JP6674365B2 - Lane marking recognition device - Google Patents

Description

  The present disclosure relates to a lane marking recognition device that recognizes lane markings on a road from an image captured by a vehicle-mounted camera.

  2. Description of the Related Art There is known an apparatus that recognizes a traveling lane from a lane marking indicating a boundary of a lane of a road. In this type of apparatus, when it is determined that the road is a branch section, a lane marking on the branch lane side is excluded from recognition targets. For example, the vehicle control device described in Patent Literature 1 performs a process by Hough transform on an edge point extracted from an image to extract an edge line which is a linear component. Then, the vehicle control device determines whether or not there is a branch section based on whether or not the extracted edge lines are plural on one side of the vehicle and two of the extracted edge lines are V-shaped. If it is determined that there is a branch section, the edge line included in the V-shape is excluded from the recognition target.

JP 2011-198110 A

  By the way, it is known that the statistical processing for extracting a straight line component from an edge point has a high load, and that the accuracy of an image is lower in a distant region than in a nearby region. Therefore, it is not realistic to perform the extraction processing of the linear component over the entire area from the vicinity of the vehicle to the distant place. Usually, the extraction processing is performed only in the vicinity area.

  In addition, the dividing line that separates the main lane and the branch lane includes an auxiliary line drawn from the near side of the branch section as a dividing line in the branch section and a auxiliary line drawn from the branch section as a starting point (hereinafter, referred to as a starting point). Discontinuous parting line). In the case of a non-continuous lane marking, a lane marking on the branch lane is recognized as a lane marking the main lane until an edge line is extracted from the non-continuous lane markings detected in the neighboring area. In addition, in the extraction processing using the statistical processing, an edge line along the non-continuous dividing line cannot be accurately extracted unless the length of the non-continuous dividing line detected in the neighboring area becomes a certain length.

  Therefore, in the related art described in Patent Literature 1, extraction of a V-shape formed by an edge line along a discontinuous lane marking and an edge line along a lane marking on the branch lane side is a branch section. This delays the determination of whether or not the vehicle control is unstable in the branch section.

  The present disclosure has been made in view of the above circumstances, and provides a lane marking recognition device that quickly determines a branch lane with a low processing load and suppresses false recognition of lane markings in a branch lane. Main purpose.

  The present disclosure relates to a lane marking recognition device (30) for recognizing a lane marking for dividing a lane of a road from an image around a vehicle (70) captured by an on-vehicle camera (11), and includes a feature point extraction unit (33). ), A candidate line extraction unit (34), a line selection unit (37), a branch determination unit (38), and a recognition unit (39). The feature point extracting unit extracts a feature point representing a boundary of road surface paint in a near area and a far area of an image, with a range from a vehicle to a predetermined distance as a nearby area, and a far area farther than the nearby area as a far area. It is configured to be. The candidate line extraction unit extracts a straight line component from the feature points extracted in the neighboring region by using a statistical process, and also determines a continuity between the extracted straight line component and the straight line component of the feature points extracted in the distant region. And a feature line is extracted from the feature point having The line selection unit is configured to select a characteristic line to be recognized as a division line from the characteristic lines extracted by the candidate line extraction unit. The branch determination unit includes one paint block, which is a rectangular area formed by the feature points extracted by the feature point extraction unit, on the inner side closer to the vehicle than the feature line selected by the line selection unit. At least one of the length of the candidate block and the width of the candidate block in the traveling direction of the vehicle separates the main lane and the branch lane, with the paint block existing as described above and existing inside the lane marking as a candidate block. It is configured to determine that there is a branch section when it has a feature of a line segment that forms a broken line. The recognizing unit is configured to, when the branch determining unit determines that there is a branch section, exclude a feature line selected on the outer side that is farther from the vehicle than the candidate block from the recognition target.

  According to the present disclosure, feature points are extracted from a near region to a far region of an image. In addition, a straight line component is detected from the feature points in the nearby area, and a feature line is extracted from the detected straight line component and feature points in a distant area having high continuity with the straight line component. Then, a feature line recognized as a division line is selected from the extracted feature lines. Usually, the division line drawn starting from the branch section is a dashed line, and such a division line is extracted as a rectangular area paint block for each dashed line segment before being extracted as a feature line. You. The length and width of the dashed line segment that is the dividing line are different from the length of the dashed segment line that separates the main lanes. Therefore, when there is a candidate block that is a paint block inside the selected feature line and at least one of the length and the width of the candidate block has a feature of a dividing line, it is determined that there is a branch section. The feature line selected outside the candidate block is excluded from the recognition target. Therefore, even when the lane marking is present in a distant area and is not extracted as a characteristic line, it is possible to determine the branching section and suppress recognition of the lane marking on the branch lane side. That is, it is possible to quickly determine a branch section with a low processing load, and suppress erroneous recognition of a lane marking in the branch section.

  Note that the reference numerals in parentheses described in this column and in the claims indicate a correspondence relationship with specific means described in the embodiment described below as one aspect, and the technical scope of the present disclosure will be described. It is not limited.

It is a figure showing the mounting position of an in-vehicle camera. It is a block diagram showing composition of a lane marking recognition device. It is a figure showing the edge point extracted in a near field and a far field. It is a figure showing the edge line extracted based on the straight line component of the neighborhood. It is a flowchart which shows the processing procedure which determines a branch area. It is a figure which shows the road surface in which the auxiliary line is drawn inside the division line from just before a branch section. It is a figure showing the road surface where a branch line is drawn from a branch section.

Hereinafter, embodiments for carrying out the invention will be described with reference to the drawings.
<Structure>
First, a configuration of a driving support system 100 to which a lane marking recognition device 30 (hereinafter, a recognition device 30) according to the present embodiment is applied will be described with reference to FIGS. 1 and 2. The driving support system 100 according to the present embodiment includes an in-vehicle camera 10, sensors 11, a vehicle control device 50, and a recognition device 30. The in-vehicle camera 10, the sensors 11, and the vehicle control device 50 are connected to the recognition device 30 via an in-vehicle LAN. The recognition device 30 is a device that is mounted on the vehicle 70 and recognizes lane markings on a road. The division lines are white lines and yellow lines drawn so as to divide the lanes of the road. In the following, a white line including a division line of a color other than white is referred to for convenience.

  As shown in FIG. 1, the on-vehicle camera 10 is installed, for example, on a rearview mirror so that a predetermined range on a road surface in front of the vehicle is a shooting range. The in-vehicle camera 10 repeatedly shoots at a preset time interval, for example, at 1/15 (second) interval, converts the captured image into a digital signal, and outputs it to the recognition device 30.

  The sensors 11 are various sensors that measure the state quantity of the vehicle, and include a vehicle speed sensor that measures the vehicle speed of the vehicle 70 and a yaw rate sensor that measures the yaw rate of the vehicle 70. The sensors 11 repeatedly measure at predetermined time intervals, and output the measurement result to the recognition device 30.

The vehicle control device 50 is an ECU, and mainly includes a known microcomputer including a CPU, a ROM, a RAM, and a semiconductor memory such as a flash memory. The vehicle control device 50 executes the warning output control and the traveling control based on the recognition result of the recognition device 30. The warning output control is a control for outputting a warning when the vehicle 70 is likely to depart from the lane. The running control is steering control and brake control of the vehicle 70 that is executed so as to run in a lane. The recognition device 30 is an ECU, and has a semiconductor memory such as a CPU, a ROM, a RAM, and a flash memory. It is composed mainly of a well-known microcomputer provided. The recognition device 30 realizes each function as shown in FIG. 2 by the CPU executing the program stored in the non-transitional substantive recording medium. In the present embodiment, the semiconductor memory corresponds to a non-transitional substantial recording medium for storing a program. In addition, by executing this program, a method corresponding to the program is executed. The number of microcomputers constituting the recognition device 30 may be one or more.

  As shown in FIG. 2, the recognition device 30 includes a feature point extraction unit 33, a candidate line extraction unit 34, a line type determination unit 35, a multiple line determination unit 36, a line selection unit 37, a branch determination unit 38, and a recognition unit 39. It has the function of. The technique for realizing these functions is not limited to software, and some or all of the elements may be realized using hardware combining logical circuits, analog circuits, and the like.

  The feature point extracting unit 33 acquires an image from the vehicle-mounted camera 10. Then, the feature point extracting unit 33 extracts, from both the near area Ra and the far area Rb of the acquired image, an edge point that is a pixel having a large change in luminance value as a feature point representing a boundary of paint on the road surface. . The vicinity area Ra is a range from the vehicle 70 to a predetermined distance. Further, the far region Rb is a region adjacent to the near region Ra and farther from the vehicle 70 than the near region Ra, and is, for example, a region up to a vanishing point.

  More specifically, the feature point extraction unit 33 searches for a pixel in which the amount of change in the luminance value is equal to or larger than a threshold in the horizontal direction from the left end to the right end of the image, and extracts an up edge point and a down edge point. The up edge point is a rising point of a luminance value that changes from a low luminance value to a high luminance value. The down edge point is a falling point of a luminance value that changes from a high luminance value to a low luminance value. FIG. 3 schematically shows the extracted up-edge points by circles and the extracted down-edge points by triangles. As shown in FIG. 3, an up-edge point is extracted on the outline on the left side of the white line, and a down-edge point is extracted on the outline on the right side of the white line.

  Further, the feature point extraction unit 33 extracts a rectangular area formed by the edge points as a paint block from the edge points extracted in the near area Ra and the far area Rb. Specifically, the feature point extraction unit 33 paints an area surrounded by an up edge point on the left side and a down edge point on the right side, and the interval between edge points in the traveling direction of the vehicle 70 is within a preset interval threshold. Extract as a block. The interval threshold value is a value shorter than the interval between general white line segments, and is a value that can be regarded as belonging to the same line segment between edge points with intervals smaller than the interval threshold value. In FIG. 3, a paint block B1 corresponding to the solid white line on the left, a paint block B2 corresponding to the solid white line on the right, and paint blocks Ba, Bb, and Bc corresponding to the broken white line on the right are extracted as paint blocks. ing. Further, the feature point extraction unit 33 calculates the length and width of the extracted paint block. The length of the paint block is the length of the paint block in the traveling direction of the vehicle 70, and the width of the paint block is the distance between the up edge point and the down edge point.

  As shown in FIG. 3, the candidate line extraction unit 34 calculates the straight line components CL1, CLr, CR1, and CRr from the edge points extracted in the neighboring area Ra by using a statistical process. In the present embodiment, the statistical processing is Hough transform. The straight line components CLl and CRl have an up edge point as an element, and the straight line components CLr and CRr have a down edge point as an element. Then, the candidate line extracting unit 34 extracts the straight line components CL1, CLr, CR1, and CRr, and the edge points having continuity with the straight line components CL1, CLr, CR1, and CRr among the edge points extracted in the distant region Rb. , Edge lines are calculated respectively. For example, it is determined that there is continuity between edge points having a difference in a horizontal position between adjacent edge points in the traveling direction that is within a preset difference threshold. Then, for example, the candidate line extracting unit 34 selects an edge point of the far region Rb having continuity with the farthest edge point of the straight line component CRl, and sequentially selects an edge point having continuity in the far direction. , An edge line is calculated from the linear component CRl and the selected edge point. The same applies to edge points having continuity with other linear components. In the present embodiment, the edge line corresponds to the feature line.

  Further, the candidate line extracting unit 34 creates a pair of an edge line having an up edge point as an element and an edge line having a down edge point as an element from the calculated edge lines. At this time, the candidate line extraction unit 34 creates a pair of edge lines having a width similar to the width of the white line. Then, the candidate line extracting unit 34 extracts, as a candidate line, an edge line on the inner side near the vehicle 70 from the pair of edge lines. That is, on the left side of the vehicle 70, the candidate line extracting unit 34 extracts an edge line having a down edge line as an element as a candidate line, and on the right side of the vehicle 70, extracts an edge line having an up edge point as an element. . As shown in FIG. 4, a candidate line L1 on the left and a candidate line R1 on the right are extracted from the image of FIG.

  The line type determination unit 35 determines from the distribution of the edge points belonging to the extracted candidate line on the road surface, what kind of paint line the extracted candidate line corresponds to. The types of the paint line include a solid white line and a broken white line that define a lane, and a dashed auxiliary line drawn inside the white line that defines a lane. Normally, since the length of the line segment and the interval of the dashed white line and the dashed auxiliary line are different, the line type is determined from the distribution of the edge points on the road surface.

  The multiple line determination unit 36 determines whether or not the extracted candidate lines constitute a multiple line in each of the left and right sides of the vehicle 70. That is, when a plurality of candidate lines have been extracted within a preset range in the horizontal direction, the multiplex line determining unit 36 determines that the plurality of candidate lines constitute a multiplex line, and If a candidate line has been extracted, it is determined that the edge line forms a single line.

  The line selecting unit 37 considers the characteristics of the white line based on the line type determined by the line type determining unit 35 and whether or not the multiple line determined by the multiple line determining unit 36 is included. A candidate line to be recognized as a white line is selected on each of the left and right sides of 70. For example, when the candidate lines form a multiple line and all of the plurality of candidate lines correspond to white lines, the line selection unit 37 selects the innermost candidate line as a recognition target. Further, for example, if the candidate lines form a multiple line, and the innermost candidate line corresponds to the auxiliary line and the other candidate lines correspond to the white line, the line selecting unit 37 may select the line immediately after the auxiliary line. The outer candidate line is selected as a recognition target. The inside here is the side approaching the vehicle 70, and the outside is the side away from the vehicle 70.

  The branch determination unit 38 determines whether there is a branch section ahead of the vehicle 70 using the paint block extracted by the feature point extraction unit 33 and the candidate line selected by the line selection unit 37. Details of the branch determination process performed by the branch determination unit 38 will be described later.

  The recognition unit 39 normally recognizes the candidate line selected by the line selection unit 37. That is, the recognition unit 39 calculates a white line parameter from the selected candidate line and estimates the road shape. The white line parameters are, for example, well-known offset, yaw angle, curvature, lane width, pitch angle, and the like.

  In addition, when the branch determination unit 38 determines that there is a branch section, the recognition unit 39 switches the processing from the normal white line recognition processing to the white line recognition processing for the branch section. The details of the white line recognition processing of the branch section will be described later.

<Branch determination processing and branch section recognition processing>
Next, branch determination processing and branch section recognition processing will be described with reference to the flowchart in FIG. This processing procedure corresponds to the processing executed by the branch determination unit 38 and the recognition unit 39, and is executed every time an edge line is selected by the line selection unit 37.

  First, in step S10, the branch determination unit 38 determines whether another candidate line exists inside the own lane, that is, inside the candidate line selected by the line selection unit 37. Among the division lines, an auxiliary line drawn inside the white line from just before the branch section becomes a division line in the branch section. As shown in FIG. 6, when the auxiliary line S1 is drawn inside the white line R1 on the right side, the candidate lines L11 and R11 corresponding to the left and right white lines L1 and R1 and the auxiliary line A candidate line S11 corresponding to S1 is extracted. Then, the candidate lines L11 and R11 are selected by the line selection unit 37, and the candidate line S11 becomes another candidate line. In step S10, if it is determined that another candidate line exists inside the selected candidate line, the process proceeds to step S20. If it is determined that no other candidate line exists, the process proceeds to step S50.

  Subsequently, in step S20, the branch determination unit 38 determines whether or not the shape of another candidate line matches the shape of the division line. Specifically, when the shape of another candidate line satisfies the following conditions (A) to (C), the branch determination unit 38 determines that the shape of the other candidate line matches the shape of the dividing line.

  Condition (A): Another candidate line is determined as an auxiliary line by the line type determination unit 35. Since the paint line drawn from the near side of the branch section becomes the dividing line when the paint line is an auxiliary line, the condition is that another candidate line is determined to be an auxiliary line.

  Condition (B): The parallelism between the first candidate line and another candidate line is higher than the parallelism between the first candidate line and the second candidate line. The first candidate line is a candidate line selected on the opposite side of the vehicle 70 from the other candidate lines, and the second candidate line is a candidate line selected on the same side and outside of the vehicle 70 as the other candidate lines. A line. In FIG. 6, the left candidate line L11 is a first candidate line, and the right candidate line R11 is a second candidate line. As shown in FIG. 6, the candidate line L11 and the other candidate line S11 are white lines that separate the main lane, and the candidate line R11 is a white line that separates the branch lane. Therefore, the parallelism between the candidate line L11 and the other candidate line S11 becomes higher than the parallelism between the candidate line L11 and the candidate line R11.

  Condition (C): The horizontal distance between the other candidate line and the candidate line selected outside gradually increases toward the far side. As shown in FIG. 6, the other candidate line S11 becomes a white line on the main lane side, the candidate line R11 becomes a white line on the branch lane side, and the candidate line R11 gradually moves away from the other candidate line S11. The horizontal distance between the candidate line S11 and the candidate line R11 gradually increases toward the far side.

  If the shapes of the other candidate lines satisfy the conditions (A) to (C), in step S30, the branch determination unit 38 determines that there is a branch section. In this case, even if the branch section itself exists in the distant area Rb, it is determined that the branch section exists because the auxiliary line is drawn from the neighboring area Ra.

  Subsequently, in step S40, the recognition unit 39 changes the recognition target from the candidate line selected outside the other candidate line to be a dividing line to another candidate line to be a dividing line. In FIG. 6, the recognition target is changed from the candidate line R11 to another candidate line S11, and the recognition targets are the candidate line L11 corresponding to the white line L1 and another candidate line S11.

  On the other hand, in step S50, the branch determination unit 38 draws one or more paint blocks (hereinafter, “candidate blocks”) inside the selected candidate line, and draws the features of the candidate blocks starting from the branch section. It is determined whether or not the feature of the segment line (hereinafter referred to as a non-continuous segment line) matches the feature of the segment. That is, it is determined whether or not a non-continuous dividing line exists in the distant region Rb. As shown in FIG. 7, when the discontinuous segment line S2 exists in the distant region Rb, the feature point extraction unit 33 detects candidate blocks Ba to Bg in the distant region Rb. Specifically, when the candidate block exists and the feature of the candidate block satisfies the following conditions (D) to (F), the branch determination unit 38 determines that the feature of the candidate block matches the feature of the segment line. It is determined that they match.

  Condition (D): the length of each candidate block is shorter than a preset length threshold. Generally, a discontinuous dividing line is a dashed line, and each line segment of the dashed line is shorter in length than a dashed white line segment that separates the main lanes and an auxiliary line segment. Therefore, the length threshold is set so as to be shorter than the line segment of the general broken line white line and the line segment of the auxiliary line, and longer than the line segment of the general non-continuous division line.

  Condition (E): the length of each candidate block is larger than the width threshold. Generally, each line segment of the discontinuous division line is thicker than the width of a broken white line segment that separates the main lanes and the width of an auxiliary line segment. Therefore, the width threshold is set to the width of the white line recognized by the recognition unit 39 so far. For example, the width threshold is set to the width of the white line recognized in the latest predetermined period or predetermined distance at the present time. Further, the width threshold value may be set to a thickness of a general white line that separates the main lanes.

  Condition (F): When there are a plurality of candidate blocks, the size of each candidate block is the same. Each segment of the non-continuous dividing line is a segment having the same size and the same length and width. Therefore, when there are a plurality of candidate blocks, the condition is that the size of each candidate block is the same. At this time, the condition may be that at least one of the length and the width of each candidate block is the same.

If there is one or more candidate blocks and the characteristics of the candidate blocks satisfy the conditions (D) to (F), in step S60, the branch determination unit 38 determines that there is a branch section.
Subsequently, in step S70, the recognizing unit 39 performs Hough transform on the edge points belonging to the candidate block, and extracts a new line which is a straight line component serving as a dividing line. The edge points of the distant region Rb are less accurate than the edge points of the nearby region Ra, and the processing load increases if the Hough transform is performed on the edge points of the distant region Rb. Is not subjected to Hough transform. Only when it is determined that there is a branch section and the candidate block of the distant region Rb is known to be a dividing line, Hough transform is performed on the edge points of the distant region Rb, and a straight line component is extracted. In FIG. 7, a new line S12 is extracted from the up-edge points of the candidate blocks Ba to Bg.

  Subsequently, in step S80, the recognition unit 39 changes the recognition target from the candidate line selected outside the candidate block to the new line extracted in step S70. In FIG. 7, the recognition target is changed from the candidate line R12 corresponding to the white line R2 to the new line S11, and the recognition target is the candidate line L12 and the new line S11 corresponding to the white line L2.

  When a negative determination is made in step S20 or a negative determination is made in step S50, in step S90, the branch determination unit 38 determines that there is no branch section. That is, the recognition unit 39 does not change the selected recognition target.

Subsequently, in step S100, the recognition unit 39 recognizes the selected recognition target as a white line. This ends the processing.
<Action>
In the flowchart of FIG. 5, in the processing of steps S10 to S30, a branch section is determined from the shape of another candidate line, and in the processing of steps S50 to S60, a branch section is determined from the characteristics of a candidate block. That is, in the processing of steps S10 to S30, after a candidate line including a straight line component is extracted from the edge point, a branch section is determined from the shape of the candidate line. On the other hand, in the processing of steps S50 to S60, before the straight line component is extracted from the edge point, the branch section is determined from the features of the candidate block.

  If the branch section is determined only by the processing of steps S10 to S30, in the branch section where the non-continuous partition line is drawn, the non-continuous partition line is extracted as a candidate line, and the candidate line corresponding to the non-continuous partition line is extracted. Is determined from the shape of. Therefore, the determination of the branch section is slower than the case of determining the branch section from the features of the candidate block. In particular, as shown in FIG. 7, in a narrow-angle branch section where the branch lane gently branches from the main lane, it takes time to satisfy the conditions (B) and (C), and the determination of the branch section is slow. Become.

  On the other hand, the processing of steps S50 to S60 is provided, and when there is no auxiliary line inside the selected candidate line, the branch section is determined from the characteristics of the candidate block, thereby quickly determining the branch section. Is realized.

<Effect>
According to the first embodiment described in detail above, the following effects can be obtained.
(1) If a candidate block exists inside the selected candidate line and the characteristics of the candidate block satisfy the conditions (D) to (F), it is determined that there is a branch section, and the outside of the candidate block is determined. The candidate line selected in is removed from the recognition target. Therefore, even when the dividing line exists in the distant region Rb and is not extracted as a candidate line, the branch section can be determined. That is, it is possible to quickly determine a branch section with a low processing load, and suppress erroneous recognition of a white line in the branch section.

  (2) If it is determined that there is a branch section, a new line that is a straight line component is extracted from the edge point of the candidate block. That is, only when it is determined to be a branch section, a straight line component is extracted even in the distant region Rb. Then, the recognition target is switched from the candidate line extracted outside the candidate block to a new line. Therefore, accurate white line recognition in a branch section can be realized while suppressing an increase in processing load.

  (3) When there are a plurality of candidate blocks, by adding the condition (F) to the conditions (D) and (E), erroneous recognition of noise is suppressed, and the branch section is determined more accurately. be able to.

  (4) When the auxiliary line drawn inside the white line from the front of the branch section is a dividing line in the branch section, the auxiliary line corresponds to the auxiliary line even if the branch section is in the distant region Rb. Candidate lines are extracted. Therefore, the branch section is determined from the shape of the candidate line corresponding to the auxiliary line, and the recognition target can be switched from the white line on the branch lane side to the auxiliary line serving as the dividing line.

(Other embodiments)
The embodiments for carrying out the present disclosure have been described above, but the present disclosure is not limited to the above-described embodiments, and can be implemented with various modifications.

  (A) In the above embodiment, when all of the conditions (D) to (F) are satisfied, it is determined that the feature of the candidate block matches the feature of the discontinuous dividing line. However, the present invention is not limited to this. Absent. For example, when at least one of the condition (D) and the condition (E) is satisfied, it may be determined that the feature of the candidate block matches the feature of the discontinuous dividing line.

  (B) Condition (G): when there are a plurality of candidate blocks, the inclination of each candidate block is in the same direction may be added to conditions (D) to (F). The slope of each candidate block may be calculated by, for example, fitting edge points belonging to each candidate block. It is sufficient that one of the inclination in the direction parallel to the traveling direction of the vehicle 70 and the inclination in the direction perpendicular to the traveling direction of each candidate block is the same. In addition, when at least one of the condition (D) and the condition (E) and at least one of the condition (F) and the condition (G) are satisfied, it is determined that the feature of the candidate block matches the feature of the discontinuous dividing line. You may.

  (C) In the above embodiment, when it is determined that there is a lane marking, the recognition target is changed from the candidate line corresponding to the white line on the branch lane side to a candidate line corresponding to the lane marking or a new line. Alternatively, the recognition process of the white line on the branch lane side may be stopped. In this case, of the two sides of the vehicle 70, only the recognition processing of the white line on the side opposite to the branching side may be continued. By continuing only the recognition processing of the white line on the opposite side, it is possible to suppress the recognition of the white line from being dragged toward the branch lane.

  (D) A plurality of functions of one component in the above embodiment may be realized by a plurality of components, or one function of one component may be realized by a plurality of components. . Also, a plurality of functions of 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. Further, a part of the configuration of the above embodiment may be omitted. Further, at least a part of the configuration of the above-described embodiment may be added to or replaced with the configuration of another above-described embodiment. In addition, all aspects included in the technical idea specified only by the language described in the claims are embodiments of the present disclosure.

  (E) In addition to the above-described lane-line recognition device, a system including the lane-line recognition device as a component, a program for causing a computer to function as the lane-line recognition device, and a non-transitional device such as a semiconductor memory storing the program. The present disclosure can be realized in various forms, such as an actual recording medium and a lane marking recognition method.

  DESCRIPTION OF SYMBOLS 10 ... In-vehicle camera, 30 ... Recognition apparatus, 33 ... Feature point extraction part, 34 ... Candidate line extraction part, 37 ... Line selection part, 38 ... Branch determination part, 39 ... Recognition part, 70 ... Vehicle.

Claims (7)

A lane marking recognition device (30) for recognizing a lane marking dividing a lane of a road from an image around a vehicle (70) taken by an on-vehicle camera (11),
Assuming that a range from the vehicle to a preset distance is a near area, and a far area than the near area is a far area, and in the near area and the far area of the image, feature points representing a boundary of road surface paint are extracted. A feature point extraction unit (33) configured as
A linear component is extracted using statistical processing from the feature points extracted in the near region, and the continuity between the extracted linear component and the feature point extracted in the distant region with the linear component. A candidate line extraction unit (34) configured to extract a feature line from the feature point having
A line selection unit (37) configured to select the characteristic line to be recognized as the partition line from the characteristic lines extracted by the candidate line extraction unit;
One paint block, which is a rectangular area formed by the feature points extracted by the feature point extraction unit, is located on the inner side closer to the vehicle than the feature line selected by the line selection unit. As described above, and the paint block present on the inner side than the lane marking as a candidate block, at least one of the length of the candidate block and the width of the candidate block in the traveling direction of the vehicle, the main lane and A branch determining unit (38) configured to determine that there is a branch section when having characteristics of a line segment forming a broken line that separates the branch lane;
When the branch determination unit determines that there is the branch section, a recognition unit configured to exclude from the recognition target the feature line selected on the outer side that is farther from the vehicle than the candidate block. And a lane marking recognition device comprising:   The recognizing unit, when the branch determining unit determines that there is the branch section, extracts a new line that is a linear component from the feature point of the candidate block using the statistical processing, and the line selecting unit The lane marking recognition device according to claim 1, wherein the extracted new line is recognized as the lane marking instead of the feature line selected outside the candidate block.   When a plurality of candidate blocks are present, the branch determination unit may further include a condition that the size of each of the candidate blocks is the same and that each of the candidate blocks has the same inclination in the same direction. The lane marking recognition device according to claim 1 or 2, configured to determine that there is a section.   4. The apparatus according to claim 1, wherein the branch determination unit is configured to determine that there is the branch section when the length of the candidate block is shorter than a preset length threshold. 5. 6. The lane marking recognition device according to 1.   The branch determination unit may determine the presence of the branch section when the width of the candidate block is wider than the width threshold, with the width of the partition line recognized by the recognition unit as a width threshold. The lane marking recognition device according to any one of claims 1 to 4, wherein the lane marking recognition device is configured as follows. The branch determination unit may include another candidate line that is the feature line extracted by the candidate line extraction unit inside the feature line selected by the line selection unit, and may include the other candidate line. If the line has the characteristics of the shape of the dividing line, it is determined that there is a branch section,
The recognizing unit, when the branch determining unit determines that there is the branch section, instead of the feature line selected outside the other candidate line by the line selecting unit, the other candidate The lane marking recognition device according to any one of claims 1 to 5, wherein the lane marking recognition device is configured to recognize a line as the lane marking.   The recognizing unit stops the process of recognizing the lane marking on the side where the candidate block exists on the inner side when the branch determining unit determines that the branch section exists, and the candidate block exists on the inner side. 2. The lane marking recognition device according to claim 1, wherein the lane marking recognition apparatus is configured to continue recognition processing of the lane markings on the side that does not.

Images