JP2003118521A - White line detecting device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To highly precisely detect a road while line. SOLUTION: A microcomputer 2 generates an edge image from an image inputted from a camera 1, sets a filter according to the inclination and the concentration value of the white line of in the edge image, and renews it as the filter used for generating the edge image. It also generates a while line model from the while line obtained from the edge image and sets it as a base in setting a white line detecting region. The white line model can be precisely calculated by renewing the edge filter by the inclination and the concentration value of the white line so that the road white line can be precisely detected.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a white line detecting device for detecting a white line from a road image captured by an image pickup means.

[0002]

2. Description of the Related Art In consideration of automatic driving of a vehicle, it is indispensable to detect the position of the vehicle within the driving lane. In order to detect the position of the host vehicle in the traveling lane with high accuracy, it is important to recognize the traveling lane itself with high accuracy. Although various methods have been studied to detect the driving lane, the driver obtains most of the information necessary for driving from the driving lanes separated by white lines. Is important for detecting the position of the own vehicle. As a means for detecting the white line of the traveling lane, there is a method of performing image processing on the road image captured by the imaging means. In order to recognize the white line of the traveling lane, edge extraction is performed to generate an edge image, and the white line is detected from the obtained edge image. An edge is a portion in the image where the density change is large, a so-called contour line, and has an advantage that it is easy to detect even if the density changes due to a shadow or dirt.

Thus, when a white line is detected, the edge image obtained by edge extraction is used. However, since the edge extraction is performed by a filter calculation, whether the filter calculation process is successful depends on the accuracy of the white line detection. Decide. Conventionally, as a technique for selecting a filter for edge detection and obtaining a clear edge image, the method is disclosed in JP 2001-101428 A.
There is one described in the official gazette. This technique is an image processing apparatus that obtains an edge image by extracting edges from an image with a filter, and includes filter selection means. Then, the filter selection means selects a filter according to the luminance of the pixel, the edge strength, and the peak value of the histogram obtained by extracting and projecting the edge.

[0004]

However, in such a conventional technique as described above, since the judgment at the time of selecting the filter is based only on the brightness difference of the image, that is, the contrast information, it is possible to determine the actual driving environment. In a state where there are many noise components such as road surface display and traveling vehicles, the edge image in which the noise component is emphasized is more emphasized than the white line which is the detection target, and it is very difficult to detect the white line from the edge image. There were challenges. Also, the filter selected is the one that reacts most strongly to vertical edge components, so if you try to detect a white line on a curved road,
There was a problem that an optimal edge image could not be obtained.

In view of such conventional problems, the present invention has an edge image in which the edge image of the white line to be detected is emphasized more than other noise components even if the ambient brightness or the road shape changes. It is an object of the present invention to provide a white line detection device that can generate the edge line and can obtain an optimum edge image even for a white line on a curved road.

[0006]

The invention according to claim 1 is
An image pickup means for picking up a predetermined area around the own vehicle mounted on the vehicle, an image storage means for storing an image obtained from the image pickup means, and an own vehicle for the image stored by the image storage means The white line detection area setting means for setting the area for detecting the white line of the traveling lane, and the edge image for generating the edge image by performing the edge extraction by the filter calculation with respect to the area set by the white line detection area setting means. A generation means, a white line detection means for performing white line detection from the edge image extracted in the white line detection area, and an edge filter update means for updating the filter coefficient used in the filter calculation in the edge image generation means, The edge filter updating means applies a filter based on both the density value in the white line detection area and the slope of the white line detected by the white line detection means. It was that the new to.

According to a second aspect of the present invention, a white line model is used to represent the shape of the road white line by a mathematical model using a plurality of parameters representing the road shape and vehicle behavior, and the white line model and the white line detecting means are used. The parameter updating means for updating a plurality of parameters of the white line model is provided so that the results detected in step 1 are matched.

According to a third aspect of the present invention, a plurality of areas for detecting a white line of the traveling lane of the host vehicle are located farther from the near side as viewed from the host vehicle in the image stored by the image storage means. It is supposed to be set to one direction.

According to a fourth aspect of the present invention, the filter calculation processing is performed for each area set by the white line detection area setting means.

According to a fifth aspect of the present invention, the next white line detection is performed based on both the density value in the white line detection area set by the white line detection area setting means and the inclination of the white line detected by the white line detection means. The edge filter is updated by predicting the filter in which the edge of the white line of the area is most emphasized.

According to a sixth aspect of the present invention, when the white line detection area to be filtered next is the area closest to the host vehicle, the white line detection area closest to the host vehicle on the previously input image is selected. The filter used for the calculated and derived region is updated as the filter used for the filter calculation.

[0012]

According to the first aspect of the present invention, the edge filter for generating the edge image is updated on the basis of both the density value in the white line detection area and the inclination of the white line. Since an appropriate filter calculation can be performed according to the curve, it is possible to eliminate the influence of the detection error due to the curve and the influence of the detection error due to the appearance of the white line when the host vehicle is traveling to the left and right in the driving lane.

According to the second aspect of the present invention, the road white line is represented by the white line model from the result detected by the white line detecting means, which can be used as a basis for setting the white line detection area. By recognizing the road white line by the white line model, the road white line can be captured only by setting a small white line detection area, so that the processing speed at the time of edge filter calculation processing is improved.

According to the third aspect of the present invention, since the white line detection area near the own vehicle has little noise, it is easy to detect an accurate road white line. Therefore, the white line detection area can accurately capture the road white line by sequentially setting the area to the far side based on the white line detected in the white line detection area closest to the own vehicle.

According to the fourth aspect of the present invention, by performing the filter calculation processing for each area set in the white line detection area setting, even when a plurality of inclinations of the white line to be emphasized are mixed, The white line segment can be emphasized accurately.

According to the fifth aspect of the invention, the next detection is performed based on both the density value in the white line detection area set by the white line detection area setting means and the inclination of the white line detected by the white line detection means. Even if the road white line is curved by predicting the filter that the edge of the white line you are trying to emphasize most and selecting an edge filter,
The edge image of the road white line can be obtained accurately.

According to the sixth aspect of the present invention, when the white line detection area to be subjected to the next filter calculation is the area closest to the host vehicle, the white line detection area closest to the host vehicle in the previously input image is detected. By setting the filter calculated and derived as the filter for edge extraction, there is no possibility of selecting a filter that is greatly deviated although the screen is changed.

[0018]

BEST MODE FOR CARRYING OUT THE INVENTION Next, embodiments of the present invention will be described. FIG. 1 is a diagram showing a configuration of an embodiment in which white line detection is applied to a vehicle control device. The microcomputer 2
Image processing is performed on the image input from the camera 1 to detect a white line on the road. Then, the shape of the road white line is represented by a mathematical model using a plurality of parameters indicating the road shape and vehicle behavior, and the road shape is recognized by updating the road parameters so that the detection result of the road white line and the white line model match. To do. The camera 1 constitutes the image pickup means in the present invention.

The memory 3 stores captured road images, edge images processed by the microcomputer 2, white line model parameters and the like. The memory 3 constitutes the image storage means in the present invention. The sensor 4 detects the vehicle speed, the steering angle, etc., which represent the behavior of the vehicle, and detects the vehicle 2.
The detection result is output to. The vehicle control unit 5 performs steering, accelerator brake control, and the like based on the output signal from the microcomputer 2. The alarm unit 6 issues an alarm such as a lane departure based on an output signal from the microcomputer 2. Display 7
Is for displaying the road shape and the like as the detection result.

FIG. 2 shows the vehicle mounting position of the camera 1. 2A is a view seen from above the vehicle, and FIG. 2B is a view seen from the side of the vehicle. Camera 1 is vehicle 2
An image of a road in front of the vehicle, which is attached to the center of the vehicle in the vehicle width direction above the front window in 3, is output and the captured image is output to the microcomputer 2.

Next, the flow of processing for detecting a road white line in the microcomputer 2 will be described with reference to the flow charts of FIGS. In step 101, parameters indicating road shape and vehicle behavior (hereinafter simply referred to as road parameters) are initialized. Figure 5
The white line model 2 on the screen coordinate system xy as shown in
2 is represented by a mathematical expression as follows using road parameters. x = (a + ie) (y−d) + b / (y−d) + c (1) In the formula (1), a to e are road parameters, and if the height of the camera 1 from the road surface is constant, the roads are The parameters represent the following road and white line shapes or vehicle behavior. a is the vehicle position of the vehicle in the driving lane,
b is the curvature of the road, c is the yaw angle of the host vehicle (optical axis of the camera 1) with respect to the road, d is the pitch angle of the host vehicle (optical axis of the camera 1) with respect to the road, and e is the lane width of the road. , Respectively. Further, c includes the pan angle of the mounting angles of the host vehicle and the camera 1, and d includes the tilt angle of the mounting angles of the host vehicle and the camera 1. Further, since the shapes of roads and white lines and vehicle behavior are unknown in the initial state, a value corresponding to, for example, the median value is set as an initial value for each road parameter. That is, the lane center is set at the position a in the lane, and the pan angle is set at the yaw angle c with respect to the lane. Further, the tilt angle in the stopped state is set to the pitch angle d with respect to the lane, and the lane width of the expressway shown in the Road Structure Ordinance is set to the lane width e.

In step 102, the image taken by the camera 1 is input to the microcomputer 2.
In step 103, the image input to the microcomputer 2 is stored in the memory 3.

In step 104, the white line detection areas 10 to 21 are set as shown in FIG. The position for setting the white line detection area is the white line model 2 expressed by the equation (1).
Set so that 2 becomes the center of the area. First, for example, the white line detection area 10 is set at a position closest to the host vehicle on the white line model 22, and after the processes from step 105 to step 109 described later are completed, the process returns from step 110 to step 104 to proceed to the next white line. The detection area 11 or the white line detection area 16 is set. That is, the white line detection area 10
From the white line detection area 15 to the white line detection areas 10, 11,
After setting 12 ... 15 in order, the white line detection area 16 to the white line detection area 21 on the right side in the figure may be set in order as white line detection areas 16, 17, 18 ,. After setting the detection area 10, the white line detection area 16 may be set, and then the white line detection area 11 may be set.
The setting order for setting the white line detection area does not matter to the left and right of the left and right white lines, but the white line detection areas are sequentially set from the side closer to the vehicle to the side farther from the white line to be detected. I shall. By repeating steps 104 to 110 in this manner, a total of twelve white line detection areas are set on the left and right road white lines 8 by six.

In the initial state, it is expected that there will be a large difference between the white line model in which the initial values are set for the road parameters and the actual road white line on the screen. Therefore, the white line detection area should be set as large as possible. It is better to do. If the road parameters have been updated in the previous process, it is considered that the difference between the actual road white line and the white line model is small, so it is better to set a small area as much as possible and to detect other than the white line. Is less likely to occur and the processing speed can be improved.

In step 105, the input image stored in the memory 3 is subjected to a filter operation using the set filter to create an edge image, and the result is stored in the memory 3. The filter used here is
For example, the SOBEL filter is set in the initial state, and the filter updated in step 109 described later is used in the subsequent processing.

In step 106, white line candidates are detected in the white line detection area. Edge images extracted in the white line detection area 10 and the white line detection area 16 in FIG. 5 are shown in FIGS. 6 (a) and 6 (b). The white line candidates are detected by selecting, for all the line segments L formed by connecting one point of the upper bottom and one point of the lower bottom of the white line detection area, the pixels whose density of pixels on the line segment is a predetermined value or more as white line candidates. To do. Further, among all the line segments L, the line segment L having the most white line candidates is set as the detection straight line in the area.

In step 107, it is judged whether or not the number of white line candidates on the detected straight line is equal to or larger than a predetermined value. If the number of pixels having the density equal to or higher than the predetermined value on the detection straight line is smaller than the predetermined ratio to the length of the detection straight line, it is considered that the white line is not detected, and the process proceeds to step 110. If it is higher than the predetermined ratio, the process proceeds to step 108.

In step 108, if the number of white line candidates on the detected straight line is greater than a predetermined ratio, the detected straight line is temporarily regarded as a white line. Then, the inclination of the white line is output as y / x. For example, in the white line detection area in which the length of the detection straight line is 15 pixels and the number of white line candidates is 1/2 or more, that is, if a white line is detected if 8 pixels or more are detected, the white line candidates on the detection straight line are detected. When the number is 7 pixels or less, it is assumed that the white line is not detected in the white line detection area. On the other hand, if the number of white line candidates is 8 pixels or more, it is assumed that the white line is detected, and the inclination of the white line is used as the detection result. The above processing is executed for all the white line detection areas. At this time, the above-mentioned predetermined ratio with respect to the length of the detection line for determining whether or not the white line is detected may be the same for all the white line detection areas, or may be set for each white line detection area. Also, the predetermined value of the density may be the same for all the white line detection areas,
It may be changed for each white line detection area.

In step 109, the above step 108 is performed.
The shape of the filter is updated based on the slope of the white line obtained in step S1, and the coefficient of the filter is updated based on the density value in the white line detection area set in step S104.

Details of the filter update in step 109 will be described with reference to the flowcharts of FIGS. 7 and 8. Here, it is assumed that the inclination takes a negative value when the line segment inside the white line rises to the right in FIG. 5, and takes a positive value when the line segment rises to the left. First, in step 701, the inclination y / x of the white line detected in the white line detection area is referred to. Step 702
Then, it is determined whether y / x ≧ -2 for y / x. When y / x ≧ -2 is true, the process proceeds to step 703. y
When / x ≧ -2 is false, the process proceeds to step 709, and the filter 3 in which the vertical edge component is most emphasized is selected. The filter 3 is, for example, as shown in FIG. In step 703, it is determined whether y / x> -1/2 for y / x. When y / x> -1/2 is true, the process proceeds to step 705. When y / x> -1/2 is false, the process proceeds to step 704, and the filter 2 in which the rising edge component is most emphasized is selected. The filter 2 is, for example, as shown in FIG.

In step 705, for y / x, y
It is determined whether / x> 1/2. When y / x> 1/2 is true, the process proceeds to step 707. When y / x> 1/2 is false, the routine proceeds to step 706, and the filter 1 in which the horizontal edge component is most emphasized is selected. The filter 1 is, for example, as shown in FIG. In step 707,
It is determined whether y / x> 2 for y / x. y / x
When> 2 is true, the process proceeds to step 709. When y / x> 2 is false, the process proceeds to step 708, and the filter 4 in which the edge component rising to the left is most emphasized is selected. This filter 4
Is, for example, as shown in FIG. Step 7
In 09, since y / x <-2 or 2 <y / x, the filter 3 in which the vertical edge component is most emphasized is selected and the process proceeds to the next step 710.

In this embodiment, there are five types of filters, but the number is not limited to this, and by increasing the types of filters, an appropriate filter can be selected according to the fine inclination of the white line. Further, although the filter shape is a 3 × 3 square as shown in FIGS. 9A to 9D, the filter shape is not limited to 3 × 3, and 6 × 3 as shown in FIG. 9E. It can also be used as a filter.

Next, the coefficient of the filter is determined based on the density value of the white line detection area. When the contrast of the image is low, a coefficient that enhances the contrast is selected. First, in step 710, the histogram of the image in the white line detection area is calculated. In step 711, the peak with the higher brightness in the histogram is considered to be the peak of the white line,
Considering the lower peak as the road peak, the difference H between the peak value of the white line and the road surface peak value is calculated.

In steps 712 to 716, the H is compared with predetermined threshold values T1 and T2, and the H
The coefficients 1 to 3 are selected according to the value of. A predetermined threshold T1,
It is assumed that T2 satisfies the relationship of T1 <T2. In step 712, the lower threshold value T1 is compared with the H value. Regarding H, it is determined whether H> T1. H>
If T1 is true, go to step 714. When H> T1 is false, the process proceeds to step 713, and the coefficient 3 for low image contrast is selected. For the coefficient 3, for example, α = 2, and a coefficient that is twice as large as the normal time is set.

In step 714, the higher threshold value T2 is compared with H. For H, it is determined whether H> T2. Step 71 if H> T2 is true
Go to 6. When H> T2 is false, the process proceeds to step 715,
Select a factor of 2 for medium image contrast.
The coefficient 2 is set as, for example, α = 1.5. In step 716, since T2 <H for the H, the coefficient 1 for high image contrast is selected. The coefficient 1 is set, for example, as α = 1. Although the number of types of coefficients is three in this embodiment, the number of types of coefficients is not limited to this, and the set value can be changed.

In step 717, the filter calculated and derived in steps 701 to 716 is replaced with the filter stored in the memory 3 which is used for setting the white line detection area and performing filter calculation next.
The new filter to be updated considers both the slope of the white line and the contrast of the edge image, and is, for example, a product of the filter type and the coefficient. As an example, FIG. 10 shows a filter to be updated when the filter 2 and the coefficient 2 are selected.

On one of the white lines, the filter used for setting the white line detection area and extracting the edge image is derived from the area immediately before the area where the white line detection area is to be set in the white line as seen from the own vehicle. Use a filtered filter. If the white line detection area for edge extraction is the closest to the host vehicle, the filter derived from the previously input image in the area is used.

Returning to the flowcharts of FIGS. 3 and 4, in step 110, it is determined whether or not the setting of the white line detection area is completed. In the example shown in FIG. 5, twelve white line detection areas, six in total, are set on the left and right road white lines 8 in order from the side closer to the host vehicle, but when all the white line detection areas are not set Returns to step 104 and sets the next white line detection area. On the other hand, if all the white line detection areas have been set, the process proceeds to step 111.

In step 111, it is confirmed whether or not the number of temporary white lines detected in all the white line detection areas is equal to or larger than a predetermined value. If the number is less than the predetermined value, it is determined that the road white line is not included in the white line detection area. , Step 120 described later
Go to. On the other hand, if the temporary white line is detected to be equal to or larger than the predetermined value, it is determined that the road white line is detected and the process proceeds to step 112.

In step 112, the amount of deviation between the road white line detected this time and the white line model obtained from the previously input image is calculated. In step 113, the road parameter fluctuation amounts Δa to Δe are calculated based on the deviation amount. As a method of calculating the fluctuation amount, for example, the method disclosed in Japanese Patent Laid-Open No. 8-5388 can be used.

Next, at step 114, the road parameters a to e are calculated based on the calculated road parameter fluctuation amounts Δa to Δe.
To correct. For example, in the case of the white line model shown in Expression (1), the road parameters a to e are corrected by the following expressions. a '
Is the road parameter of the white line model obtained from the previous input image for the correction result a. b ', c',
The same applies to d'and e '. a = a '+ Δa, b = b' + Δb, c = c '+ Δc, d = d' + Δd, e = e '+ Δe, (2)

In step 115, it is confirmed whether the parameter representing the road shape is normal among the road parameters, and if it is not normal, the routine proceeds to step 116, and the parameter representing the road shape road shape is initialized. Formula (1)
In the case of the white line model shown in, the parameter b reflects the road curvature and the parameter e reflects the lane width. Therefore, when the road curvature b estimated from the parameter b becomes a road curvature which cannot be determined by the vehicle behavior detection value by the sensor 4 on the road currently running, the parameter b is initialized. Similarly, when the lane width estimated from the parameter e becomes a lane width that cannot be determined on the road on which the vehicle is currently traveling based on the vehicle behavior detection value detected by the sensor 4, the parameter e is initialized. If the parameter indicating the road shape is normal, the process proceeds to step 117.

Next, at step 117, it is confirmed whether or not the parameter indicating the vehicle behavior is normal. If not, the routine proceeds to step 118, where the parameter indicating the vehicle behavior is initialized. In the case of the white line model shown in Expression (1), the parameter a reflects the position of the host vehicle in the lane, the parameter c reflects the yaw angle with respect to the road, and the parameter d reflects the pitch angle with respect to the road surface. Therefore, when the vehicle position estimated from the parameter a becomes a vehicle position which is not possible on the road currently running, as judged from the estimated value of the road curvature or the vehicle behavior detection value by the sensor 4, the parameter a is initialized. Turn into. Similarly, when the yaw angle estimated by the parameter c becomes an angle that is not possible on the road currently running, as judged from the estimated value of the road curvature or the detected value of the vehicle behavior by the sensor 4, the parameter c is set. initialize. Further, when the pitch angle estimated from the parameter d becomes an angle which is not possible on the road currently traveling, which is judged by the sensor 4 from the vehicle behavior detection value, the parameter d is initialized. . When the parameter indicating the vehicle behavior is normal, the process proceeds to step 119.

In step 119, the corrected road parameters a to e are updated as the road parameters of the new white line model and stored in the memory 3. Finally, in step 120, a white line model is calculated from the updated road parameters. If the number of white lines detected is less than or equal to the predetermined value in step 111, the white line model is calculated from the previous estimated values of road parameters. Information from the road parameters is output to the vehicle control unit 5, the alarm unit 6, the display unit 7 and the like as needed. When the above process is completed, the process returns to step 102 and the above process is repeated.

In this embodiment, step 104 constitutes the white line detection area setting means of the present invention, step 105 constitutes the edge image generation means, and step 108 constitutes the white line detection means. Further, step 109 constitutes edge filter updating means, and step 119 constitutes parameter updating means.

Since the present embodiment is constructed as described above and the optimum filter is obtained according to the inclination of the white line, the influence of the detection error due to the curve and the own vehicle traveling to the left and right in the lane of travel. In the case of being present, the influence of the detection error due to the appearance of the white line can be eliminated. For example, in FIG.
When a white line on a road with a curving direction is detected, a plurality of slopes of the white line to be emphasized are mixed (for example, an area near the vehicle on the left white line in the figure is a right upward 4
The line segment of 5 degrees, the slope of the white line in the area far from the vehicle on the left side white line is close to the horizontal line segment, and the slope of the white line in the area on the right side white line in the figure near the vehicle is the 45 degree line segment that rises to the left and the right side white line. Even if the white line in the area distant from the vehicle has an inclination close to vertical, the white line can be accurately emphasized.

By setting the area based on the white line model when setting the white line detection area, it is possible to catch the road white line in a small area, and the processing speed of the edge filter calculation becomes faster as the white line detection area becomes smaller. Since it is less likely to detect a noise component, an accurate road white line can be detected.

When setting the white line detection area, the white line detection area may be set at a position offset in the direction of change of the white line model from the past change state of the white line model. The white line detection area is set on the memory 3, but if the microcomputer has a cache memory or the like, that memory may be used.

In FIG. 5, the white line detection area is set to the inside edge portion of the white line when viewed from the host vehicle, but when the white line model is set to the outside of the white line when viewed from the host vehicle,
The area is set so that the white line model becomes the center of the white line detection area.

When setting the road parameter in the initial state, the initial value of the road parameter may be set on the basis of the value representing the behavior of the vehicle detected by the sensor 4.
For example, in the initial state, when the steering wheel is steered to the right or left, it is determined that the vehicle is traveling on a road having a curvature corresponding to the steering angle, and the parameter b is set to a value corresponding to the steering angle. You may.

[Brief description of drawings]

FIG. 1 is a diagram showing an embodiment of the present invention.

FIG. 2 is a diagram showing a mounting position of a camera.

FIG. 3 is a flowchart showing road white line detection processing.

FIG. 4 is a flowchart showing road white line detection processing.

FIG. 5 is a diagram showing a white line detection area.

FIG. 6 is a diagram showing a detected white line in a white line detection area.

FIG. 7 is a flowchart showing edge filter update processing.

FIG. 8 is a flowchart showing an edge filter update process.

FIG. 9 is a diagram illustrating an example of an edge filter.

FIG. 10 is a diagram showing an example of an updated filter.

FIG. 11 is a diagram showing a curved road.

[Explanation of symbols]

1 camera 2 microcomputer 3 memory 4 sensors 5 Vehicle control unit 6 Alarm section 7 Display 8 white line 9 vanishing point 10-21 White line detection area 22 White line model 23 vehicles

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ identification code FI theme code (reference) G06T 7/60 200 G06T 7/60 200J G08G 1/16 G08G 1/16 C F term (reference) 5B057 AA16 CA08 CE03 CE06 CH01 CH11 DA15 DA16 DB09 DC08 DC16 DC22 DC33 5H180 AA01 CC04 CC24 LL01 LL02 LL07 LL08 LL15 5L096 AA06 BA04 CA02 CA24 DA01 DA03 FA06 FA14 GA55 HA07

Claims (6)

[Claims] 1. An image pickup means for picking up a predetermined area around an own vehicle mounted on a vehicle, an image storage means for storing an image obtained from the image pickup means, and an image stored by the image storage means. On the other hand, a white line detection area setting means for setting an area for detecting the white line of the traveling lane of the own vehicle,
Edge image generation means for generating an edge image by performing edge extraction by a filter operation on the area set by the white line detection area setting means, and a white line for white line detection from the edge image extracted in the white line detection area. It has a detection means and an edge filter update means for updating a filter coefficient used in the filter calculation in the edge image generation means, and the edge filter update means uses the density value in the white line detection area and the white line detection means. A white line detection device characterized in that a filter is updated based on both of the detected inclinations of the white line. 2. A white line model is obtained by expressing the shape of a road white line by a mathematical expression model using a plurality of parameters representing the road shape and vehicle behavior, and the white line model and the result detected by the white line detection means are obtained. The white line detection device according to claim 1, further comprising a parameter updating unit that updates a plurality of parameters of the white line model so as to match each other. 3. The white line detection area setting unit, which is closer to the image stored by the image storage unit than the own vehicle, has a plurality of regions for detecting a white line in the traveling lane of the own vehicle. The white line detection device according to claim 1 or 2, wherein the white line detection device is sequentially set to a position farther from. 4. The white line detection device according to claim 3, wherein the edge image generation means performs a filter calculation process for each area set by the white line detection area setting means. 5. The edge filter updating means, based on both the density value in the white line detection area set by the white line detection area setting means and the slope of the white line detected by the white line detection means, determines the next white line. The white line detecting apparatus according to claim 3 or 4, wherein the edge filter is updated by predicting a filter in which the edge of the white line in the detection area is most emphasized. 6. The edge filter updating means, when the white line detection area to be subjected to the next filter calculation is the area closest to the host vehicle, the white line detection area closest to the host vehicle on the previously input image. The white line detection device according to claim 3, 4 or 5, wherein the filter used for the region calculated and derived in step S3 is updated as a filter used in the filter calculation.