JP2014127026A - Branch road determination device, boundary line recognition device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a branch road determination device that enables a suppression of a wrong detection of a branch road, and to provide a boundary line recognition device that uses the branch road determination device.SOLUTION: The branch road determination device comprises: a camera 11 that is mounted onto a vehicle, and photographs a travelling road; a boundary line candidate extraction part 20 that extracts a candidate line of a pair of right and left boundary lines on the basis of image information photographed by the camera 11; a vehicle information acquisition device 12 that calculates a curvature of the travelling road; and a branch road determination part 33 in which the larger a degree of a travelling road including a plurality of characteristics as a branch road is, the higher a probability that the travelling road is the branch road is calculated, which integrates a calculated probability about a plurality of characteristics and determines whether or not the travelling road is the branch road. The plurality of characteristics includes at least any of the candidate line extracted by the boundary line candidate extraction part 20 being a solid line and the curvature calculated by the vehicle information acquisition device 12 being smaller than a predetermined value.

Description

  The present invention relates to a branch path determination apparatus and a boundary line recognition apparatus that determine a branch path that branches from a main road based on image information captured by a camera mounted on a vehicle.

  In recent years, for the purpose of improving the vehicle's automatic control and driving safety, development of a recognition device for recognizing a boundary line of a road such as a white line based on image information taken by a camera mounted on the vehicle has been promoted ( For example, see Patent Document 1).

  In order to realize the automatic control of the vehicle and the improvement of traveling safety, it is necessary to suppress the erroneous recognition of the boundary line by the recognition device and recognize the boundary line appropriately. For this purpose, when there is a branch road that branches off from the main road, it is necessary to accurately discriminate between the main road and the branch road and suppress erroneous detection of the branch road.

  In Patent Literature 1, when the road curvature and the pitch angle of the vehicle are calculated based on the detected lane boundary position, and the absolute value of the difference between the calculated road curvature and the pitch angle and the past average value is larger than the threshold value In addition, the detected lane boundary position is recognized as the lane boundary position of the branch road.

JP 2005-346383 A

  The recognition device recognizes the boundary line by extracting an edge component representing the boundary line from the image information. For this reason, when a disturbance is applied to the edge component, the boundary line may be erroneously recognized, and the curvature and pitch angle calculated based on the boundary line may vary greatly. Therefore, when branching is determined based on the change in the boundary line as in the branch road determination method of Patent Document 1, there is a possibility that the main road is erroneously detected as a branch road.

  In view of the above circumstances, a main object of the present invention is to provide a branch path determination apparatus capable of suppressing erroneous detection of a branch path and a boundary line recognition apparatus using the branch path determination apparatus.

  In order to solve the above-described problem, a branch road determination device according to claim 1 is provided with a camera mounted on a vehicle and photographing a runway, and a pair of left and right that divide the runway based on image information photographed by the camera. A boundary candidate extraction unit that extracts a candidate line for a boundary line, a curvature calculation unit that calculates a curvature of the runway, and the greater the degree that the runway has a feature as a branch road, the more the runway is a branch road A branch path determination unit that calculates a certain probability high and integrates the probabilities calculated for a plurality of the characteristics to determine whether or not the road is a branch path, and the plurality of features are the boundary line candidate extraction At least one of the candidate line extracted by the unit being a solid line and the curvature calculated by the curvature calculation unit being smaller than a predetermined value.

  According to the branch path determination device of the first aspect, a candidate line of a pair of left and right boundary lines that divides the runway is extracted from image information captured by a camera mounted on the vehicle. In addition, the curvature of the runway is calculated. Furthermore, the probability that the runway is a branch road is calculated higher as the degree that the runway has the feature as a branch road is larger. The probabilities calculated for the characteristics of the plurality of branch roads including at least one of the extracted candidate lines being solid lines and the curvature of the road being smaller than a predetermined value are integrated, and the road is a branch road. It is determined whether or not.

  Here, the inventor of the present application pays attention to the fact that the boundary line of the branch road is often not a broken line but a solid line, and that a branch road is rarely installed on a sharply curved road having a curvature larger than a predetermined value. did. In addition, the determination of whether the candidate line is a solid line or a broken line and the determination that the runway is not a sharp curve are not easily affected by the disturbance applied to the edge component of the boundary line. Therefore, a branch road is erroneously detected by determining whether or not the road is a branch road with at least one of the candidate line being a solid line and the curvature of the road being smaller than a predetermined value. Can be suppressed.

  According to a second aspect of the present invention, there is provided a branch road determination device that is mounted on a vehicle and photographs a road, and a pair of left and right boundary lines that separate the road based on image information captured by the camera. The higher the probability that the runway has a feature as a branch road, the higher the probability that the runway is a branch road, the higher the probability that the runway has a feature as a branch road And a branch path determination unit that determines whether the runway is a branch path by integrating the probabilities calculated for a plurality of the features, and the plurality of features are extracted by the boundary line candidate extraction unit It includes both that the candidate line is a solid line and that the curvature calculated by the curvature calculation unit is smaller than a predetermined value.

  In the configuration of claim 2, the feature as the plurality of branch roads includes both that the extracted candidate line is a solid line and that the curvature of the runway is smaller than a predetermined value. This is different from the invention described in item 1. Therefore, since two features that improve the reliability are added and whether or not the branch path is determined, it is possible to determine the branch path more accurately.

The block diagram which shows the structure of the branch path determination apparatus which concerns on one Embodiment. The flowchart which shows the process sequence which determines a branch path. The table | surface which shows the probability which is a branch path with respect to a continuous line and a broken line. The figure which shows the probability which is a branch road with respect to the curvature of a runway. An arithmetic expression that integrates probabilities. The figure which shows an example of the determination result of a branched path.

  With reference to FIG. 1, the structure of the branch path determination apparatus (boundary line recognition apparatus) which concerns on one Embodiment is demonstrated. The branch path determination device according to the present embodiment includes a camera 11, a vehicle information acquisition device 12 (curvature calculation unit), and an image processing device 70.

  The camera 11 is a CCD camera, for example, and is fixed, for example, on the back side of a rearview mirror so as to face the front of the vehicle so as to capture the road ahead of the vehicle. The camera 11 captures the traveling road ahead of the vehicle and outputs the captured image information to the image processing device 70.

  The vehicle information acquisition device 12 includes a yaw rate sensor 12a and a vehicle speed sensor 12b. The vehicle information acquisition device 12 calculates the curvature at the center of the road from the angular velocity (yaw rate) in the turning direction of the host vehicle detected by the yaw rate sensor 12a and the vehicle speed detected by the vehicle speed sensor 12b, and calculates the curvature of the calculated road. The curvature is output to the image processing device 70.

  The image processing apparatus 70 is configured as a microcomputer including a CPU, a ROM, a RAM, an I / O, a bus line connecting these, and the like. In the present embodiment, the functions of the boundary line candidate extraction unit 20, the boundary line feature calculation unit 30, the boundary line feature integration unit 40, and the boundary line selection unit 50 are realized by the CPU executing a program stored in the ROM. ing.

  The boundary line candidate extraction unit 20 extracts a candidate line for a boundary line such as a pair of left and right white lines that divide the runway. The boundary line candidate extraction unit 20 continuously processes the image information acquired from the camera 11 at a predetermined sampling period, and a plurality of luminances that change abruptly in the horizontal direction intersecting the vehicle traveling direction on the image. Extract points as edge points. Then, Hough transformation is performed on the extracted plurality of edge points to extract a plurality of candidate lines for boundary lines that divide the left and right sides of the traveling road on which the host vehicle is traveling.

  The boundary line feature calculation unit 30 includes a vote count calculation unit 31, an edge strength calculation unit 32, and a branch path determination unit 33. The boundary line feature calculation unit 30 calculates the degree of having a feature as a boundary line of the main lane for each of the plurality of candidate lines extracted by the boundary line candidate extraction unit 20, and the degree of having the feature is calculated. The larger the value, the higher the probability that the candidate line is the boundary of the main road.

  As the characteristics of the boundary line of the main track, for example, the number of votes by the Hough transform is larger than a predetermined number, the edge strength represented by the luminance differential value in the horizontal direction is larger than the predetermined value, and is delimited by the boundary line The runway is not a branch road.

  The vote number calculation unit 31 is configured such that, when the number of votes by the Hough transform is smaller than a predetermined number, the probability that the candidate line is the boundary line of the main road is minimized, and the candidate line is recognized as the boundary line of the main road Suppress. When the edge strength at the edge point included in the candidate line is smaller than a predetermined value, the edge strength calculating unit 32 recognizes the candidate line as a boundary line with the lowest probability that the candidate line is a boundary line of the main road. It suppresses that. The branch road determination unit 33 determines whether or not the road delimited by the candidate line is a branch road, and if it is determined to be a branch road, the probability that the candidate line is a boundary line of the main road is set to 0, and the candidate line Is not recognized as the boundary of the main track.

  The boundary line feature integration unit 40 combines the probabilities calculated by the boundary line feature calculation unit 30 with respect to the boundary line features of a plurality of running roads, and integrates the probability that the candidate line is the boundary line of the main road running road. Note that the characteristics of the boundary line are not limited to the above three characteristics. For other boundary line features, the probability that the candidate line is the boundary line of the main road may be calculated, and the calculated probabilities may be integrated.

  The boundary line selection unit 50 selects a candidate line to be recognized as the boundary line of the main road based on the probability integrated by the boundary line feature integration unit 40. Specifically, among candidate lines having a high probability of being integrated, a candidate line that is paired on the left and right sides of the vehicle and that is closest to the vehicle is selected.

  When the vehicle is automatically controlled, the steering amount of the vehicle is set based on the shape of the candidate line selected by the boundary line selection unit 50. Further, when the vehicle is outside the selected candidate line, an alarm sound is output. That is, the candidate line selected as the boundary line of the main road is set as the boundary line of the control target, and various automatic controls are executed on the vehicle.

  Next, with reference to FIG. 2, the branch path determination processing procedure executed by the branch path determination unit 33 will be described in detail. In this branch road determination, the probability that the road is a branch road is calculated higher as the degree that the road divided by the candidate line has a feature as a branch road is larger. Then, the probabilities of the branch roads calculated for a plurality of features are integrated, and it is determined whether the running road is a branch road based on the integrated probabilities. In this branch path determination, it is determined whether the left candidate line is branched to the left and whether the right candidate line is branched to the right.

  The inventor of the present application paid attention to the fact that the boundary line of the branch road is often a solid line instead of a broken line, and that the branch road is rarely installed on a sharply curved road having a curvature larger than a predetermined value. Therefore, in S11 and S12, it is determined that the candidate line extracted by the boundary line candidate extraction unit 20 is a solid line and that the curvature of the road calculated by the vehicle information acquisition device 12 is smaller than a predetermined value, respectively. As a feature, the probability that the runway is a branch road is calculated.

  First, in S11, the probability that the running road is a branch road is calculated based on whether the candidate line is a solid line or a broken line. It is determined whether the candidate line is a solid line or a broken line. As shown in the table of FIG. 3, when the candidate line is a solid line, the probability that the runway is a branch road is set to a predetermined high probability, and the candidate line is a broken line. In this case, the probability that the runway is a branch road is set to a predetermined low probability. The predetermined high probability is a value that does not increase or decrease the integrated probability when the probabilities calculated for a plurality of features are integrated. That is, when the candidate line is a solid line, the probability that it is a branch path is not lowered, but it is not increased so as to be positively determined as a branch path. On the other hand, the predetermined low probability is a value that lowers the integrated probability when the probabilities calculated for a plurality of features are integrated, but is a value that does not set the probability of determining a runway as a branch road to zero. That is, even if the candidate line is a broken line, the runway may be determined as a branch road if the probability calculated for other features is sufficiently high.

  To determine whether a candidate line is a solid line or a broken line, edge points are continuously included on the candidate line, and if there is no missing part where no edge point exists, the candidate line is determined to be a solid line, and the edge point on the candidate line If there is a missing part that does not exist, the candidate line is determined to be a broken line.

  Next, in S12, the probability that the road is a branch road is calculated from the curvature of the road calculated by the vehicle information acquisition device 12. As shown in FIG. 4, the larger the curvature of the runway, the lower the probability that the runway is a branch road. Specifically, if the curvature of the runway is greater than the predetermined value, the runway has a very sharp curve and branch roads are rarely installed, so the probability of being a branch road is set to the predetermined minimum probability. Suppressing being determined as a branch road. If the curvature of the runway is smaller than the predetermined minimum curvature, that is, if the runway is a gentle curve or straight line, it is expected that a branch road is installed on the runway with normal probability, so it is a branch road The probability is set to a predetermined maximum probability, and it is not suppressed that the runway is determined to be a branch road. When the curvature of the runway is between a predetermined minimum curvature and a predetermined value, the probability of being a branch road is reduced as the curvature of the runway is larger. In S12, the probability that the left candidate line and the right candidate line are common branches is calculated.

  Next, in S13 to S17, the probability that the running road is a branch road is calculated for each feature as another branch road. In S13, the characteristic of the branch road is that the parallelism of the left and right candidate lines is lower than a predetermined value, and the higher the parallelism of the left candidate line and the right candidate line of the runway, the lower the probability of being a branch road. To do. The degree of parallelism decreases as the difference between the curvature calculated from the shape of the left candidate line and the curvature calculated from the shape of the right candidate line increases. For straight roads with the highest degree of parallelism, the probability of being a branch road is minimized. In S13, the probability that the left candidate line and the right candidate line are common branches is calculated.

  Next, in S14, the characteristic of the branch path is that the deviation amount of the curvature of the candidate line is larger than a predetermined amount, and the probability of the branch path is calculated to be higher as the curvature deviation amount is larger. The curvature divergence amount is a curvature divergence amount calculated from the shapes of the left and right candidate lines with respect to the curvature at the center of the road calculated by the vehicle information acquisition device 12.

  Next, in S15, the characteristic of the branch road is that the width of the runway is a branch pattern, and the probability that the runway is a branch road is calculated higher than when the width of the runway is wider than when it is not widened. When the runway extends to the left, the probability of being a left branch road is increased, and when the runway is extended to the right, the probability of being a right branch road is increased.

  Next, in S16, the fact that the branch information is displayed by the road surface display is characterized by the branch road, and the probability that the runway is a branch road is lower than when there is no display when the branch information is not displayed by the road surface display. calculate. Generally, branch information such as an arrow is often displayed on the road surface at a point where the branch road branches off from the main road. Therefore, a road surface display is acquired from the image information, and the acquired road surface display is compared with branch information such as arrows registered in the database. The less the branch information included in the road surface display, the lower the probability that the road is a branch road.

  Next, in S17, the fact that the branch information is displayed by the road marking is characterized by the branch road, and the probability that the runway is a branch road is lower than when there is no display when the branch information is not displayed by the road sign. calculate. In general, branch information such as an arrow is often displayed on a road sign at a point where a branch road branches off from a main road. Therefore, a road sign is acquired from the image information, and the acquired road sign is compared with branch information such as an arrow registered in the database. The less the branch information included in the road sign, the lower the probability that the road is a branch road.

  Subsequently, in S18, the integrated probabilities are calculated by integrating the probabilities of the branched paths calculated for the features as the plurality of branched paths in S11 to S17. Specifically, the integration probability is calculated using the arithmetic expression shown in FIG. First, a probability X obtained by integrating the probabilities calculated in S11 and S12 is calculated, where A is the probability calculated in S11 and B is a probability calculated in S12. Further, a probability X obtained by integrating the probabilities calculated in S11 to S13 is calculated with A being a probability X obtained by integrating the probabilities calculated in S11 and S12 and B being a probability calculated in S13. In this way, the probabilities are sequentially integrated, and the integrated probability obtained by integrating the probabilities calculated in S11 to S17 is calculated.

  Subsequently, in S19, it is determined whether or not the integration probability calculated in S18 is 50% or more. If the integration probability is not 50% or more (NO), it is determined in S20 that the runway is not a branch road. If the integration probability is 50% or more (YES), it is determined in S21 that the runway is a branch road.

  Subsequently, in S22, the determination result of the branch path in S20 and S21 is output to the boundary line feature integration unit 40. When the branch road determination unit 33 determines that the road is a branch road, the probability that the candidate line is a boundary line of the main road is 0. Therefore, the boundary line selection unit 50 recognizes the candidate line extracted by the boundary line candidate extraction unit 20 as the boundary line of the main road from candidate lines excluding the boundary line of the road determined to be a branch road. Select a candidate line.

  In FIG. 6, as an example of the branch path determination, the probability that the branch path is calculated in S11 to S14 when the curvature estimation accuracy calculated from the shape of the left candidate line is reduced due to the effect of the pitching of the vehicle. The result of determining whether or not a branch road is integrated is shown. That is, the candidate line is a solid line, the curvature of the runway is smaller than a predetermined value, the parallelism of the left and right candidate lines is lower than a predetermined value, and the deviation amount of the curvature of the candidate line is larger than the predetermined amount An example of the result of determination of a branch path is shown as a feature of the branch path.

  Due to the effect of pitching, the probability that the left candidate line is a left branch road calculated from the divergence amount of the left candidate line is increased by the process of S14, but it is determined that the left candidate line is a broken line by the process of S11. Reduce the integration probability that is the road. As a result, the probability of being an integrated left branch is 0.3348 (33.48%) and the probability of being an integrated right branch is 0.3260 (32.60%), both of which are less than 50% Thus, the main road is not erroneously determined as a branch road.

  According to this embodiment described above, the following effects are obtained.

  ・ The boundary line of the branch road is not a broken line but a solid line in many cases, and a branch road is rarely installed on a sharply curved road having a curvature larger than a predetermined value. In addition, the determination of whether the candidate line is a solid line or a broken line and the determination that the runway is not a sharp curve are not easily affected by the disturbance applied to the edge component of the boundary line. Therefore, it is determined whether the runway is a branch road, with both the fact that the candidate line is a solid line and the curvature of the runway calculated by the vehicle information acquisition device 12 being smaller than a predetermined value. Thus, erroneous detection of the branch path can be suppressed.

  • If the candidate line is not a solid line, the probability that the runway delimited by the candidate line is a branch road is low. Therefore, when the candidate line is not a solid line, the branch path can be accurately determined by setting the probability that the runway is a branch path to be lower than the probability when the candidate line is a solid line.

  ・ The higher the track curvature, the lower the probability of branching. Therefore, the larger the curvature of the runway, the lower the probability that the runway is a branch road, thereby suppressing erroneous detection of the branch road.

  ・ If the curvature of the runway is larger than the specified value, the probability that the runway is a branch road is set to the predetermined minimum probability, so if the probability calculated for the characteristics of other branch roads is not high enough, the runway will branch Not determined as a road. Therefore, erroneous detection of the branch path can be suppressed.

  -If the curvature of the runway is smaller than the predetermined minimum curvature, it is expected that a branch road is installed on the runway with normal probability. Therefore, when the curvature of the runway is smaller than a predetermined minimum curvature, it is not suppressed that the runway is determined to be a branch road by setting the probability that the runway is a branch road to a predetermined maximum probability.

  -By using the curvature of the runway calculated by the vehicle information acquisition device 12, it is possible to determine a branch road using the curvature of the runway that is not easily affected by disturbance. Therefore, the erroneous detection of the branch path can be further suppressed.

  When there is no display of branch information by road surface display, erroneous detection of a branch road can be further suppressed by lowering the probability that the candidate line is a branch road than when there is a display.

  -When there is no display of branch information by road signs, erroneous detection of branch roads can be further suppressed by lowering the probability that the candidate line is a branch road than when there is a display.

(Other embodiments)
The present invention is not limited to the description of the above embodiment, and may be modified as follows.

  In the branch path determination, it is only necessary to execute at least one of S11 and S12 and at least one of S13 to S17 without executing all the processes of S11 to S17. That is, the probability that the runway is a branch road may be calculated for a plurality of features of the branch road including at least one of the candidate line being a solid line and the curvature of the runway being smaller than a predetermined value. In addition, for at least one of the features that the candidate line is a solid line and the curvature of the runway is smaller than a predetermined value, the probability of being a branch road is calculated, and features other than the features of S11 to S17 are calculated on the branch road. A certain probability may be calculated.

  When the curvature calculated by the vehicle information acquisition device 12 is infinite, the probability of being a branch road is calculated to be smaller as the calculated curvature is larger so that the probability of being a branch road is a predetermined minimum probability. You may do it. That is, it is not necessary to set a predetermined value for the calculated curvature.

  -When the curvature calculated by the vehicle information acquisition device 12 is 0, the probability of being a branch road is calculated higher as the calculated curvature is smaller, so that the probability of being a branch road is calculated to a predetermined maximum probability. You may make it do. That is, the predetermined minimum curvature may not be set for the calculated curvature.

  When the curvature calculated by the vehicle information acquisition device 12 is larger than the intermediate curvature set between the predetermined value and the predetermined minimum curvature, the probability of being a branch road is calculated as, for example, a predetermined low probability, If it is small, the probability that it is a branch path may be calculated as, for example, a predetermined high probability. That is, the probability may be set to a discontinuous binary value.

  When the branch road determination unit 33 determines that the road is a branch road, the probability that the candidate line is a boundary line of the main road may be calculated as a probability other than 0, for example, 0.1.

  The curvature of the runway may be a curvature obtained by averaging the curvature calculated from the shape of the left candidate line and the curvature calculated from the shape of the right candidate line.

  DESCRIPTION OF SYMBOLS 11 ... Camera, 12 ... Vehicle information acquisition apparatus, 12a ... Yaw rate sensor, 12b ... Vehicle speed sensor, 20 ... Boundary line candidate extraction part, 33 ... Branch line determination part, 40 ... Boundary line feature integration part, 50 ... Boundary line selection part 70: Image processing apparatus.

Claims (10)

A camera (11) mounted on the vehicle and shooting the runway;
A boundary line candidate extraction unit (20) that extracts a candidate line of a pair of left and right boundary lines that divides the runway, based on image information captured by the camera;
A curvature calculator (12) for calculating the curvature of the runway;
The greater the degree that the runway has a feature as a branch road, the higher the probability that the runway is a branch road, and the probability calculated for a plurality of the characteristics is integrated to determine whether the runway is a branch road or not. A branch path determination unit (33) for determining whether or not
The plurality of features include at least one of the candidate line extracted by the boundary line candidate extraction unit being a solid line and the curvature calculated by the curvature calculation unit being smaller than a predetermined value. A branch path judging device characterized by the above. A camera mounted on the vehicle and shooting the runway;
Based on image information photographed by the camera, a boundary line candidate extraction unit that extracts a candidate line of a pair of left and right boundary lines that divides the runway;
A curvature calculator for calculating the curvature of the runway;
The greater the degree that the runway has a feature as a branch road, the higher the probability that the runway is a branch road, and the probability calculated for a plurality of the characteristics is integrated to determine whether the runway is a branch road or not. A branch path determination unit for determining whether or not
The plurality of features include both that the candidate line extracted by the boundary line candidate extraction unit is a solid line and that the curvature calculated by the curvature calculation unit is smaller than a predetermined value. A branch path determination device.   3. The branch path determination device according to claim 1, wherein the branch path determination unit calculates the probability lower than the case where the candidate line is a solid line when the candidate line is not a solid line.   4. The branch path determination device according to claim 1, wherein the branch path determination unit calculates the probability lower as the curvature calculated by the curvature calculation unit is larger.   The branch path according to any one of claims 1 to 4, wherein the branch path determination unit calculates the probability to a predetermined minimum probability when the curvature calculated by the curvature calculation unit is larger than the predetermined value. Judgment device.   The branching path determination unit calculates the probability to a predetermined maximum probability when the curvature calculated by the curvature calculation unit is smaller than a predetermined minimum curvature set smaller than the predetermined value. The branch path determination device according to any one of 1 to 5.   The branch path determination device according to any one of claims 1 to 6, wherein the curvature is calculated from a yaw rate detected by a yaw rate sensor (12a) mounted on the vehicle. The plurality of features includes the display of branch information by road surface display acquired from the image information,
The branch according to any one of claims 1 to 7, wherein the branch road determination unit calculates a probability that the running road is a branch road lower than when there is a display when the branch information is not displayed by the road surface display. Road determination device. A plurality of the features include the display of branch information by a road sign obtained from the image information;
The branch according to any one of claims 1 to 8, wherein the branch path determination unit calculates a probability that the runway is a branch path lower than when there is a display when there is no display of branch information by the road sign. Road determination device. The branch path judging device according to any one of claims 1 to 9,
Of the candidate lines extracted by the boundary line candidate extraction unit, the candidate line recognized as the boundary line of the main road from candidate lines excluding the boundary line of the road determined to be a branch road by the branch road determination device Boundary line recognition device to select.