JP2018181093A - Mark line recognition device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide technology capable of accurately recognizing a pair of right and left mark lines for regulating boundaries of a driving lane in a provisional section in service.SOLUTION: A mark line recognition device which is loaded on a vehicle is provided with: a line extraction part; a boundary candidate line selection part; a provisional service determination part; a learning part; and a correction processing part. The line extraction part extracts lines on a road surface from an image photographed by a photographing part. The boundary candidate line selection part selects a pair of right and left boundary candidate lines which become candidates of a pair of right and left boundary mark lines from among the extracted lines. The provisional service determination part determines whether or not the vehicle is traveling on the provisional section in service. The learning part learns lane width when the vehicle enters the provisional section in service. The correction processing part corrects a position of a right side boundary candidate line based on the lane width learned by the learning part while the vehicle travels on the provisional section in service.SELECTED DRAWING: Figure 2

Description

  The present disclosure relates to technology for recognizing lane markings based on an image captured from a vehicle.

Patent Document 1 describes a technique capable of estimating the left and right dividing lines with high accuracy even when double white lines are drawn on the left and right of the traveling lane.
Specifically, in the technology described in Patent Document 1, detection points are detected in each of the lane lines on the left and right of the traveling lane, and the detection points correspond to the widths of a plurality of types of lane lines on opposing lane lines The candidate points of the dividing lines are set in the respective positions. Then, in each of the left and right of the traveling lane, a virtual line is set from the curve approximation formula set based on the detection point and the candidate point, and the type of the virtual line is estimated, and the type of the estimated virtual line The position of the dividing line is determined from.

JP 2012-58983 A

  According to the technique described in Patent Document 1, when there are plural types of dividing lines on the left and right of the own lane, the relationship (for example, the width of the lane) is considered to trust those whose relationship does not change much. We estimate the parcel line based on it. This concept basically assumes that the on-vehicle camera always detects the lane markings normally. Therefore, with the technique described in Patent Document 1, it is difficult to accurately recognize the dividing line in a scene where the dividing line is not always detected properly by the on-vehicle camera.

  As a scene where a division line is not always detected normally by a vehicle-mounted camera, the scene which is drive | working a temporary service area is mentioned. In the temporary service section, there is a high possibility that a situation in which it is difficult to recognize may occur, particularly because the division lines on the central belt side may be rubbed at some places or a change in brightness may occur inside and outside the tunnel.

  The present disclosure provides a technique capable of accurately recognizing a pair of left and right demarcation lines defining a boundary of a traveling lane in a temporary service section.

  The lane marking recognition apparatus (20) of the present disclosure defines the boundary of the lane (51) in which the vehicle is traveling, from the image in front of the vehicle taken by the imaging unit (10) mounted on the vehicle (101). It is comprised so that a pair of left and right boundary division lines (53, 61) may be recognized.

The marking line recognition apparatus according to the present disclosure includes a line extraction unit (21 to 24), a candidate boundary line selection unit (26), a temporary service determination unit (27), a learning unit (29), and a correction processing unit (30). And.
The line extraction unit is configured to extract lines (53, 54, 61 to 64) extending along the traveling direction of the vehicle from the image captured by the imaging unit. The boundary candidate line selection unit is configured to select a pair of left and right boundary candidate lines that are candidates for the pair of left and right boundary division lines (53, 61, 64) from the lines extracted by the line extraction unit. .

  The temporary service determining unit is configured to determine whether the vehicle is traveling on the temporary service section (50). The learning unit is configured to learn the lane width when the temporary service determining unit determines that the vehicle is traveling on the temporary service section. The vehicle width is the distance between the pair of left and right boundary divisions, that is, the distance between the pair of left and right boundary divisions in the direction perpendicular to the traveling direction.

  The correction processing unit determines the position of the right boundary candidate line out of the pair of left and right boundary candidate lines selected by the boundary candidate line selection unit when the temporary service determining unit determines that the vehicle is traveling on the temporary service section. The correction is made based on the lane width learned by the learning unit.

  According to such a configuration, when the vehicle enters the temporary service section, the lane width is learned. Then, when the lane width is learned, thereafter, the position of the right boundary candidate line among the pair of left and right boundary candidate lines selected by the boundary candidate line selection unit is corrected based on the learned lane width. .

  If the right border division line is worn or in a state of being hard to recognize due to the surrounding environment, a line outside the actual border division line may be selected as the right border candidate line.

  On the other hand, according to the present disclosure, even if a line outside the actual boundary division line is selected as a boundary candidate line on the right side of the vehicle, the position of the boundary candidate line is based on the learning value of the lane width. It is corrected to the actual right border line position. As a result, the pair of left and right boundary candidate lines after the correction can be finally recognized with high accuracy as the pair of left and right boundary division lines.

Therefore, according to the configuration of the present disclosure, it is possible to accurately recognize the pair of left and right boundary division lines defining the boundary of the traveling lane in the temporary service section.
In addition, the reference numerals in parentheses described in this column and the claims indicate the correspondence with the specific means described in the embodiment described later as one aspect, and the technical scope of the present disclosure It is not limited.

It is an explanatory view showing composition of vehicles of an embodiment. It is a block diagram showing a parting line recognition device of an embodiment. It is an explanatory view showing an example of a temporary service section of an embodiment. It is an explanatory view showing an example in the state where it is unclear, and the state where it is indistinct in the right boundary division line in the temporary service section of an embodiment. It is a flow chart of division line recognition processing of an embodiment. It is an explanatory view for explaining a procedure of edge point extraction processing of an embodiment. It is explanatory drawing which shows the specific case where the right side boundary candidate line is not recognized normally. It is an explanatory view for explaining the contents of correction processing of an embodiment. It is an explanatory view for explaining the contents of correction processing of an embodiment.

Hereinafter, embodiments of the present disclosure will be described with reference to the drawings.
(1) Configuration of Vehicle As shown in FIG. 1A, a vehicle 101 of the present embodiment includes an on-vehicle camera 10, a lane marking recognition device 15, and a vehicle control device 20.

  As shown in FIGS. 1A and 1B, the on-vehicle camera 10 is mounted in the vicinity of the top of the windshield 103 in the room of the vehicle 101. As shown in FIGS. 1A and 1B, the on-vehicle camera 10 captures an image of the front of the vehicle 101 and generates image data of the captured image of the front. The image taken by the on-vehicle camera 10 includes the road surface in front of the vehicle 101.

  The on-vehicle camera 10 repeatedly captures images at preset time intervals (for example, at intervals of 1/15 seconds). The on-vehicle camera 10 can capture a color image. Image data of an image captured by the on-vehicle camera 10 is output to the lane line recognition device 15.

  The lane line recognition device 15 recognizes the lane line on the road, estimates the shape of the traveling road based on the recognized lane line, and outputs the estimation result to the vehicle control device 20. The dividing lines are line paints such as white and yellow drawn on the road mainly for dividing the lanes.

  The lane line recognition apparatus 15 includes a known microcomputer provided with a CPU, a ROM, a RAM, and a semiconductor memory such as a flash memory. The lane line recognition device 15 is configured such that each function is realized by the CPU executing a program stored in the non-transitional tangible storage medium. In the present embodiment, the semiconductor memory corresponds to a non-transitional tangible storage medium storing a program.

  The number of microcomputers provided in the lane line recognition apparatus 15 may be one or more. Further, the various functions realized by the marking line recognition device 15 are not limited to those realized by the CPU executing a program, and some or all of them are realized using one or more hardware. You may

  The vehicle control device 20 can execute various functions for supporting the traveling of the vehicle 101, such as an alarm output function and a lane keeping function, based on the information such as the estimation result of the road shape input from the marking line recognition device 15. It is. The alarm output function is a function of monitoring the relative positional relationship of the vehicle 101 in the left and right direction in the traveling lane, and outputting an alarm to the driver when the vehicle 101 is about to deviate from the lane. The lane keeping function is a function that assists the steering operation and the braking operation so that the vehicle 101 can appropriately travel in the lane.

(2) Configuration of lane marking recognition device As shown in FIG. 2, the lane marking recognition device 15 includes an edge point extraction unit 21, a marker extraction unit 22, a line candidate calculation unit 23, a line feature extraction unit 24, and a multiple line determination unit 25, candidate candidate line selection unit 26, temporary service determination unit 27, state determination unit 28, lane width learning unit 29, correction processing unit 30, alternative line search unit 31, road shape estimation unit 32, and recognition result output unit 33 Prepare.

  In the present embodiment, the functions of the units 21 to 33 included in the lane marking recognition apparatus 15 shown in FIG. 2 are all realized by the CPU executing a program. That is, FIG. 2 illustrates the configuration of the dividing line recognition device 15 as a functional block for each of various functions realized by the CPU executing a program. The specific contents of the functions implemented by the above-described units 21 to 33 included in the lane marking recognition apparatus 15 will be described later with reference to FIG.

  As shown in FIG. 2, a sensor unit 11 is provided in the vehicle 101. The sensor unit 11 is a generic name of various sensors that measure various state quantities of the vehicle 101. The sensor unit 11 includes, for example, various sensors such as a vehicle speed sensor that measures the vehicle speed of the vehicle 101, and a yaw rate sensor that measures the yaw rate of the vehicle 101. The sensor unit 11 repeatedly measures the state quantity of the measurement target at preset time intervals, and outputs the measurement result to the lane line recognition apparatus 15.

The dividing line recognition device 15 recognizes a pair of left and right boundary dividing lines that define the boundary of the lane on which the vehicle is traveling based on the image captured by the on-vehicle camera 10.
Here, the temporary service section is included in the road on which the vehicle 101 may travel. The temporary service section is a known section called a temporary common use section or a temporary two lanes, and basically, a total of two lanes, one lane on each side, are arranged to face each other across the central zone. Further, in the temporary service section, one solid dividing line is drawn on the left side in the traveling direction, and at least a yellow dividing line is drawn on the central band on the right side in the traveling direction. In the central zone, white lane lines are drawn separately for each lane, and lane separation markers are installed between the yellow lane lines on the own lane side and the yellow lane lines on the opposite lane side. , Curbs may be installed.

  An example of the temporary service section is shown in FIG. FIG. 3 is an image of the temporary in-service section 50 taken by the on-vehicle camera 10 mounted on the vehicle 101 traveling in the temporary in-service section 50. The temporary service section 50 shown in FIG. 3 is divided into two lanes 51 and 52, which are the lane in which the vehicle 101 is traveling, the on-coming lane 51 on which the oncoming vehicle travels, and the on-coming lane 52. It comprises outer lines 53, 54 and a central band 55.

  The roadway outside line 53 drawn on the left side of the own lane 51 is a left side demarcation line among a pair of left and right demarcation lines defining the boundary of the own lane 51, and hereinafter also referred to as the left side demarcation line 53. A side wall 71 is provided at a further left end of the roadway outside line 53 on the left side as viewed from the host vehicle 101. Also on the opposite lane 52 side, the roadway outside line 54 drawn on the left side of the opposite lane 52 is the left boundary division line among the pair of left and right boundary division lines defining the boundary of the opposite lane 52. A side wall 72 is provided at a further left end of the roadway outside line 54 on the left side of the vehicle traveling on the opposite lane 52.

  The center zone 55 provided in the temporary service zone 50 has a multiple line structure in which a plurality of dividing lines are drawn. Specifically, the central zone 55 has a yellow center line 61 drawn at a position closest to the own lane 51, a white center line 62 drawn parallel to and adjacent to the yellow center line 61, and the opposite lane 52. And a white center line 64 drawn parallel to and adjacent to the yellow center line 63. Also, in the central portion of the central zone 55 in the lane direction (that is, the traveling direction of the vehicle 101), that is, between the white center lines 62 and 64, the curb 65 and the lane separation mark 66 are discrete along the lane direction. Is located in

  The yellow center line 61 drawn at a position closest to the own lane 51 among the demarcation lines 61 to 64 included in the central zone 55 is the right side of the pair of left and right demarcation lines defining the boundaries of the own lane 51 It is a dividing line, and hereinafter, it is also referred to as a right boundary dividing line 61.

  In the temporary service section 50, among the division lines on both sides of the lane, particularly the division lines on the central zone side are more likely to be worn or soiled, and it is likely to be difficult to recognize. Further, the temporary service section 50 has a relatively high possibility of being provided with a tunnel as compared with other general roads. Within the tunnel, there is a high possibility that it becomes relatively difficult to recognize the dividing line due to various factors such as the difference in light and dark, as compared with the outside of the tunnel.

As described above, during the temporary service zone 50, there is a high possibility that the boundary division line on the right side among the pair of boundary division lines on both sides of the lane is difficult to recognize as compared with other general roads. .
FIG. 4 shows an example in which an unclear area 70 is present in the central zone 55 in the middle of the temporary service zone 50. The unclear area 70 is an area in which both the yellow center line 61 and the white center line 62 on the side of the own lane 51 in the center zone 55 are rubbed or soiled and are not recognized normally by the marking line recognition device 15 It is.

  In the example of FIG. 4, when the boundary division lines 53 and 61 on both left and right sides of the own lane 51 are both clear, the vehicle control device 20 has a normal lane width based on the information from the division line recognition device 15 It can be recognized that the value is W0, and the vehicle 101 can be controlled to travel substantially at the center of the normal lane width W0.

  On the other hand, when the vehicle 101 enters a section in which the unclear area 70 exists on the central band 55 side, the marking line recognition device 15 temporarily sets the white central line 64 on the opposite lane 52 side to the own lane 51. If it is misidentified as the right side boundary division line 61 of the side, the vehicle control device 20 misrecognizes the width Wx from the left side boundary division line 53 to the white center line 64 on the opposite lane 52 side as the lane width of the own lane 51 there's a possibility that. Thus, when the lane width is erroneously recognized as Wx larger than the actual width W0, the vehicle 101 is controlled to travel substantially at the center of the erroneously recognized lane width Wx, and as a result, the center There is a possibility of approaching the belt 55 side.

  Therefore, while the vehicle 101 is traveling on the temporary in-service section 50, the lane marking recognition apparatus 15 of the present embodiment has a pair of left and right boundary divisions in a method different from the case of traveling on a road other than the temporary in-service section. It is configured to recognize lines.

(3) Sectioning line recognition process Next, the sectioning line recognition process performed by the sectioning line recognition apparatus 15 will be described with reference to FIG. The lane marking recognition apparatus 15 periodically and repeatedly executes the lane marking recognition processing of FIG. 5 while the vehicle 101 is traveling.

  When the marking recognition process of FIG. 5 is started, the marking recognition apparatus 15 executes edge point extraction processing in S110. The edge point extraction process is a process implemented by the edge point extraction unit 21 illustrated in FIG. 2 and is a process of extracting an edge point from a captured image. Specifically, in S110, the edge point extraction unit 21 acquires an image captured by the on-vehicle camera 10, and extracts an edge point that is a pixel with a large change in luminance value from the acquired image.

  As a specific method of extracting edge points from a photographed image is well known, the detailed description is omitted here and only the outline is briefly described. While scanning from the left end to the right end of the image in the horizontal direction, the edge point extraction unit 21 searches for pixels for which the amount of change in luminance value is equal to or greater than a threshold, and extracts up edge points and down edge points. The up edge point is a rising point of the luminance value that changes from a low luminance value to a high luminance value, and more specifically, a point where the amount of change in luminance value from a low luminance value to a high luminance value is equal to or greater than the luminance threshold. It is. The down edge point is a falling point of the luminance value changing from a high luminance value to a low luminance value, more specifically, the amount of change in luminance value from a high luminance value to a low luminance value is equal to or higher than the luminance threshold It is a point. The edge point extraction unit 21 repeats the extraction of the edge points by such scanning in the horizontal direction while shifting the position in the vertical direction of the image, that is, the depth direction of the image, so that edge points can be extracted from almost the entire area of the image. Extract.

  The extracted edge point 81 is schematically shown in FIG. FIG. 6 shows an example in which edge points 81 are extracted on both sides of the left boundary division line 53 and on both sides of the right boundary division line 61. In FIG. 6, the distinction as to whether the edge point 81 is an up edge point or a down edge point is omitted. Moreover, although illustration is abbreviate | omitted in FIG. 6, an edge point is similarly extracted about division lines other than these each boundary division lines 53 and 61. FIG. Furthermore, when there is a point where the amount of change in luminance value is equal to or greater than the luminance threshold, not only the dividing line, that point is also extracted as an edge point.

  In S120, a marker extraction process is performed. This marker extraction process is a process implemented by the marker extraction unit 22 shown in FIG. 2 and is a process of extracting markers that are lumps corresponding to division lines by connecting the edge points extracted in S110.

  Specifically, in S120, the marker extraction unit 22 extracts a substantially rectangular area formed by a plurality of edge points as a marker from each edge point extracted in S110. Specifically, the marker extraction unit 22 pairs an up edge point with a down edge point located on the right side of the up edge point and closest to the up edge point. Then, a substantially rectangular area in which the pair is arranged at an interval within a preset interval threshold in the traveling direction of the vehicle 101 is extracted as a marker.

  In the central zone 55, since the yellow center line 61 on the side of the own lane 51 and the white center line 62 are adjacent to each other, edge points extracted at both ends of the yellow center line 61 on the side of the own lane 51 are also up. It becomes an edge point. Further, in the central zone 55, the white center line 64 on the opposite lane 52 side and the yellow center line 63 are also adjacent, so the edge points extracted at both ends of the yellow center line 63 on the opposite lane 52 side are all down edge It becomes a point.

  That is, where the yellow center line and the white center line are adjacent to each other, edge points of the same type are continuously extracted at both ends of the yellow center line. Therefore, in the present embodiment, when the same edge point is continuously extracted, a color component between the two continuously extracted edge points is detected, and when the yellow component is at a predetermined level or more, Pair two edge points.

  For example, when three edge points of an up edge point, an up edge point, and a down edge point are extracted sequentially from the left, if there is a yellow component of a certain level or more between the two up edge points, the two up Edge points are paired. Furthermore, the second up edge point and the third down edge point are paired.

  The interval threshold is a value shorter than the interval between line segments of a general division line, and edge points having an interval smaller than the interval threshold can be regarded as belonging to the same line segment. In S120, the marker extraction unit 22 calculates the length and width of each extracted marker. The length of the marker is the length in the traveling direction of the vehicle 101, and the width of the marker is the horizontal length perpendicular to the traveling direction, that is, the distance between the up edge point and the down edge point in the lane width direction. .

  In S130, line candidate calculation processing is executed. This line candidate calculation process is a process implemented by the line candidate calculation unit 23 shown in FIG. 2 and extends along the traveling direction of the vehicle 101 by connecting the markers extracted in S120 in the traveling direction. It is a process of extracting existing lines.

  According to this line candidate calculation process, when the vehicle 101 is traveling on the temporary service zone 50 as shown in FIG. 3, for example, six line candidates corresponding to the six division lines 53, 54, 61 to 64. And at least two line candidates corresponding to the side walls 71 and 72 are extracted. However, when the vehicle 101 travels, for example, a section in which the unclear area 70 shown in FIG. 4 is present, the central zone 55 corresponds to the two dividing lines 61 and 62 on the own lane 51 side. Line candidates are not extracted, and two line candidates corresponding to the two demarcation lines 63 and 64 on the opposite lane 52 side are extracted.

  In S140, line feature extraction processing is performed. This line feature extraction process is a process implemented by the line feature extraction unit 24 shown in FIG. 2, and for each line candidate calculated in S130, various information indicating the properties of the line candidate such as line type and color are It is a process to extract.

  The line feature extraction unit 24 determines the line type based on the distribution on the road surface of the edge point belonging to the line candidate. As the line type, there are, for example, solid dividing lines, broken dividing lines, and broken auxiliary lines drawn inside the dividing lines. Usually, since the lengths of the line segments and the intervals between the line segments are different between the dashed dividing line and the broken auxiliary line, they can be discriminated from the distribution on the road surface of the edge points.

Further, the line feature extraction unit 24 can distinguish and determine at least two or more colors of yellow and white.
The line feature extraction unit 24 also determines the number of edge points, the strength of edge points, the line length, and the like for each line candidate in the line feature extraction process of S140. The number of edge points is the number of edge points possessed by one line candidate, and when the line candidate comprises a plurality of markers such as broken lines, it is the total number of edge points possessed by the plurality of markers. Edge point intensity is the brightness difference of each edge point. The line length is the length in the traveling direction of one line candidate, and when the line candidate consists of a plurality of markers such as a broken line, for example, one line of the plurality of markers connected in the traveling direction It is a length.

  In S150, a multiline determination process is performed. The multiple line determination process is a process implemented by the multiple line determination unit 25 shown in FIG. In S150, specifically, the multiple line determination unit 25 determines whether or not each line candidate calculated in S130 constitutes a multiple line. When a plurality of line candidates are calculated within a range preset in the horizontal direction, the multiple line determination unit 25 determines that the multiple line is configured by the plurality of line candidates.

  By this multiple line determination processing, for example, while the vehicle 101 is traveling on the temporary service zone 50 as shown in FIG. 3, six line candidates corresponding to the six dividing lines 53, 54, 61 to 64 are calculated. For example, the following determination is made. That is, among the calculated six line candidates, it is determined that four line candidates corresponding to the four division lines 61 to 64 in the central band 55 constitute multiple lines, and the other two lines It is determined that each of the two line candidates corresponding to the division lines 53 and 54 in the first group constitutes a single line.

  However, when the vehicle 101 travels, for example, a section in which the unclear area 70 shown in FIG. 4 is present, the central zone 55 corresponds to the two dividing lines 63 and 64 on the opposite lane 52 side. It is determined that a multiple line is configured by two line candidates.

  In S160, a boundary candidate line selection process is performed. The boundary candidate line selection process is a process implemented by the boundary candidate line selection unit 26 shown in FIG. In S160, specifically, the boundary candidate line selection unit 26 is a candidate for a pair of left and right boundary division lines 53 and 61 that define the boundary of the own lane 51 in travel among the line candidates calculated in S130. Select a pair of border candidate lines.

  When selecting the pair of left and right boundary candidate lines, the boundary candidate line selection unit 26 refers to the results of the line feature extraction process of S140 and the multiple line determination process of S150. That is, among the line candidates calculated in S130, the line candidates determined to be solid lines in the line feature extraction process of S140 are narrowed down. Then, in each of the left and right of the vehicle 101, among the line candidates determined as solid lines, the line candidate closest to the vehicle 101 is selected as a boundary candidate line.

  When, for example, the vehicle 101 is traveling in the temporary service zone 50 as shown in FIG. 3, on the left side of the vehicle 101, one line candidate is calculated and the one line candidate is a solid line, The one line candidate is selected as a left boundary candidate line which is a candidate for the left boundary partition line 53. On the other hand, on the right side of the vehicle 101, the determination result that there is a multiple line consisting of four line candidates, and the determination result that all four line candidates constituting the multiple line are solid lines The line candidate closest to the vehicle 101 among the four line candidates constituting the multiple line is selected as the right boundary candidate line which is a candidate for the right boundary division line 61 based on the above.

  However, when the vehicle 101 is traveling in a section in which, for example, the unclear area 70 shown in FIG. 4 is present, two line candidates constituting a multiple line are on the opposite lane 52 side on the right side of the vehicle 101 Since two line candidates corresponding to the demarcation lines 63 and 64 are selected, the line candidate closest to the vehicle 101 is selected as the right side boundary candidate line among the two line candidates. That is, the line candidate corresponding to the white center line 64 on the opposite lane 52 side is selected as the right side boundary candidate line.

  In S170, it is determined whether the road on which the vehicle 101 is traveling is a temporary service section. This determination process is a process implemented by the temporary availability determination unit 27 shown in FIG. In S170, the temporary availability determination unit 27 can use various methods to determine whether the road on which the vehicle 101 is traveling is a temporary service area.

  The temporary service determination unit 27 may determine, for example, whether or not the road being traveled is a temporary service section, using the results of the line feature extraction process of S140 and the multiple line determination process of S150. Specifically, for example, when there is a multiplex line on the right side of the vehicle 101 and the multiplex line includes a composite line consisting of a yellow line and a white line, it is determined to be a temporary service zone. May be In this case, for example, a white solid line exists as a line candidate on the left side of the vehicle 101, and a structure such as a lane separation mark or a curb is provided on the right side of the vehicle 101 along the line candidate. In addition, conditions such as the presence of the oncoming traffic lane on the right side of the traveling lane may be appropriately added to the determination.

  Also, for example, the temporary availability determination unit 27 may set conditions specific to the temporary service section such as no sharp curve, a number of intersections being equal to or less than a threshold value, or a number of signals being less than or equal to a threshold value within a certain traveling range. It may be determined whether or not it is a temporary in-use section depending on whether it is satisfied or not. Also, for example, based on information obtained by data communication with the outside of the vehicle 101, such as acquiring GPS information, it may be determined whether the vehicle 101 exists in the temporary service zone.

  In addition, when it is determined that the temporary determination section is determined to be a temporary in-service section by the above-described determination method as a temporary determination, the interval between the pair of left and right boundary candidate lines selected in S160 is further determined. It may be determined whether a candidate line width which is an interval in a direction perpendicular to the traveling direction of each boundary candidate line is within a predetermined width range. Then, when the candidate line width is within the predetermined width range, it may be formally determined to be a temporary service section.

  If it is determined in S170 that the road on which the vehicle 101 is traveling is not a temporary service section, the process proceeds to S240. If it is determined in S170 that the road on which the vehicle 101 is traveling is the temporary service section, the process proceeds to S180.

  When it is determined in S170 that the road on which the vehicle 101 is traveling is the temporary service section, the lane width learning unit 29 performs initial learning of the lane width. Specifically, the candidate line width of each currently selected boundary candidate line is stored in the memory as the initial value of the lane width learning value. The lane width learning value stored in the memory is updated to the value calculated in the lane width learning process each time the lane width learning process of S200 described later is executed.

  At S180, a state determination process is performed. This state determination process is a process implemented by the state determination unit 28 shown in FIG. 2 and is a process of determining in which state the pair of left and right border candidate lines selected in S160 are recognized. .

  Specifically, in S180, the state determination unit 28 performs at least the clearness determination and the parallelism determination. Clarity judgment is judgment of whether each boundary candidate line is recognized in a clear state. Parallelism judgment is judgment of whether each boundary candidate line is parallel to each other.

  For clarity determination, for each boundary candidate line, the number of edge points, edge point intensity, line length, whether it is a solid line or not, variation in edge point interval, maximum value of edge point intervals, estimation line Based on at least one of various parameters such as an edge point error, it is determined whether or not it is clearly recognized. The estimation line is a curve fitted to an edge point.

  For example, the number of edge points is equal to or greater than the threshold value, the ratio of edge points having a strength equal to or greater than the threshold value among all edge points is equal to or greater than the threshold value, the line length is equal to or greater than the threshold value, and The difference between the values is within the threshold value, the maximum value of the distance between the edge points is less than the threshold value, and the error between the estimated line and the edge point is less than the threshold value. If satisfied, it may be determined to be clear.

  In the temporary service zone, in general, the boundary dividing line on the left in the traveling direction is less likely to be hard to recognize as compared with the boundary dividing line on the right in the traveling direction. Therefore, in the present embodiment, the left boundary candidate line is described on the premise that it is clearly recognized.

  For parallelism judgment, for each boundary candidate line, calculate at least one of various parameters such as curvature, inclination with respect to traveling direction, degree of change from recognition shape in the past, density of edge points, etc. Based on the comparison result of at least one parameter, it is determined whether the right boundary candidate line is parallel to the left boundary candidate line. For example, the difference between the left and right curvatures at the same position in the running direction is less than or equal to the threshold, the difference between the left and the right with respect to the traveling direction at the same position in the running direction is less than the threshold, Left and right border candidate lines if at least one of the predetermined conditions is satisfied among the fact that the difference in density of points is less than or equal to the threshold and the degree of change from the recognition shape in the past is less than or equal to the threshold on both left and right May be determined to be parallel.

  Furthermore, the state determination unit 28 also determines whether or not the lane width learning process of S200 and the correction process of S230 described later may be performed based on the result of the clarity determination and the parallelism determination. Specifically, the state determination unit 28 determines that both of the left and right border candidate lines are determined to be clear in the clarity determination and that the left and right border candidate lines are determined to be parallel in the parallelism determination. It is determined that the vehicle width learning process may be performed. If it is determined that the left and right boundary candidate lines are parallel in the parallelism determination, it is determined that the correction process of S230 may be executed regardless of the determination result of the clarity determination.

  In addition to the clearness determination and the parallelism determination, the state determination unit 28 may perform, for example, an aptitude determination to be learned in S180. The learning object suitability judgment is a judgment as to whether or not the currently selected pair of left and right boundary candidate lines has a suitability as a target for the lane width learning processing in terms of clarity and parallelism. .

  There are various possible specific contents of the judgment of aptitude for learning. For example, it may be determined whether the candidate line width is within a predetermined width range. Then, when the candidate line width is within the predetermined width range, it may be determined that each of the boundary candidate lines is suitable for the lane width learning process. Further, for example, it may be determined whether the color of the right boundary candidate line is yellow among the boundary candidate lines. Then, when the color of the right side boundary candidate line is yellow, it may be determined that each of the boundary candidate lines has the suitability as a target of the lane width learning process.

  When the state determination unit 28 is configured to perform the learning object suitability judgment, the judgment as to whether or not the lane width learning processing may be performed may be determined based on the clarity judgment, the parallelism judgment, and the learning object suitability judgment. It may be performed based on the result. Specifically, the state determination unit 28 determines that both the left and right border candidate lines are clear in the clarity determination, and determines that the left and right border candidate lines are parallel in the parallelism determination, and in the learning object suitability determination It may be determined that the lane width learning process may be performed when it is determined that each of the boundary candidate lines is suitable for the lane width learning process.

  In S190, it is determined whether the lane width can be learned. Specifically, it is determined whether it is determined that the lane width learning process may be performed in the state determination process of S180. If it is not determined that the lane width learning process may be performed in the state determination process of S180, it is determined that the lane width can not be learned, and the process proceeds to S210. If it is determined that the lane width learning process may be performed in the state determination process of S180, it is determined that the lane width can be learned, and the process proceeds to S200.

  In S200, a lane width learning process is performed. This lane width learning process is a process implemented by the lane width learning unit 29 shown in FIG. Specifically, in S200, the lane width learning unit 29 calculates the distance between the pair of left and right boundary candidate lines selected in S160, that is, the distance in the direction perpendicular to the traveling direction. Then, the calculated interval is learned as the lane width of the temporary service section during travel. Specifically, the calculated interval is stored in the memory as a lane width learning value. When the lane width learning value is already stored, the stored lane width learning value is updated to the lane width learning value calculated at S200 this time.

  In S210, it is determined whether the correction based on the lane width learning value stored in the memory is necessary for the right boundary candidate line among the pair of left and right boundary candidate lines selected in S160. Specifically, for example, if the difference between the candidate line widths of the pair of left and right boundary candidate lines selected in S160 and the lane width learning value is equal to or larger than a predetermined width threshold, it may be determined that the correction is necessary. Good. Conversely, when the difference between the candidate line width and the lane width learning value is smaller than the width threshold, it may be determined that the correction is unnecessary. Also, if it is determined that the left and right boundary candidate lines are not parallel in the parallelism determination by the state determination unit 28, it may be determined that the correction is necessary.

  Although it is possible that the line candidate is not calculated at all on the central band side and the right boundary candidate line is not selected, it is also determined that the correction is necessary in S210.

  In S220, it is determined whether correction based on the lane width learning value may actually be performed on the right boundary candidate line among the pair of left and right boundary candidate lines selected in S160. Specifically, it is determined whether or not it is determined in the state determination process of S180 that the correction process may be performed. If it is not determined that the correction process may be performed in the state determination process of S180, it is determined that the correction process for the currently selected right boundary candidate line should not be performed, and the process proceeds to S260.

  As an example in which the right side boundary candidate line is not judged to be parallel to the left side boundary candidate line, for example, an example as shown in FIG. 7 can be considered. In FIG. 7, the left boundary candidate line 86 is normally extracted along the corresponding left boundary division line 53, while the right boundary candidate line 87 is separated from the corresponding right boundary division line 61 and the lane separation mark is separated. It shows an example extracted as a line connecting to 66. That is, FIG. 7 shows an example in which the left and right boundary candidate lines 86 and 87 are not parallel.

  In such a case, even if the correction process is performed on the right boundary candidate line 87, the corrected right boundary candidate line is corrected to a line shifted from the actual right boundary division line 61. Therefore, in the present embodiment, even if it is determined that the correction is necessary in S210, if the right boundary candidate line to be corrected is not parallel to the left boundary candidate line as in the example of FIG. It is configured not to perform processing.

  When the line candidate is not calculated at all on the central band side and the right side boundary candidate line is not selected, it is determined in S220 that the correction process should not be performed, and the process proceeds to S260.

  On the other hand, when it is determined that the correction process may be performed in the state determination process of S180, it is determined that the correction process may be performed on the currently selected right boundary candidate line in the determination process of S220. , S230.

  In S230, correction processing is performed on the right boundary candidate line. This correction process is a process realized by the correction processing unit 30 shown in FIG. More specifically, in S230, the correction processing unit 30 offsets the position of the right side boundary candidate line, that is, the position in the lane width direction, according to the difference between the candidate line width and the lane width learning value.

  For example, as illustrated in FIG. 8, while the lane width learning value is W0, the candidate line width is set by the currently selected right boundary candidate line being the white center line 64 on the opposite lane 52 side. It is assumed that Wx1 is longer than the vehicle width learning value W0. In this case, in the correction process of S230, the right boundary candidate line is offset to the left by the difference between the candidate line width Wx1 and the lane width learning value W0. As a result, the right side boundary candidate line is corrected to the position indicated by the two-dot chain line in FIG. 8, and the candidate line width Wc after correction has the same value as the lane width learning value W0.

  If it is determined in S220 that the correction should not be performed, an alternative line search process is executed in S260. This alternative line search process is a process implemented by the alternative line search unit 31 shown in FIG. In S260, among the line candidates calculated in S130, the alternative line search unit 31 is a line existing on the right side of the vehicle 101, among other line candidates different from the currently selected right boundary candidate line. From the above, alternative lines which are line candidates that can be targets of correction processing are searched.

  More specifically, a line candidate existing on the left side in the running direction of the oncoming lane or a line candidate corresponding to a structure provided along the running direction on the left side in the running direction on the oncoming lane is used as a substitute line. Explore. In the temporary service zone 50 illustrated in FIG. 3, a line candidate corresponding to the roadway outside line 54 on the opposite lane 52 side and a line candidate corresponding to the side wall 72 on the opposite lane 52 side may be alternative lines.

  The alternative line search unit 31 determines, as an alternative line, a line candidate that satisfies a predetermined condition among the line candidates that can be alternative lines. The predetermined condition may be determined as appropriate, and for example, it may be determined to be parallel to the left boundary candidate line by performing the same determination as the above-described parallelism determination. Also, for example, it may be determined to be clear by performing the same determination as the above-described clearness determination.

  If there are a plurality of line candidates satisfying a predetermined condition, which line candidate is to be determined as the alternative line may be appropriately determined. For example, among the plurality of line candidates, the line candidate closest to the own lane 51 may be determined as the alternative line. Alternatively, one of the plurality of line candidates having the highest clarity may be determined as the alternative line. Alternatively, among the plurality of line candidates, one having the highest parallelism with the left boundary candidate line may be determined as the alternative line.

  In S270, as a result of the alternative line search processing in S260, it is determined whether or not an alternative line has been searched, that is, whether any line candidate has been determined as an alternative line. If no alternative line is found, the process proceeds to S240. In this case, the right boundary candidate line selected in S160 is invalidated.

  If an alternative line is searched in S270, the process proceeds to S230, and correction processing is performed on the searched alternative line. That is, the substitute line is corrected to the position corresponding to the right boundary division line 61 by offsetting the substitute line to the left.

  For example, as illustrated in FIG. 9, it is assumed that the vehicle 101 travels in a section where an unclear area 80 in which no dividing line is recognized is present on the central band 55 side. Then, it is assumed that a line candidate corresponding to the roadway outside line 54 on the opposite lane 52 side is determined as the alternative line by the alternative line search processing of S260.

  In this case, in the correction process of S230, the distance Wx2 between the left boundary candidate line and the alternative line is set as a candidate line width, and the position of the alternative line is offset to the left by the difference between the candidate line width Wx2 and the lane width learning value W0. . Thus, the substitute line is corrected to the position indicated by the two-dot chain line in FIG. 9, and the substitute line after the correction becomes the right boundary candidate line. Further, the candidate line width Wc after correction is equal to the lane width learning value W0.

  In S240, a road shape estimation process is performed. In S240, the pair of left and right boundary candidate lines currently selected is treated as the recognition result of the pair of left and right boundary division lines 53 and 61. That is, as the recognition result of the pair of left and right boundary division lines 53 and 61, the currently selected pair of left and right boundary candidate lines is adopted, and the process of S240 is performed based on the currently selected left and right pair of boundary candidate lines. Run.

  The road shape estimation process of S240 is a process implemented by the road shape estimation unit 32 shown in FIG. In S240, the road shape estimation unit 32 estimates the shape of the currently traveling road including the own lane 51 based on the pair of left and right border candidate lines currently selected. When the correction process is performed in S230, for the right boundary candidate line, the shape of the road is estimated based on the right boundary candidate line after the correction process (that is, after the offset).

  Specifically, the road shape estimation unit 32 estimates, for example, various road parameters such as lane position, lane inclination, lane curvature, and lane width as road parameters indicating the characteristics of the road currently being traveled.

  Here, the lane width is as described above, which is the distance between the pair of left and right boundary division lines. The lane position is a center position on the width of the lane with respect to the center of the vehicle 101, and can be represented by the distance from the center of the vehicle 101 to the center position. For example, when the center of the vehicle 101 is positioned on the left side of the center position, the position of each boundary division line with respect to the vehicle 101 can be specified by setting it to a negative value and setting it to a positive value. The lane inclination is the inclination of the tangent at the center position of the virtual lane center passing through the center of the left and right boundary division lines with respect to the vehicle traveling direction, for example, a positive value when inclined to the left, and an inclination to the right For example, the angle can be specified by setting it to a negative value. The lane curvature is the curvature at the center of the lane.

  If the right boundary candidate line is invalidated because the alternative line is not searched in S260, the road shape estimation process is executed only with the left boundary candidate line in S240. At this time, the lane width uses information calculated up to the previous frame.

  In S250, recognition result output processing is performed. This recognition result output process is a process implemented by the recognition result output unit 33 shown in FIG. In S250, the recognition result output unit 33 outputs the road shape estimated in S240, that is, various road parameters, to the vehicle control device 20.

(4) Effects of the Embodiment According to the embodiment described above, the following effects (1a) to (1) can be obtained.
(1a) In the present embodiment, when the vehicle 101 enters the temporary service zone 50, the lane width is learned. Then, when the lane width is learned, thereafter, the position of the right side boundary candidate line is corrected based on the lane width learning value. Thereby, even if a line outside the actual right side boundary division line is selected as the right side boundary candidate line on the right side of the vehicle, the position of the right side boundary candidate line is actually based on the lane width learning value. It is corrected to the position corresponding to the dividing line. As a result, the pair of left and right boundary candidate lines after the correction can be finally recognized with high accuracy as the pair of left and right boundary division lines.

  (1b) In the present embodiment, when the vehicle 101 enters the temporary service zone 50, the state determination unit 28 determines whether or not the lane width learning process may be performed. Then, when it is determined that the lane width learning process may be performed, learning of the lane width is performed. Therefore, in the lane width learning process, the lane width can be learned accurately.

  (1c) Further, the state determination unit 28 determines whether or not the lane width learning process may be performed based on at least two determination results of the clear line determination and the parallel line determination. Specifically, it is determined that the lane width may be learned when it is determined to be clear by the clearness determination and to be parallel by the parallelism determination. Therefore, the accuracy of the lane width learning value can be further improved.

  (1d) In the present embodiment, when the vehicle 101 enters the temporary service zone 50 and lane width learning is performed, the right boundary candidate line is corrected based on the lane width learning value. However, the correction process is not always performed, and the correction process is performed when it is determined by the state determination unit 28 that the correction process may be performed. Therefore, when the right boundary candidate line is in an inappropriate state as the correction target, the correction of the right boundary candidate line can be suppressed.

  (1e) Further, the state determination unit 28 determines whether or not the correction process may be performed based on the result of the parallelism determination. Specifically, when it is determined by the parallelism determination that the right boundary candidate line is parallel to the left boundary candidate line, it is determined that the correction processing may be performed on the right boundary candidate line. Therefore, the correction process can be appropriately performed.

  (1f) In the present embodiment, when it is determined that the correction process should not be performed on the selected right boundary candidate line, an alternative line is searched. Then, when the alternative line is searched, the correction process is performed on the alternative line to appropriately acquire the right boundary candidate line. Therefore, even if the right side boundary candidate line is not selected from the center zone, the right side boundary division line can be appropriately recognized based on the alternative line existing on the oncoming lane side.

(5) Correspondence Relationship with the Terms of the Claims Here, the correspondence relationship between the terms of the present embodiment and the terms of the claims will be described.
The on-vehicle camera 10 corresponds to an example of a photographing unit. The side wall 72 corresponds to an example of a structure. The edge point extraction unit 21, the marker extraction unit 22, the line candidate calculation unit 23, and the line feature extraction unit 24 correspond to an example of a line extraction unit. The state determination unit 28 corresponds to an example of a learning availability determination unit, a clearness determination unit, a parallelism determination unit, and a correction availability determination unit. The vehicle width learning unit 29 corresponds to an example of a learning unit. The alternative line search unit 31 corresponds to an example of a search unit.

  Moreover, in FIG. 5, S110 to S140 correspond to an example of processing of the line extraction unit. S160 corresponds to an example of processing of the boundary candidate line selection unit. S170 corresponds to an example of the process of the temporary availability determination unit. S180 corresponds to an example of processing of the learning availability determination unit, the clearness determination unit, the parallelism determination unit, and the correction availability determination unit. S200 corresponds to an example of processing of the learning unit. S230 corresponds to an example of processing of the correction processing unit. S260 corresponds to an example of processing of the search unit. S210 corresponds to an example of processing of the correction necessity determination unit.

(6) Other Embodiments Although the embodiments of the present disclosure have been described above, the present disclosure is not limited to the above-described embodiments, and can be variously modified and implemented.

  (6-1) In the above embodiment, whether or not the correction is necessary is determined in S210, and when it is determined that the correction is necessary, the correction process is performed through S220, but the determination process of S210 is It may be omitted. That is, the correction processing may be performed on the selected right boundary candidate line every time as long as the correction is possible regardless of the positional relationship with the actual right boundary division line. In this case, for example, when the candidate line width and the lane width learning value are the same value, the offset amount is 0 in the correction process, and the correction is not substantially performed.

  (6-2) The on-vehicle camera 10 according to the above-described embodiment is an example of an imaging unit for imaging the front of the vehicle 101. The configuration of the imaging unit may be determined as appropriate. Further, a plurality of imaging units may be provided, and various processing such as recognition of boundary division lines and estimation of a road shape may be performed based on images from each imaging unit.

  (6-3) The timing at which the learning of the lane width is performed may be determined as appropriate while the vehicle 101 is traveling in the temporary service zone. For example, learning of the lane width may be performed only once after the vehicle 101 enters the temporary service section. More specifically, after the vehicle 101 enters the temporary service section, once learning of the lane width is performed and the lane width learning value is stored in the memory, thereafter, until the vehicle 101 leaves the temporary service section The learning may not be performed, and the correction process may be performed using the lane width learning value stored in the memory.

  Also, for example, conditions for learning may be predetermined, and learning may be performed each time the above conditions are satisfied after the vehicle 101 enters the temporary service zone. The conditions include, for example, that the vehicle 101 has traveled a certain distance, and that the difference between the candidate line width and the currently stored lane width learning value is a certain value or more.

  (6-4) The plurality of functions of one component in the above embodiment may be implemented by a plurality of components, or one function of one component may be implemented by a plurality of components. It is also good. Also, a plurality of functions possessed by a plurality of components may be realized by one component, or one function realized by a plurality of components may be realized by one component. In addition, part of the configuration of the above embodiment may be omitted. In addition, at least a part of the configuration of the above-described embodiment may be added to or replaced with the configuration of the other above-described embodiment. In addition, all the aspects contained in the technical thought specified from the wording described in the claim are an embodiment of this indication.

  (6-5) In addition to the dividing line recognition apparatus described above, a system including the dividing line recognition apparatus as a component, a program for causing a computer to function as the dividing line recognition apparatus, a medium storing the program, and partial line recognition The present invention can also be realized in various forms such as methods.

  DESCRIPTION OF SYMBOLS 10 Vehicle camera 15 marking line recognition device 21 edge point extraction unit 22 marker extraction unit 23 line candidate calculation unit 24 line feature extraction unit 25 multiple line determination unit 26 boundary candidate Line selection unit 27 Provisional service determination unit 28 State determination unit 29 Lane width learning unit 30 Correction processing unit 31 Alternative line search unit 32 Road shape estimation unit 33 Recognition result output unit , 50: temporary service section, 51: own lane, 52: opposite lane, 53: roadway outer line (left boundary division line), 54: roadway outer line, 55: central zone, 61: yellow central line (right boundary division line 62, 64: white central line, 63: yellow central line, 65: curb, 66: lane separation mark, 70, 80: unclear area, 71, 72, side wall, 81: edge point, 86: left boundary candidate Line 87: right side boundary candidate line 101: vehicle 101: vehicle

Claims (8)

From the image in front of the vehicle taken by the imaging unit (10) mounted on the vehicle (101), a pair of left and right boundary division lines (53, 61) defining the boundary of the lane (51) in which the vehicle is traveling A lane marking recognition device (20) for recognizing
A line extraction unit (21 to 24) configured to extract lines (53, 54, 61 to 64) extending along the traveling direction of the vehicle from the image captured by the imaging unit; ,
Boundary candidate line selection configured to select a pair of left and right boundary candidate lines that are candidates for the pair of left and right boundary division lines (53, 61, 64) from the lines extracted by the line extraction unit Part (26),
A temporary service determining unit (27) configured to determine whether the vehicle is traveling on the temporary service zone (50);
And a learning unit (29) configured to learn a lane width which is an interval between the pair of left and right boundary division lines when the temporary service determining unit determines that the vehicle is traveling on the temporary service section. ,
When the temporary service determining unit determines that the vehicle is traveling on the temporary service section, the position of the right boundary candidate line among the pair of left and right boundary candidate lines selected by the boundary candidate line selecting unit is selected. A correction processing unit (30) configured to correct based on the lane width learned by the learning unit;
A lane marking recognition device (20). The lane marking recognition apparatus according to claim 1, wherein
Furthermore, based on the extraction state of the pair of left and right boundary candidate lines selected by the boundary candidate line selection unit by the line extraction unit, the distance between the selected pair of left and right boundary candidate lines is learned as the lane width. A learnability determination unit (28) configured to determine whether or not to
The learning unit is configured to learn an interval of the pair of left and right boundary candidate lines selected by the boundary candidate line selection unit as the lane width when it is determined that the learning may be performed by the learning possibility determination unit. Is configured to
Lane marking recognition device. The dividing line recognition apparatus according to claim 2, wherein
The line extraction unit is configured to detect an edge point existing in a direction perpendicular to the traveling direction of the vehicle from the image captured by the imaging unit, and extract the line based on the detected edge point. Yes,
The learning availability determination unit
For each of the pair of left and right boundary candidate lines selected by the boundary candidate line selection unit, at least one of the number of edge points, the length in the traveling direction, whether or not it is a solid line, and the intensity of the edge points A clearness judging unit (28) configured to judge whether or not extraction is performed in a clear state based on
For each of the pair of left and right boundary candidate lines selected by the boundary candidate line selection unit, based on at least one of the curvature, the inclination with respect to the traveling direction, the degree of change from the recognition shape in the past, and the density of the edge points And a parallelism determination unit (28) configured to determine whether or not both are parallel;
The lane line recognition apparatus according to claim 1, further comprising: when it is determined that the clarity determination unit is clear and the parallelism determination unit is determined to be parallel, the learning may be performed. The dividing line recognition apparatus according to any one of claims 1 to 3, wherein
Furthermore, based on the extraction condition by the line extraction unit of the right boundary candidate line among the pair of left and right boundary candidate lines selected by the boundary candidate line selection unit, the correction processing unit of the right boundary candidate line is A correction possibility determination unit (28) configured to determine whether the correction may be performed according to
The correction processing unit is configured to perform the correction on the boundary candidate line on the right side when it is determined by the correction possibility determination unit that the correction may be performed.
Lane marking recognition device. The dividing line recognition apparatus according to claim 4, wherein
The line extraction unit is configured to detect an edge point existing in a direction perpendicular to the traveling direction of the vehicle from the image captured by the imaging unit, and extract the line based on the detected edge point. Yes,
The correction possibility determination unit determines both of the pair of left and right boundary candidate lines based on at least one of curvature, inclination with respect to the traveling direction, change degree from a recognition shape in the past, and density of the edge point. Are determined to be parallel, and if it is determined to be parallel, it is determined that the correction may be performed,
Lane marking recognition device. The dividing line recognition apparatus according to claim 4 or 5, wherein
Furthermore, when the correction possibility determination unit determines that the pair of left and right boundary candidate lines are not parallel, a line existing on the right side of the vehicle among the lines extracted by the line extraction unit, An alternative line that is a line that can be the target of the correction by the correction processing unit instead of the right boundary candidate line among other lines different from the line extracted as the right boundary candidate line And a search unit (31) configured to search for
When the correction processing unit determines that the pair of left and right boundary candidate lines is not parallel by the correction possibility determination unit, the alternative line is searched when the search unit searches the alternative line. Are configured to perform the correction as the right boundary candidate line,
Lane marking recognition device. The dividing line recognition apparatus according to claim 6, wherein
The search unit is a line (54) present on the left in the traveling direction of the oncoming lane in the oncoming lane (52) adjacent to the lane in which the vehicle is traveling, and along the traveling direction on the left in the traveling direction of the oncoming lane Searching for any of the lines formed by the provided structures (72) as the alternative lines,
Lane marking recognition device. The lane marking recognition apparatus according to any one of claims 1 to 7, wherein
Furthermore, a correction necessity determination unit (judged as to whether or not the correction by the correction processing unit is necessary for the boundary candidate line on the right side among the pair of left and right boundary candidate lines selected by the boundary candidate line selection unit ( 30),
The correction processing unit is configured to perform the correction on the boundary candidate line on the right side when the correction necessity determination unit determines that the correction is necessary.
Lane marking recognition device.

Images