the host vehicle has deviated from the center in the host lane, and appropriately performs a steering operation so that the host vehicle travels along the center in the host lane. be able to. Alternatively, driving assistance for controlling the steering itself can be performed in accordance with the alarm. In general, the former is called LDW (Lane Departure Warning) and the latter is called LKA (Lane Keep Assist).

JP 2007-179386 A

  However, in such a lane marking recognition apparatus described in Patent Document 1, the validity of recognition is determined based on the number of times of successful extraction of lane marking edges, but is once undetected. There is no particular consideration or mention regarding the re-detection after that.

  Here, when the lane line on the unrecognized side is detected even in one frame after it has once been unrecognized, if the detected lane line is confirmed, if the detection of that one frame is a false detection, A reduction in output accuracy due to erroneous detection and an inappropriate alarm associated with a reduction in accuracy will be caused. In particular, there is a problem that it is difficult to cope with erroneous detection due to road surface noise, tar marks, tire marks, and the like that show characteristics close to lane markings in an image.

  In view of the above problems, an object of the present invention is to provide a lane marking recognition device, a lane marking recognition method, a program, and a medium that can more appropriately redetect a lane marking on a side that has not been detected once. .

In order to solve the above problem, the lane marking recognition device according to the present invention is:
Imaging means for imaging including the road surface in front of the vehicle;
Edge point extraction means for extracting edge points in the road surface from image information output by the imaging means;
Line segment extraction means for extracting a line segment group having the possibility of corresponding to the partition line from the edge point;
Detecting means for detecting a line segment corresponding to the partition line from the line segment group,
When the detection by the detection means is not performed more than a predetermined number of undetected times, a confirmation means is provided for confirming the detection when the detection by the detection means is continued for a predetermined number of times of re-detection. .

In the lane marking recognition device,
The detecting means may detect the line segment group constituting the line segment group as the line segment when a predetermined condition is satisfied.

In order to solve the above problem, a lane marking recognition method according to the present invention includes:
An imaging step for imaging including a road surface in front of the vehicle;
An edge point extracting step for extracting an edge point in the road surface from the image information output in the imaging step;
A line segment extracting step of extracting a line segment group having a possibility of corresponding to the partition line from the edge point;
Detecting a line segment corresponding to the lane marking from the line segment group, and
When the detection in the detection step is not performed more than a predetermined number of undetected times, the method includes a confirmation step for determining the detection when the detection in the detection step is performed continuously for a predetermined number of times of re-detection or more. .

  The program according to the present invention is a program for executing the lane marking recognition method, and the medium according to the present invention is a medium storing the program.

  According to the present invention, when re-detecting the lane marking on the side once undetected, it is possible to prevent re-detection as much as possible and perform re-detection more appropriately.

It is a block diagram which shows one Embodiment of the lane marking recognition apparatus 1 of the Example based on this invention. It is a schematic diagram which shows the specific aspect of the road surface to which the control content of one Embodiment of the lane marking recognition apparatus 1 of the Example which concerns on this invention is applied. It is a schematic diagram which shows the concept of the predetermined | prescribed undetected frequency | count N used for control of one Embodiment of the lane marking recognition apparatus 1 of the Example which concerns on this invention. It is a schematic diagram which shows the concept of the predetermined | prescribed redetection frequency M used for control of one Embodiment of the lane marking recognition apparatus 1 of the Example which concerns on this invention. It is a flowchart which shows the control content of one Embodiment of the lane marking recognition apparatus 1 of the Example which concerns on this invention. It is a schematic diagram which shows the state transition aspect in the control content of one Embodiment of the lane marking recognition apparatus 1 of the Example which concerns on this invention. It is a schematic diagram which shows the correlation aspect of each step and state transition of the control content of one Embodiment of the lane marking recognition apparatus 1 of the Example which concerns on this invention.

  Hereinafter, embodiments for carrying out the present invention will be described with reference to the accompanying drawings.

  As shown in FIG. 1, the lane marking recognition device 1 of this embodiment includes a front recognition camera 2, an LDWECU 3 (Electronic Control Unit), and a buzzer 4.

  The front recognition camera 2 (imaging means) is, for example, a CCD camera or a CMOS camera. For example, the front recognition camera 2 (imaging means) is disposed at the center upper part of the windshield in the own vehicle room so that the optical axis coincides with the front, An image is captured in a range including both sides of the lane, and the captured image information is output to the LDWECU 3.

  The LDWECU 3 includes, for example, a CPU, a ROM, a RAM, a data bus connecting them, and an input / output interface. The CPU performs predetermined processing according to a program stored in the ROM, and an edge point for performing the processing described below. The extraction unit 3a, the line segment extraction unit 3b, the detection unit 3c, the determination unit 3d, the alarm unit 3e, the state determination unit 3f, and the state transition unit 3g are configured.

  The edge point extraction means 3a of the LDWECU 3 acquires image information from the forward recognition camera 2, and binarizes and differentiates edge points in the road lane located in the image information captured by the forward recognition camera 2. Extract by etc. The line segment extracting means 3b performs Hough transform on the group of edge points and extracts a line group having a possibility of corresponding to a white line (division line) on the road surface. The detection unit 3c detects a line segment corresponding to a white line when the line segment candidates constituting the line segment group satisfy a predetermined condition. The predetermined condition refers to a condition in which, for example, two pairs of line segments separated from each other by the width of the road on which the vehicle passes have a width dimension equivalent to a white line.

  The determination means 3d of the LDWECU 3 estimates and determines the road parameters (road shape: lane width, radius R, yaw angle, pitch angle) by the Kalman filter from the detected position of the line segment corresponding to the detected white line. The Kalman filter refers to a general function for estimating the position of a target in the present, future, and past by utilizing a law that depends on time change, but a detailed description is omitted.

  From the road parameter estimated as a line segment corresponding to the detected white line, the alarm means 3e of the LDWECU 3 recognizes the white line on the left, right or both sides of the road surface, and the white line exists on both the left and right sides. In this case, the offset distances W1 and W2 from them are calculated, and when only one of them exists, the offset distance W3 with the left or right white line is calculated.

  In addition, the determination unit 3d of the LDWECU 3 performs the operation again only when the white line is detected again after the predetermined re-detection number M or more when the detection unit 3c does not detect the white line more than the predetermined non-detection number N. A process for confirming the detection of the white line is performed. Further, the alarm means 3e of the LDWECU 3 outputs a buzzer 4 so as to output an alarm to the driver when any of the offset distances W1 to W2 or the offset distance W3 is less than the threshold value Wth and the alarm condition is satisfied. Ring.

  In this embodiment, when a white line has not been detected for a predetermined number of undetected times N or more, the detection is again input to the Kalman filter and confirmed unless detected again for the predetermined number of redetections M or more. Describe the significance of not doing In general, both LDW and LKA are assumed for vehicles traveling on highways and automobile roads. In this example, this is applied when there are noise objects such as irregularities, tar marks, and tire marks on general roads. Assumes that

  For example, consider the form of a general road with two or more lanes on one side as shown in FIG. As shown in FIG. 2, when the vehicle is traveling on the left lane, a solid white line is located on the left side of the vehicle, and a broken white line is located on the right side. At this time, the length of the broken line in the traveling direction of the vehicle is L1 (m), the interval between the broken lines is L2, and the length of the noise target on the road surface is L3. It is assumed that the noise target is similar in shape to the white line, for example, an elongated rectangular shape or parallelogram shape surrounded by parallel straight lines, or an elliptical shape elongated in the traveling direction.

  Here, in general, when extracting edge points from image information, line scanning is not performed for the entire area of the image information, but the vehicle bonnet and night illumination range are taken into account in order to reduce the processing load. On the other hand, the range of the effective distance L4 is line scanned. The vehicle speed is V (m / s), and the number of images acquired per second (number of frames) is A / s.

  The predetermined undetected number of times N is basically set based on a broken line cycle (passing time between broken lines). In the left diagram of FIG. 3, the lower limit of the range of the effective distance L4 is t1 seconds from the time when it touches the lower broken line, the vehicle moves upward at the speed V, and the upper broken line is the range of the effective distance L4. It shall be in contact with the upper limit of.

  In this case, the time t1 for passing through the broken line interval is expressed by t1 = (L2−L4) / V, and it is necessary to re-detect the non-detection due to the broken line interval once. The lower limit value needs to satisfy N ≧ A × t1. When securing the amount of information in the image information necessary for the extraction by the edge point extracting means 3a is considered, both the left and right diagrams in FIG. When considering the existence of a noise source showing a discrete and general distribution such as road surface reflected light that is different from the noise target in FIG. 2 described above, an appropriate correction that raises the lower limit value to the larger side is considered. I do.

  The predetermined number of re-detections M is also basically set based on the duration of the noise target image information. In the left diagram of FIG. 4, the upper limit of the range of the effective distance L4 is t2 seconds from the time when it is in contact with the upper side of the noise target, the vehicle moves upward at the speed V, and the lower side of the noise target is the effective distance L4. It shall touch the lower limit of the range. Note that the predetermined re-detection number M is set larger than the predetermined non-detection number N.

  In this case, the passage time t2 for the noise target to be excluded from the detection target to pass through the range of the effective distance L4 is represented by t2 = (L4−L3) / V, and it is necessary to exclude redetection due to the detection of the noise target. Therefore, the lower limit value of the predetermined undetected number needs to satisfy M ≧ A × t2. In this case as well, when securing the amount of information in the image information necessary for the extraction by the edge point extraction means 3a is considered, the value of L4-L3 is set to be small, and the noise object in FIGS. In consideration of the presence of noise sources having different discrete distributions, etc., appropriate correction is performed to raise the lower limit value to the larger side.

  That is, in this embodiment, the predetermined number of non-detections N and the predetermined number of re-detections M are defined separately, and in the former, non-detection due to the interval between the broken lines on the general road is prevented, and in the latter, noise is detected. It is supposed to prevent erroneous detection at the time of re-detection after it is once undetected.

  Hereinafter, the control content of the lane marking recognition apparatus 1 of the first embodiment will be described using the flowchart shown in FIG. 5 and the state transition diagram shown in FIG.

  In step S1 of FIG. 4, the state determination means 3f of the LDWECU 3 shown in FIG. 1 has detected whether the detection means 3c has continuously detected one or the other of the white lines that are assumed to be located on the left and right twice or more. Determine.

  If it is determined as affirmative in step S1, the state transition means 3g shown in FIG. 1 of the LDWECU 3 turns on the detection transient state flag and changes the recognition status from the undetected state C0 shown in FIG. 6 to the detected transient state C1. Transition.

  In step S3, the state determination means 3f of the LDWECU 3 determines whether the determined road parameter is deviating from the parameter stored in advance and is abnormal, or whether both of the white lines are not detected. If the determination is affirmative, the state transition means 3g proceeds to step S29 to turn off the detected transient state flag, shifts from the detected transient state C1 shown in FIG. 6 to the undetected state C0, and proceeds to step S4 if negative. .

  In step S4, the state determination means 3f of the LDWECU 3 determines whether or not one has been detected ten times in succession. If affirmative, the process proceeds to step S5, the detection transient state flag is turned off, and the state transition means 3g If one is right, the process shifts to the right detection state C2 shown in FIG. 6; if the other is the left, the process moves to the left detection state C6 shown in FIG.

  In step S6, the state determination unit 3f of the LDWECU 3 determines whether or not both have been detected ten times in succession. If yes, the process proceeds to step S7, and the state transition unit 3g turns on the both-side detection success state flag. The process proceeds to both detection success states C5 shown in FIG. 6, and if no, the process proceeds to step S19.

  In step S8, the state determination means 3f of the LDWECU 3 determines whether or not the determined road parameter is deviated from the parameter stored in advance and is abnormal, or whether the white line is not detected ten times continuously. If it is affirmative, the state transition means 3g proceeds to step S9, turns off the both detection success state flag, shifts from the both detection success state C5 shown in FIG. If so, the process proceeds to step S10.

  In step S10, the state determination means 3f of the LDWECU 3 determines whether or not the other of the ones in step S4 is undetected, and if the state transition means 3g is affirmative, it proceeds to step S11 and sets the both detection success state flag. Then, the process proceeds to step S12, the other instantaneous lost state flag is turned on, and both detection successful states C5 shown in FIG. 6 change to the left instantaneous lost state C4 (right detection in progress), or both detection successful states C5 to right Transition to the instantaneous lost state C8 (while left detection is in progress). If no, return to the step before step S10.

  In step S13, the state determination means 3f of the LDWECU 3 determines whether or not the other non-detection is continued 10 times. If the result is affirmative, the process proceeds to step S17, and the state transition means 3g turns off the other instantaneous lost flag. 6, if the left instantaneous lost state C4 (right detection is in progress) in FIG. 6, the right detection success state C2 is entered, and if the right instantaneous lost state C8 (left detection is in progress), the left detection success state C6 is entered, If not, the process proceeds to step S14.

  In step S14, the state determination means 3f of the LDWECU 3 determines whether or not the number of undetected times in step S13 is equal to or greater than the predetermined number of undetected times N. If affirmative, the process proceeds to step S15; Proceed.

  In step S15, the state determination means 3f determines whether or not the other redetection continues for a predetermined redetection number M or more. If the determination is affirmative, the process proceeds to step S16. If the determination is negative, the process returns to step S13. .

  In step S18, the state determination means 3f determines whether or not the other redetection is continued once. If the determination is affirmative, the process proceeds to step S16. If the determination is negative, the process returns to the step before step S13.

  In step S16, the state transition means 3g of the LDWECU 3 turns off the other instantaneous lost state flag, proceeds to step S7, turns on the both-side detection success state flag, and shifts to the both detection success state C5 shown in FIG. Thereafter, the process returns to B before step S7.

  After the negative determination in step S6 or after the processing of step S17, step S19 is executed, and the one-side detection success flag is turned on by the state transition means 3g, and the right detection success state C2 or the left detection success shown in FIG. Transition is made to state C6.

  In step S20, the state determination means 3f of the LDWECU 3 determines whether or not one of the steps S19 is undetected, and the state transition means 3g proceeds to step S21 if affirmative, and turns off the one detection success state flag. Further, proceeding to step S22, the instantaneous lost state flag is turned on, and the right detection successful state C2 shown in FIG. 6 is shifted to the right instantaneous lost state C3, or the left detection successful state C6 is shifted to the right instantaneous lost state C7. If no, the process returns to the step S20.

  In step S23, the state determination means 3f of the LDWECU 3 determines whether or not one undetected state is continuous ten times, and if affirmative, the process proceeds to step S27, and the state transition means 3g turns off the one detection successful state flag. Then, the state is shifted from the right instantaneous lost state C3 or the left instantaneous lost state C7 in FIG. 6 to the undetected state C0, and if negative, the process proceeds to step S24.

  In step S24, the state determination means 3f of the LDWECU 3 determines whether or not the number of undetected times in step S13 is equal to or greater than a predetermined number of undetected times N. If affirmative, the process proceeds to step S15; Proceed.

  In step S25, the state determination unit 3f determines whether one redetection continues for a predetermined redetection number M or more. If the determination is affirmative, the process proceeds to step S26, and if the determination is negative, the process returns to the step before step S23. .

  In step S28, the state determination means 3f determines whether or not one redetection is continued once. If the determination is affirmative, the process proceeds to step S26. If the determination is negative, the process returns to step S23.

  In step S26, the state transition means 3g of the LDWECU 3 turns off the one-time instantaneous lost state flag, and from the right instantaneous lost state C3 shown in FIG. 6 to the right detection successful state C2, or from the left instantaneous lost lost state C7 to the left detection successful state. Move to C6. Thereafter, the process returns to A before step S6.

  That is, step S10 to step S16 control the transition between the both detection success state C5 of FIG. 6 and the left instantaneous lost state C4 (right detection in progress) or right instantaneous lost state C8 (left detection in progress). S26 controls the transition between the right detection success state C2 and the right instantaneous lost state C3, and the left detection success state C6 and the left instantaneous lost state C7 in FIG. Note that the correlation between the step number in the flowchart of FIG. 5 and the state transition of FIG. 6 is shown in FIG.

  The lane marking recognition method of the present invention is executed according to the control contents described above. According to the lane marking recognition apparatus 1 of the present embodiment realized by these things, the following operational effects can be obtained.

  That is, the predetermined number of undetected times N and the predetermined number of redetected times M are defined separately, and the former is determined based on the broken line interval passage time t1 and the number of image acquisitions A. Detection can be prevented. The latter is determined based on the noise target passage time t2 and the image acquisition number A, and prevents false detection at the time of re-detection after the noise target is once undetected.

  As a result, even when the application target of LDW is extended to the general road, the accuracy of detecting the white line of the broken line is improved, and even for the noise target that is more frequently generated on the general road, it is caused by the noise target. The occurrence of erroneous detection can be prevented, and the detection accuracy can also be improved by this.

  In particular, for the purpose of reducing the processing load, the following advantages can be obtained when the range of the effective distance L4 described above is further limited in the range in the width direction. That is, when the scan range is narrowed down to a region narrower than the vehicle width centered on the already detected white line and the scan range is not narrowed on the side where the white line is not detected, the detection range is from non-detection to re-detection. In the process, there is a high possibility of picking up a noise object and erroneously detecting it, but according to the present embodiment, such erroneous detection can also be effectively prevented.

  By increasing the detection accuracy in this way, it is possible to suppress disturbance components corresponding to so-called observation noise before being input to the Kalman filter, and to increase the output accuracy of road parameters. As a result, the reliability of the alarm of the LDW itself can be improved.

  In addition, when the number of undetected times is not equal to or greater than the predetermined number of undetected times N in step S14 or step S24 of FIG. 5, redetection is determined by one detection. As a result, re-detection can be performed more smoothly when one of the broken white lines is faint due to wear.

  Although the preferred embodiments of the present invention have been described in detail above, the present invention is not limited to the above-described embodiments, and various modifications and substitutions are made to the above-described embodiments without departing from the scope of the present invention. be able to.

  Further, in the embodiment described above, the edge point extracting means 3a, line segment extracting means 3b, detecting means 3c, determining means 3d, alarm means 3e, determining means 3f, state determining means 3g, state transition means 3h Although the LDWECU 3 is used in configuring the above, it is possible to replace it with the LKA ECU. In this case, in addition to the alarm, the reliability of the support control can be improved.

  Further, in the above-described embodiment, a case where the left side of the road surface that is actually assumed to be applied is a solid white line and the right side is a broken white line. Even when solid white lines exist discontinuously, accurate re-detection can be performed based on the state transitions shown in the left detection success state C6 and the left instantaneous lost state C7 shown in FIG.

  The same applies to the right-side traffic road surface overseas. For example, even when a solid white line discontinuously exists only on the right side on a road surface without a center line, the right detection success state C2 and the right side shown in FIG. Based on the state transition indicated by the instantaneous lost state C3, accurate redetection can be performed.

  Further, regardless of whether inside or outside the country, even if there are white lines on both sides of the one-way road surface, but one or the other is discontinuous, both detection success state C5, left instantaneous lost state C4 shown in FIG. Based on the state transition of the right instantaneous lost state C8, accurate re-detection can be performed.

  The present invention relates to a lane marking recognition device, which prevents false detection by a noise object, particularly at the time of re-detection, eliminates an inappropriate alarm, and appropriately performs a more appropriate alarm. The driver can be encouraged to drive the vehicle in the center of the lane. In other words, the safety can be further improved, which is useful when applied to various vehicles such as passenger cars, trucks, and buses.

DESCRIPTION OF SYMBOLS 1 Marking line recognition apparatus 2 Front recognition camera (imaging means)
3 LDWECU
3a Edge point extraction means 3b Line segment extraction means 3c Detection means 3d Determination means 3e Alarm means 3f State determination means 3g State transition means 4 Buzzer C0 Undetected state C1 Detection transient state C2 Right detection success state C3 Right instantaneous lost state C4 Left instant Lost state (right detection in progress)
C5 Both detection success status C6 Left detection success status C7 Left instantaneous lost status C8 Right instantaneous lost status (left detection in progress)
N Predetermined number of detections M Predetermined number of redetections L1 Broken line length L2 Broken line interval L3 Noise target length L4 Effective distance V Vehicle speed A Number of image acquisitions per second

Claims (5)

  An imaging unit that captures an image including a road surface ahead of the vehicle, an edge point extraction unit that extracts an edge point in the road surface from image information output from the imaging unit, and a possibility that the edge point corresponds to the lane marking A line segment extracting means for extracting a line segment group having a line segment group and a detection means for detecting a line segment corresponding to the lane marking from the line segment group, and when the detection by the detection means is not performed a predetermined undetected number of times or more. A lane marking recognition apparatus comprising: a determination unit that determines the detection when the detection unit detects the number of times of re-detection continuously.   The lane marking recognition apparatus according to claim 1, wherein the detection unit detects the line segment when the line segment candidate constituting the line segment group satisfies a predetermined condition.   An imaging step for imaging including a road surface in front of the vehicle, an edge point extracting step for extracting an edge point in the road surface from image information output in the imaging step, and a possibility corresponding to the lane marking from the edge point A line segment extraction step for extracting a line segment group having a property and a detection step for detecting a line segment corresponding to the partition line from the line segment group, and the detection in the detection step is not performed more than a predetermined number of undetected times. The method further comprises a confirmation step of confirming the detection when the detection in the detection step is continuously performed for a predetermined number of times of re-detection.   A program for executing the lane marking recognition method according to claim 3.   A medium storing the program according to claim 4.

Images