JP2003187252A - Traffic lane boundary detector - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a traffic lane boundary detector capable of stably detecting a traffic lane boundary. <P>SOLUTION: An outermost contour extraction part 15 extracts outermost contour information of a group of white lines based on original image data stored in a frame buffer part 13 and contour data including positional information of an edge detected by an edge detection part 14. For details, the outermost contour extraction part 15 judges whether or not the edge corresponds to gaps generated between white lines to constitute a group of the white lines based on the contour data including the positional information of the edge extracted from the original image data and deletes the edge corresponding to the gaps from the contour data. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a lane boundary detecting device for detecting a lane boundary position of a marking line from image data obtained by an image pickup device.

[0002]

2. Description of the Related Art Conventionally, as an apparatus for controlling the traveling of a moving body such as an automobile, there is an automatic driving apparatus for recognizing a running path shape and automatically performing steering control. In order to realize this automatic driving, it is necessary to recognize the road,
For example, a detection device for detecting the lane boundary position based on the marking line is required. As this type of detection device, for example, Japanese Patent Laid-Open No.
Those described in JP-A-3-142478, JP-A-6-20189, and JP-A-6-119594 are known.

These detecting devices utilize the characteristic that the white line is brighter than the road surface, and the image data is subjected to processing such as differentiation to extract the contour data of the high-luminance portion from the photographed image. After that, among the contour data extending in the direction along the road, the data of the interval corresponding to the predetermined interval (for example, the design reference value) is regarded as the outline of the white line.

[0004]

By the way, the lane marking is
The shape differs depending on the condition of the track. Fig.7 (a)-
(D) is a figure which shows the example. As shown in FIG. 7A, in a general section, one side (left side) is a solid demarcation line 5
1 is provided, and a broken line 52 is provided on the other side (right side). On the other hand, as shown in FIG. 7B, in sections such as entrances, exits, junctions, and uphill lanes of a highway, the broken demarcation line 53 provided on one side (left side) is It is formed wider than the partition line 52 on the right side. In this case, according to the conventional method, the division line 5 in FIG.
3 may not be recognized.

Further, as shown in FIG. 7 (c), in a section such as a curve where attention is required, a marking line (composite line) having a wide broken line is provided on either the left or right side of the solid line or the broken line. 54 and 55 are provided. In the opposite lane section, a dividing line (called a multiple white line) 56 provided with a plurality of solid lines is provided. In these cases, the contour of the lane marking may not be stably detected. The reason is as follows.

Since a wide-angle lens is used in the image pickup device, when the composite white lines 54, 55 and the multiple white line 56 are reflected in the peripheral portion of the screen, they are slightly degenerated, so that a gap between the white lines and the white line can be seen. I can't see it. Also, due to the dirt and peeling of the white line itself, the gaps between them are visible or invisible.

The white line portion is saturated due to improper exposure of the camera and the gap cannot be seen. For example, when the shadow of the vehicle occupies a large area on the screen, the white line portion is overexposed, and it is difficult to determine the boundary position, and it is difficult to determine the gap.

Since the actual white line is thinned due to dirt / peeling and the contrast is lowered, it is difficult to determine the gap. Due to these factors, the appearance of the gap is not always constant, so the contour of the lane markings is not stable, and there is a problem that the lane boundary position cannot be detected stably.

An object of the present invention is to provide a lane boundary detecting device which can stably detect a lane boundary.

[0010]

In order to solve the above-mentioned problems, the invention according to claim 1 is based on image data of a traveling road photographed by an image pickup device and including a marking line in its field of view, based on the lane boundary position of the road surface. In the lane boundary detection device that detects
A contour extracting means for extracting the outermost contour information of the white line group from the image data is provided, and the lane boundary position is set based on the outermost contour information.

According to a second aspect of the present invention, in the lane boundary detecting apparatus according to the first aspect, the contour extracting means, based on the contour data including the position information of the edge extracted from the image data, the edge is extracted. Determines whether or not corresponds to a gap generated between the white lines forming the white line group, deletes the edge corresponding to the gap from the contour data, and extracts the outermost contour information of the white line group.

The invention according to claim 3 is the invention according to claim 1 or 2.
In the lane boundary detection device described above, the contour extraction means measures an interval between two edges, compares the measured value with a set value, and determines whether or not the measured value is smaller than the set value. Means and the density difference determining means for comparing the density differences on both sides of the two edges based on the determination result of the edge interval measuring means, and based on the determination results of the edge interval measuring means and the density difference determining means. And a gap determining unit that determines whether or not the two edges correspond to a white line gap and deletes the corresponding edges from the outermost contour information.

According to a fourth aspect of the present invention, in the lane boundary detecting apparatus according to any one of the first to third aspects, the offset setting means stores offset information, and the offset setting means stores the offset information. Boundary setting means for setting the lane boundary position based on the offset information.

According to a fifth aspect of the present invention, in the lane boundary detecting apparatus according to the fourth aspect, the boundary setting means indicates the outermost contour indicating the positions of two edges corresponding to the white line group of the outermost contour information. A first lane boundary position based on the data is calculated, a second lane boundary position obtained by offsetting one position of the two outermost contour data based on the offset information, and the second lane boundary position based on the offset information. And a third lane boundary position obtained by offsetting the other position of the one outermost contour data, and outputting one of the first to third lane boundary positions according to a selection instruction. .

(Operation) According to the invention described in claim 1,
The contour extracting means extracts the outermost contour information of the white line group from the image data, and sets the lane boundary position based on the outermost contour information. Therefore, the outline of the white line group (outermost outline information)
Is stable, and the lane boundary position is stably detected.

According to the second aspect of the invention, based on the contour data including the position information of the edge extracted from the image data by the contour extracting means, the edge is placed between the white lines forming the white line group. It is determined whether or not it corresponds to the generated gap, the edge corresponding to the gap is deleted from the contour data, and the outermost contour information of the white line group is extracted. Therefore, the outermost contour information of the white line group is reliably extracted, and the lane boundary position is stably detected.

According to the third aspect of the invention, the contour extracting means measures the interval between the two edges, compares the measured value with the set value, and determines whether the measured value is smaller than the set value. Based on the determination result of the edge interval measuring unit and the density difference determining unit, the density difference determining unit that compares the density differences on both sides of the two edges based on the determination result of the edge interval measuring unit. And a gap determining means for determining whether or not the two edges correspond to the gap of the white line and deleting the corresponding edges from the outermost contour information. Therefore, the edges corresponding to the gaps between the white lines are reliably determined and deleted.

According to the invention described in claim 4, the lane boundary position is set based on the offset information stored in the offset setting means. Therefore, the lane boundary position is set to an arbitrary position with respect to the outermost contour information of the white line group.

According to the invention described in claim 5, the boundary setting means calculates the first lane boundary position based on the outermost contour data indicating the positions of the two edges corresponding to the white line group of the outermost contour information. Then, the second lane boundary position where one position of the two outermost contour data is offset based on the offset information, and the other position of the two outermost contour data is offset based on the offset information. The third lane boundary position is calculated, and one of the first to third lane boundary positions is output from the output means according to the selection instruction. Therefore, the lane boundary position corresponding to the request (selection instruction) can be easily obtained.

[0020]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of a lane boundary detecting apparatus embodying the present invention will be described below with reference to FIGS.

The lane boundary detection device is installed on a moving body,
A lane boundary position (lane boundary position) is determined by processing an image obtained by photographing a running path on which the moving body travels. FIG. 1 is a schematic block diagram of a lane boundary detection device 10.

Lane boundary detecting device (hereinafter, detecting device) 10
Is an image pickup unit 11, an A / D converter 12, a frame buffer
(Frame Buffer) unit 13, edge detection unit 14, outermost contour extraction unit 15, offset setting unit 16, lane boundary setting unit 1
7 and an output unit 18. The detection device 10 is, for example, a CPU
Is connected to the control unit 19, which controls the above units 11 to 18.

The image pickup section 11 outputs image data of a road image. The A / D converter 12 performs analog-digital conversion on the image data input from the imaging unit 11, and outputs the converted data. The frame buffer unit 13 is A /
The output signal of the D converter 12 is stored as original image data.

The edge detection unit 14 performs edge detection processing on the original image data stored in the frame buffer unit 13 to extract data of a contour line portion (hereinafter, contour data). The edge detection process is performed by a predetermined operator (for example,
Prewitt operator) is applied to obtain differential image data. Then, the edge detection unit 14 outputs, for a region having an intensity equal to or higher than the noise level in the differential image data, contour data in which the center of the region is regarded as an edge point. It should be noted that, for an area having an intensity equal to or higher than the noise level, the point having the maximum intensity may be regarded as an edge point and the contour data may be output.

The outermost contour extraction unit (hereinafter referred to as extraction unit) 15 is
Based on the contour data and the original image data stored in the frame buffer unit 13, the position information of the outermost contour of the white line and the white line group is obtained. The white line group is composed of a plurality (a plurality of types) of white lines, and is composed of the composite white lines 54 and 55 in FIG.
It includes the multiple white line 56 shown in (d).

The extraction unit 15 determines whether or not each edge point included in the contour data corresponds to a gap generated between white lines, and if so, deletes the edge point from the contour data. Specifically, the extraction unit 15 pays attention to a negative edge (edge point having a negative differential intensity) of the contour data, and after the negative edge occurs, a plus edge (an edge point having a positive differential intensity) occurs. Measure the interval until. The extraction unit 15 has an interval within a predetermined set value (for example, 4 pixels, and the set value is stored in advance in the extraction unit 15 or another register based on white lines forming a general white line group). If so, consider that these edges may indicate gaps between white lines. Next, the extraction unit 15 reads the densities of a plurality of pixels outside each of the two edges that are considered to indicate a gap from the frame buffer unit 13 and compares the respective average values. If the density difference is equal to or smaller than the set value, the densities of the read pixels represent the densities of the white lines. Therefore, the extraction unit 15 determines that these edges are edges corresponding to the gaps between the white lines. . Then, the extraction unit 15 deletes the edge data from the contour data. Therefore, the contour data includes the contour data of the white line group and does not include the edge data corresponding to the gap between the white lines. Then, the extraction unit 15 outputs the contour data as contour position information. When the contour data are separated in a broken line shape along the traveling direction, the extraction unit 15 complements the contour data along the traveling direction, and among the contour data of the maximum width (contour data of the white line group). , And outputs two pieces of contour data having wide widths as contour position information.

The offset setting unit 16 stores an offset value from the outermost lane boundary position used by the lane boundary setting unit 17. This offset value is used by the control unit 19
Is set in advance.

The lane boundary setting unit 17 generates the first lane boundary position data, which is the intermediate position between both edges of the contour data, based on the contour position information (contour data) from the extraction unit 15. In addition, the lane boundary setting unit 17 is a second lane boundary that is a position where one edge (for example, right side) of the contour data is offset by the first offset value from the other edge based on the outermost contour data and the offset value. Generate position data. Furthermore, the lane boundary setting unit 17
Based on the outermost contour data and the offset value, third lane boundary position data, which is a position where the other edge (for example, the left side) is offset to the one edge by the second offset value, is generated. Then, the lane boundary setting unit 17 outputs the generated first to third lane boundary position data.

The output unit 18 selects one of the first to third lane boundary position data output by the lane boundary setting unit 17 according to an instruction from the control unit 19 and outputs the selected data. This selection data is sent to a system control computer such as an alarm device and a speed control device, and the moving body is controlled based on the selection data.

FIG. 2 is a block diagram showing the structure of the extraction unit 15. In the figure, the extraction unit 15 and the output unit 18
The lane boundary setting unit 17 between the two is omitted. Extraction unit 1
5 includes an edge interval measuring unit 21, a density difference determining unit 22, and a gap determining unit 23.

The edge interval measuring unit 21 pays attention to a negative edge (edge point having a negative differential intensity) of the contour data, and after the negative edge occurs, a plus edge (edge point having a positive differential intensity). Measure the interval until the occurrence of. If the distance is within a predetermined set value, the edge distance measuring unit 21 considers that these edges may indicate a gap generated between the white lines, and the information thereof is used as the density difference determining unit 22 and the gap determining unit. To 23. The set value is the number of pixels corresponding to the gap between the white lines (eg 4 pixels)
This set value is stored in advance in the edge interval measuring unit 21 or another register based on the white lines forming the general white line group.

The density difference determination unit 22 reads the densities of a plurality of pixels outside each of the two edges that are considered to indicate a gap from the frame buffer unit 13, compares the respective average values, and compares the results. Is output to the gap determination unit 23.

The gap determining unit 23 is the edge gap measuring unit 21.
It is determined whether the focused edge is the edge corresponding to the gap between the white lines, based on the measurement result of 1. and the measurement result of the density difference determination unit 22. Then, the gap determination unit 23 deletes the edge data from the contour data when the edge data corresponds to the gap between the white lines. That is, the gap determination unit 2
3 outputs the contour position information including the contour data of the white line group but not the edge data corresponding to the gap between the white lines.
When the contour data are separated in a broken line shape along the traveling direction, the gap determination unit 23 complements the contour data along the traveling direction, and the contour data having the maximum width (of the contour data of the white line group , Two contour data having a wide width are output as contour position information.

Next, the operation of the lane boundary detecting device configured as described above will be described. Now, the original image data 31 shown in FIG. 5A is recorded in the frame buffer unit 13. 5A to 5C are visualizations of image data.

The edge detecting section 14 of FIG. 1 reads out the original image data from the frame buffer section 13 along the scanning line.
FIG. 3A shows a schematic waveform of original image data obtained by reading the original image recorded in the frame buffer unit 13 on the scanning line S1 of FIG. 5A. The horizontal axis represents time and the vertical axis represents concentration value.

The edge detecting section 14 applies a predetermined operator to the original image data to generate differential image data. Further, the edge detection unit 14 outputs contour data in which the center of the area is regarded as an edge point for the area having the intensity equal to or higher than the noise level of the differential image data. FIG. 5B shows the contour data 31a for one screen.

The extraction unit 15 in FIG. 1 pays attention to the minus edge B shown in FIG. 3C, and measures the interval from the occurrence of the minus edge B to the next plus edge C. When this interval is within the set value, the extraction unit 15 checks whether or not both sides of the edges B and C are white lines.

First, as shown in FIG.
The density values of the pixels a and b on the front side (left side in the drawing) of the pixel corresponding to the minus edge B in the scanning direction are obtained from the original image (that is, the original image data stored in the frame buffer unit 13). Next, the extraction unit 15 similarly obtains the density values of the pixels c and d on the rear side (right side in the drawing) in the scanning direction of the pixel corresponding to the plus edge C. Then, the extraction unit 15 calculates the density difference between the average value of the density values of the pixels a and b and the average value of the density values of the pixels c and d. Since the calculated density difference is less than or equal to the set value, the extraction unit 15
The edges B and C are deleted from the contour position information. as a result,
As shown in FIG. 3D, a plus edge A and a minus edge D remain in the contour position information.
Is the outermost contour edge.

Similarly, the extraction unit 15 performs the same processing on the original image data of the composite white line 55 appearing on the right side of FIG. 5A (the right side portion in FIG. 3), and FIG. Leave the outermost contour edge shown. FIG. 6A is information (image data 32) of the remaining outermost contour edge.

Next, the extraction unit 15 uses the contour position information (see FIG. 6).
Outermost contour data (contour position information) 33a shown by a solid line in FIG. 6B by complementing the edge data by the white line group on the left side and the edge data by the white line group on the right side in (a) along the traveling direction. 33d is generated.

Based on these outermost contour data 33a to 33d, the lane boundary setting unit 17 of FIG. 1 calculates the intermediate position of the outermost contour data 33a to 33d, and the first position shown by the dashed line in FIG. 6B. The lane boundary position data 34a and 34b are generated.

Although not shown, the lane boundary setting unit 17
Generates the second and third lane boundary position data based on the outermost contour data shown in FIG. 6 (b). As described in detail above, according to this embodiment, the following effects can be obtained.

(1) In this embodiment, the outermost contour extraction unit 1
In No. 5, the outermost contour information of the white line group is extracted based on the original image data stored in the frame buffer unit 13 and the contour data including the edge position information detected by the edge detection unit 14. As a result, the edges corresponding to the gaps between the white lines are not extracted, so that the outline of the white line group (outermost outline information) is stable and the lane boundary position can be detected stably.

(2) In this embodiment, the outermost contour extraction unit 1
Reference numeral 5 determines whether or not the edge corresponds to a gap generated between the white lines forming the white line group, based on the contour data including the position information of the edge extracted from the original image data. The corresponding edge is deleted from the contour data. As a result, the outermost contour information of the white line group is reliably extracted, so that the lane boundary position can be stably detected.

(3) In this embodiment, the outermost contour extraction unit 1
Reference numeral 5 denotes an edge interval measuring unit 21 that measures an interval between two edges, compares the measured value with a set value, and determines whether the measured value is smaller than the set value, and a determination result of the edge interval measuring unit. Based on the density difference determination unit 22 that compares the density differences on both sides of the two edges, and whether the two edges correspond to the white line gap based on the determination results of the edge interval measurement unit and the density determination unit. And a gap determining unit 23 that determines and deletes the corresponding edge from the outermost contour information. As a result, it is possible to easily determine whether or not both sides of the edge corresponding to the gap are white lines by comparing their density differences, and it is possible to reliably delete the edge corresponding to the gap between the white lines. .

(4) In this embodiment, the offset setting section 16 for storing the offset information is provided. As a result, by setting the lane boundary position from the outermost contour information of the white line group based on the offset information, the lane boundary position can be set at an arbitrary position.

(5) In this embodiment, the lane boundary setting section 17 calculates the first lane boundary position based on the outermost contour data indicating the positions of the two edges corresponding to the white line group of the contour position information, and offsets it. A second lane boundary position obtained by offsetting one position of the two outermost contour data based on the information, and a third lane boundary position obtained by offsetting the other position of the two outermost contour data based on the offset information. The lane boundary position is calculated. The output unit 18 is configured to output one of the first to third lane boundary positions according to the selection instruction. As a result, the lane boundary position corresponding to the request (selection instruction) can be easily obtained.

The present invention is not limited to the above embodiment, but may be modified as follows. It is embodied in a lane boundary detection device in which at least one of the image pickup unit 11, the A / D converter 12, and the frame buffer unit 13 is a separate body.

In the edge detection unit 14, as an operator to be used, Sobel, Roberts, Kirsch, Robinson, a simple three-level operator, etc. are used to perform edge detection. In addition, edge detection is performed using a simple Laplacian filter, a noise suppression type Laplacian filter, a bridge filter, a surfboard filter, or the like.

The output unit 18 stores the instruction of the control unit 19 in advance and outputs one of the first to third lane boundary position data selected based on the instruction. The lane boundary setting unit 17 generates one of the first to third lane boundary position data according to the instruction of the control unit 19.

Edge detector 14, outermost contour extractor 1
5. The CPU may execute a program to provide at least one of the functions provided by the lane boundary setting unit 17.

Next, the technical idea understood from the above embodiment will be described below together with its effects. (A) The contour extracting means pays attention to a minus edge,
The interval from the occurrence of the negative edge to the occurrence of the next edge is measured, and when the measured value is within the set value, it is determined whether or not both sides of both edges are white lines. The lane boundary detection device according to claim 2, wherein both edges are deleted from the contour data. Therefore, the edge corresponding to the gap between the white lines can be surely removed.

(B) With respect to the edge extracted from the original image data, paying attention to the negative edge, measuring the interval from the occurrence of the negative edge to the occurrence of the next edge, and setting the measured value. An edge determination method comprising: comparing a value with a value, and determining whether or not both edges correspond to a gap generated between white lines based on the comparison result. Therefore, the edge corresponding to the gap between the white lines can be surely removed.

[0054]

As described above in detail, according to the present invention, it is possible to provide a lane boundary detecting device capable of stably detecting a lane boundary.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a lane boundary detection device.

FIG. 2 is a block diagram for explaining an outermost contour extraction unit.

FIG. 3 is an explanatory diagram of contour line detection.

FIG. 4 is an explanatory diagram of gap determination.

FIG. 5 is an explanatory diagram of a white line.

FIG. 6 is an explanatory diagram of a white line.

FIG. 7 is an explanatory diagram of a white line.

[Explanation of symbols]

15 ... Outermost contour extraction section (contour extraction means), 16 ... Offset setting section (offset setting means), 17 ... Boundary setting section (boundary setting means), 18 ... Output section (output means), 21 ...
Edge interval measuring unit (edge interval measuring unit), 22 ... Density difference determining unit (density difference determining unit), 23 ... Gap determining unit (gap determining unit).

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Munehiro Takayama             Aichi, 2-chome, Asahi-cho, Kariya city, Aichi prefecture             N Seiki Co., Ltd. (72) Inventor Toshiaki Kakinami             Aichi, 2-chome, Asahi-cho, Kariya city, Aichi prefecture             N Seiki Co., Ltd. (72) Inventor Hisashi Satonaka             1 Toyota Town, Toyota City, Aichi Prefecture Toyota Auto             Car Co., Ltd. (72) Inventor Makoto Nishida             1 Toyota Town, Toyota City, Aichi Prefecture Toyota Auto             Car Co., Ltd. (72) Inventor Shin Takahashi             Aichi Prefecture Nagachite Town Aichi District             Ground 1 Toyota Central Research Institute Co., Ltd. (72) Inventor Yoshiki Ninomiya             Aichi Prefecture Nagachite Town Aichi District             Ground 1 Toyota Central Research Institute Co., Ltd. F term (reference) 5B057 AA16 BA02 DA08 DB02 DB05                       DB09 DC16 DC22                 5H180 AA01 CC04 LL06 LL09                 5L096 AA03 AA06 BA04 CA02 DA01                       FA06 FA14 GA05

Claims (5)

[Claims] 1. A lane boundary detecting device for detecting a lane boundary position of a road surface from image data of a traveling road imaged by an imaging device and including lane markings in its field of view, wherein outermost contour information of a white line group is detected from the image data. A lane boundary detecting device comprising a contour extracting means for extracting, and setting a lane boundary position based on the outermost contour information. 2. The contour extracting means, based on the contour data including the position information of the edge extracted from the image data, determines whether the edge corresponds to a gap generated between the white lines forming the white line group. The lane boundary detection apparatus according to claim 1, wherein the outermost contour information of the white line group is extracted by determining whether or not the edge corresponding to the gap is deleted from the contour data. 3. The edge extracting means measures the distance between two edges, compares the measured value with a set value, and judges whether the measured value is smaller than the set value or not. Based on the determination result of the edge interval measuring means,
Density difference determining means for comparing the density differences on both sides of one edge, and based on the determination results of the edge interval measuring means and the density difference determining means, it is determined whether or not the two edges correspond to a white line gap. The lane boundary detection device according to claim 1 or 2, further comprising: a gap determination unit that deletes a corresponding edge from the outermost contour information. 4. An offset setting means for storing offset information, and a boundary setting means for setting a lane boundary position based on the offset information stored in the offset setting means. 4. The lane boundary detection device according to claim 3. 5. The boundary setting means calculates a first lane boundary position based on outermost contour data indicating positions of two edges corresponding to a white line group of the outermost contour information, and based on the offset information. A second lane boundary position in which one position of the two outermost contour data is offset, and a third lane in which the other position of the two outermost contour data is offset based on the offset information. The lane boundary detection device according to claim 4, further comprising output means for calculating a boundary position and outputting one of the first to third lane boundary positions in accordance with a selection instruction.