JP7007856B2 - White line recognition device for vehicles - Google Patents

Description

The present invention relates to a vehicle white line recognition device that recognizes a white line based on an image captured by an image pickup means such as an in-vehicle camera.

In recent years, in order to improve the safety and convenience of vehicles, driving support devices that actively support the driving operation of drivers have been developed. As a driving support device of this type, for example, an in-vehicle camera is used to image the environment outside the vehicle, and from the captured image, a three-dimensional object such as a white line laid on the road surface or a preceding vehicle on the road surface is recognized, and these recognition information is used. Based on this, those that perform various driving support controls such as vehicle speed control, steering support control, and automatic driving are known.

As a vehicle white line recognition device used for such a driving support device, for example, in Patent Document 1, a vehicle is provided for each of a plurality of search lines extending horizontally with respect to a search area set on a captured image. By investigating the change in brightness from the inside to the outside in the horizontal direction, edge points whose brightness changes from dark to light by a predetermined value or more are detected as white line candidate points, and the minimum square method is used for the detected white line candidate points. A technique for recognizing a white line by calculating a quadratic curve or the like is disclosed.

2012-22574 Gazette

However, as in the technique disclosed in Patent Document 1 described above, in a process of only detecting an edge point whose brightness changes from dark to light by a predetermined value or more as a white line candidate point, an image of light projected on the road surface is obtained. An edge point may be erroneously detected as a white line candidate point.

For example, when a structure such as a guardrail or a fence is provided along the side of the road, the sunlight projected on the road surface from the gap of the structure may form a high-intensity band-shaped image. When such an image of light extends along the white line in the search area, the edge point of the image may be erroneously recognized as a white line candidate point.

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a vehicle white line recognition device capable of performing appropriate white line recognition by eliminating the influence of an image of light projected on a road surface.

In the vehicle white line recognition device according to one aspect of the present invention, the brightness changes predeterminedly on the image pickup means for capturing the image of the vehicle exterior environment in front of the own vehicle and the search line extending in the horizontal direction of the image captured by the image pickup means. An edge point detecting means for detecting an edge point and extracting the edge point whose luminance changes from dark to bright as a white line start point and extracting the edge point whose luminance changes from bright to dark as a white line end point. , The luminance difference between the white line estimation section estimated based on the white line start point and the white line end point and the road surface other than the white line estimation section, and the luminance transition between the white line estimation section and the road surface. The edge point verification means includes an edge point verification means for verifying whether or not the white line start point and the white line end point are edge points corresponding to the actual white line based on the length of the transition section. It is evaluated that the longer the distance from the own vehicle to the white line start point, the higher the possibility that the white line start point and the white line end point are edge points corresponding to the actual white line .

Further, in the vehicle white line recognition device according to another aspect of the present invention, the luminance is predetermined on the image pickup means for capturing the image of the vehicle exterior environment in front of the own vehicle and the search line extending in the horizontal direction of the image captured by the image pickup means. Edge inspection that detects the edge point that changes to and extracts the edge point whose brightness changes from dark to bright as the white line start point, and extracts the edge point whose brightness changes from bright to dark as the white line end point. The white line start point and the white line end point are the actual white line start point and the white line end point based on the luminance difference between the white line estimation section estimated based on the white line start point and the white line end point and the road surface other than the white line estimation section. The edge point verification means includes an edge point verification means for verifying whether or not the edge point corresponds to the white line, and the edge point verification means increases the longer the distance from the own vehicle to the white line start point, the more the white line start point and the white line start point. It is evaluated that there is a high possibility that the end point of the white line is an edge point corresponding to the actual white line .

Further, in the vehicle white line recognition device according to another aspect of the present invention, the luminance is predetermined on the image pickup means for capturing the image of the vehicle exterior environment in front of the own vehicle and the search line extending in the horizontal direction of the image captured by the image pickup means. Edge inspection that detects the edge point that changes to and extracts the edge point whose brightness changes from dark to bright as the white line start point, and extracts the edge point whose brightness changes from bright to dark as the white line end point. The white line start is based on the length of the output means and the transition section in which the brightness transitions between the white line estimation section estimated based on the white line start point and the white line end point and the road surface other than the white line estimation section. The edge point verification means includes an edge point verification means for verifying whether or not the point and the white line end point are edge points corresponding to the actual white line, and the edge point verification means is a distance from the own vehicle to the white line start point. Is evaluated as having a higher possibility that the white line start point and the white line end point are edge points corresponding to the actual white line .

According to the vehicle white line recognition device of the present invention, it is possible to eliminate the influence of the image of the light projected on the road surface and perform appropriate white line recognition.

Schematic configuration diagram of a vehicle driving support device Explanatory drawing schematically showing an example of a reference image that captures the environment outside the vehicle Explanatory drawing which shows the white line candidate point detected from FIG. Explanatory diagram showing the relationship between the distribution of luminance and the differential value of luminance in the image search direction. Explanatory drawing showing the luminance characteristics of the white line and the image of light in the near and far directions of the own vehicle. Explanatory diagram showing white line candidate points projected in real space (A) is an explanatory diagram showing the amount of movement of the own vehicle due to running, and (b) is an explanatory diagram showing the amount of movement of the white line candidate point detected in the front frame relative to the own vehicle. Explanatory diagram showing the white line approximation line and the white line search area calculated using the white line candidate points. Flowchart showing white line recognition routine White line Flowchart showing the start point verification subroutine A map showing the relationship between the distance to the white line candidate point and the evaluation score A map showing the relationship between the brightness difference between the target area and the road surface and the evaluation score A map showing the relationship between the length of the inner transition region and the evaluation score A map showing the relationship between the length of the outer fiber region and the evaluation score A map showing the relationship between the brightness difference between the maximum brightness and the minimum brightness of the inner transition region and the evaluation score. A map showing the relationship between the maximum brightness of the outer fiber region, the brightness difference between the minimum device, and the evaluation score. A map showing the relationship between the brightness value of the target area and the evaluation score

Hereinafter, embodiments of the present invention will be described with reference to the drawings. The drawings relate to one embodiment of the present invention, FIG. 1 is a schematic configuration diagram of a vehicle driving support device, FIG. 2 is an explanatory diagram schematically showing an example of a reference image of an image of the environment outside the vehicle, and FIG. 3 is from FIG. An explanatory diagram showing the detected white line candidate points, FIG. 4 is an explanatory diagram showing the relationship between the distribution of luminance and the differential value of luminance in the image search direction, and FIG. 5 is an image of white lines and light in the near and far directions of the own vehicle. FIG. 6 is an explanatory diagram showing the luminance characteristics of the above, FIG. 6 is an explanatory diagram showing a white line candidate point projected in the real space, FIG. 7 (a) is an explanatory diagram showing the amount of movement of the own vehicle by traveling, and (b) is an explanatory diagram in front. An explanatory diagram showing the relative movement amount of the white line candidate points detected in the frame with respect to the own vehicle, FIG. 8 is an explanatory diagram showing a white line approximation line and a white line search area calculated using the white line candidate points, and FIG. 9 is a white line recognition. A flowchart showing the routine, FIG. 10 is a flowchart showing the white line start point verification subroutine, FIG. 11 is a map showing the relationship between the distance to the white line candidate point and the evaluation score, and FIG. 12 is the brightness difference between the target area and the road surface and the evaluation score. 13 is a map showing the relationship between the length of the inner transition region and the evaluation score, FIG. 14 is a map showing the relationship between the length of the outer fiber region and the evaluation score, and FIG. 15 is the map showing the relationship between the length of the inner transition region and the evaluation score. A map showing the relationship between the brightness difference between the maximum brightness and the minimum brightness and the evaluation score, FIG. 16 is a map showing the relationship between the maximum brightness in the outer fiber region and the brightness difference between the minimum device and the evaluation score, and FIG. 17 is a map showing the relationship between the evaluation score and the maximum brightness in the outer fiber region. It is a map showing the relationship between the luminance value and the evaluation score.

In FIG. 1, reference numeral 1 is a vehicle such as an automobile (own vehicle), and the vehicle 1 is equipped with a driving support device 2. The driving support device 2 includes, for example, a stereo camera 3 as an image pickup means, a stereo image recognition device 4, a control unit 5, and the like. Of these configurations, in the present embodiment, a vehicle exterior monitoring device that recognizes the vehicle exterior environment in front of the own vehicle is configured mainly by the stereo camera 3 and the stereo image recognition device 4.

Further, the own vehicle 1 is connected to a vehicle speed sensor 11 that detects the own vehicle speed V, a yaw rate sensor 12 that detects the yaw rate γ, a main switch 13 that switches ON / OFF of each function of driving support control, and a steering wheel. A steering angle sensor 14 that is opposed to the steering shaft and detects the steering angle θst, an accelerator opening sensor 15 that detects the accelerator pedal depression amount (accelerator opening) θacc by the driver, and the like are provided.

The stereo camera 3 is composed of a set of CCD cameras using a solid-state image pickup device such as a charge-coupled device (CCD) as a stereo optical system. These left and right CCD cameras are attached to the front of the ceiling in the vehicle interior at regular intervals, stereo-image an object outside the vehicle from different viewpoints, and output image data to the stereo image recognition device 4. In the following description, one of the stereo-captured images (for example, the image on the right side) is referred to as a reference image, and the other image (for example, the image on the left side) is referred to as a comparative image.

The stereo image recognition device 4 performs, for example, various image processings shown below on the captured reference image and comparative image. That is, the stereo image recognition device first sequentially extracts a small area of, for example, 4 × 4 pixels from the reference image, compares the luminance or color pattern of each small area with the comparison image, finds the corresponding area, and uses the reference. Find the distance distribution over the entire image. Further, the stereo image recognition device 4 examines the brightness difference between each pixel on the reference image and the adjacent pixel, extracts the pixel whose brightness difference exceeds the threshold value as an edge point, and extracts the extracted pixel (edge). By adding the distance information to the points), a distance image (a distribution image of edge points having the distance information) is generated. Then, the stereo image recognition device 4 performs a known grouping process on the generated distance image, and compares it with a three-dimensional frame (window) stored in advance, so that the white line and the side wall in front of the vehicle are used. , Recognize three-dimensional objects, etc.

Here, the white line to be recognized in the present embodiment is, for example, a single lane marking line or a multiplex line (double line or the like) in which a line-of-sight guide line or the like is provided inside the lane marking line. It is a general term for lane markers that extend on the road and divide the own vehicle traveling lane, and the form of each line is not limited to a solid line, a broken line, or the like, and further includes a yellow line or the like. Further, in the white line recognition of the present embodiment, even if the white line actually existing on the road is a double white line or the like, it is recognized by approximating each of the left and right sides by a single approximation formula (a straight line or a curve approximation formula of a first order or higher). It shall be.

By the way, there are various variations of the white line laid on the actual road as described above, and the shape of the white line changes with the branching or merging of the traveling path. In addition, there are various noises such as road surface repair marks, puddles, snow, and images of light on the road. Therefore, it is difficult to recognize the white line accurately in all the scenes by the uniform processing. Therefore, the stereo image recognition device 4 performs white line recognition using various methods without relying only on white line recognition using pattern matching based on the distance image as described above, and comprehensively determines these recognition results. And recognize the final white line.

As one of such white line recognition, the stereo image recognition device 4 recognizes the left and right white lines based on the horizontal luminance change on one of the images (for example, the reference image shown in FIG. 2) captured in front of the vehicle. I do.

Specifically, for example, as shown in FIG. 3, the stereo image recognition device 4 sets left and right white line search areas Ad (Adl, Adr) on the image, and horizontally in each white line search area Ad. The brightness change is examined from the inside to the outside in the vehicle width direction for each search line l value string consisting of a plurality of extending pixel strings. Then, in the stereo image recognition device 4, for example, the luminance of the pixel outside in the vehicle width direction is relatively high with respect to the luminance of the inner pixel, and the differential value of the luminance indicating the amount of change is the setting threshold value on the + side. The brightness of the edge points equal to or higher than (dBth) and the brightness of the outer pixel in the vehicle width direction is relatively low with respect to the brightness of the inner pixel, and the amount of change is shown, and the differential value of the brightness is the set threshold value on the-side. -DBth) or less, the pair with the edge point is extracted as the white line start point and the white line end point (see FIGS. 4A and 4B).

Then, the stereo image recognition device 4 sets the white line start point located on the outermost side in the vehicle width direction as the white line candidate point Pd (left side white line candidate point Pdl, right side) on each search line l in each white line search area Ad. It is detected as a white line candidate point Pdr) and projected onto the real space (see FIG. 6).

Then, the stereo image recognition device 4 obtains the white line approximation lines Ll and Lr representing the left and right white lines based on the respective point sequences of the left and right white line candidate points Pdl and Pdr (see FIG. 8).

Here, in order to prevent erroneous detection of an edge point such as light projected on the road surface as a white line candidate point Pd in the white line recognition as described above, the stereo image recognition device 4 detects the white line start point and the white line. Verify whether the end point corresponds to the actual white line.

Specifically, the stereo image recognition device 4 has the luminance of the white line estimation section estimated based on the white line start point and the white line end point on the search line l and the road surface other than the white line estimation section in the search area Ad. Verify whether the white line start point and white line end point are edge points corresponding to the actual white line based on the difference and the length of the transition section where the luminance transitions between the white line estimation section and the road surface. ..

More specifically, the stereo image recognition device 4 has, for example, the distance from the own vehicle 1 to the white line start point, the difference between the average brightness of the white line estimation section and the average brightness other than the white line estimation section, and before and after the white line start point. The length of the first transition section where the brightness transitions from dark to light, the length of the second transition section where the brightness changes from light to dark before and after the white line end point, and the maximum and minimum brightness in the first transition section. Evaluation values S1 to S7 are calculated using the difference, the difference between the maximum brightness and the minimum brightness in the second transition section, and the brightness value in the white line estimation section as parameters, respectively, and the evaluation values calculated from these evaluation values S1 to S7. When S0 is less than the set threshold Sth, it is determined that the white line corresponding to the white line start point and the white line end point is likely to exist, and when the evaluation value S0 is equal to or more than the set threshold Sth, the white line start point and the white line end point are determined. It is judged that there is a high possibility that the white line corresponding to the point does not actually exist (there is a high possibility that it is an image of light, etc.).

In the present embodiment, the evaluation values S1 to S7, for example, are calculated by the following equations (1) to (7). That is,
S1 = (distance from own vehicle 1 to the start point of the white line) x k1 ... (1)
S2 = (difference between the average brightness of the white line estimation section and the average brightness other than the white line estimation section) × k2… (2)
S3 = (length of first transition section) x k3 ... (3)
S4 = (length of second transition section) x k4 ... (4)
S5 = (difference between maximum luminance and minimum luminance in the first transition section) × k5 ... (5)
S6 = (difference between maximum brightness and minimum brightness in the second transition section) × k6 ... (6)
S7 = (luminance value in the white line estimation section) × k7 ... (7)

Further, the evaluation value S0 is calculated by, for example, the following equation (8). That is,
S0 = S1 + S2 + S3 + S4 + S5 + S6 + S7 ... (8)

Here, in the equations (1) to (7), k1 to k7 are preset coefficients, and are set by using, for example, a statistical method such as "t-test". These coefficients k1 to k8 are set by statistics based on the white lines captured in each scene and the feature amount of the light image, and are evaluated values when the white line estimation area corresponds to the actual white lines. S0 is set to a value that converges to "0", and the evaluation value S0 converges to "1" when the white line estimation region corresponds to an image of light or the like.

That is, for example, when comparing the white line and the image of light as shown in FIG. 5, the difference between the average luminance in the white line estimation section and the average luminance in the non-white line estimation section tends to be larger in the white line. .. Therefore, for example, the coefficient k2 is set so that the evaluation value S2 becomes smaller as the difference between the average luminance in the white line estimation section and the average luminance other than the white line estimation section becomes larger (see, for example, FIG. 12).

Further, for example, as shown in FIG. 5, when the white line and the image of light are compared, the length of the transition region (the length of the first and second transition regions) tends to be shorter in the white line. .. Therefore, for example, the coefficients k3 and k4 are set so that the evaluation values S3 and S4 become smaller as the length of the transition region becomes shorter (see, for example, FIGS. 13 and 14).

Further, for example, as shown in FIG. 5, when the white line and the image of light are compared, the difference in luminance between the maximum luminance and the minimum luminance in the transition region tends to be larger in the white line. Therefore, for example, the coefficients k5 and k6 are set so that the evaluation values S5 and S6 become smaller as the difference between the maximum luminance and the minimum luminance in the transition region increases (see, for example, FIGS. 15 and 16).

Further, for example, as shown in FIG. 5, when the white line and the image of light are compared, the white line tends to have a higher luminance in the white line estimation region. Therefore, for example, the coefficient k7 is set so that the evaluation value S7 becomes smaller as the brightness of the white line estimation region increases (see, for example, FIG. 17).

It should be noted that, in the distance of the own vehicle 1, even if the image of light is mistakenly recognized as a white line, the distance from the own vehicle 1 to the start point of the white line increases as the distance from the own vehicle 1 increases, based on the idea that the traveling control and the like are not significantly affected. , The coefficient k1 is set so that the evaluation value S1 becomes smaller (see, for example, FIG. 11).

As described above, in the present embodiment, the stereo image recognition device 4 realizes each function as an edge point detecting means and an edge point verification means.

The driving environment information in front of the own vehicle 1 recognized by the stereo image recognition device 4 is input to the control unit 5. Further, the control unit 5 is input with the vehicle speed V from the vehicle speed sensor 11, the yaw rate γ from the yaw rate sensor 12, etc. as the traveling information of the own vehicle 1, and is input from the main switch 13 as the operation input information by the driver. An operation signal, a steering angle θst from the steering angle sensor 14, an accelerator opening θacc from the accelerator opening sensor 15, and the like are input.

Then, for example, when the driver is instructed to execute the ACC (Adaptive Cruise Control) function, which is one of the driving support control functions, through the operation of the main switch 13, the control unit 5 recognizes the stereo image recognition device 4. The direction of the preceding vehicle is read, and whether or not the preceding vehicle to be followed is running on the own vehicle driving path is identified.

As a result, when the preceding vehicle to be followed is not detected, the constant speed running control that maintains the vehicle speed V of the own vehicle 1 at the set vehicle speed set by the driver through the open / close control of the throttle valve 16 (engine output control). To execute.

On the other hand, when the preceding vehicle which is the tracking target vehicle is detected and the vehicle speed of the preceding vehicle is equal to or less than the set vehicle speed, the following traveling control is performed in a state where the vehicle-to-vehicle distance to the preceding vehicle is converged to the target vehicle-to-vehicle distance. Will be executed. At the time of this follow-up travel control, the control unit 5 basically converges the inter-vehicle distance with the preceding vehicle to the target inter-vehicle distance through the open / close control (engine output control) of the throttle valve 16. Further, when it is determined that sufficient deceleration cannot be obtained only by controlling the throttle valve 16 due to sudden deceleration of the preceding vehicle or the like, the control unit 5 controls the output hydraulic pressure from the active booster 17 (automatic intervention of the brake). Control) is also used to converge the inter-vehicle distance to the target inter-vehicle distance.

Further, when the driver is instructed to execute the lane departure prevention function, which is one of the driving support control functions, through the operation of the main switch 13, the control unit 5 is, for example, the left and right white lines defining the own vehicle traveling lane. The warning determination line is set based on (the approximate line of the white line recognized by the stereo image recognition device 4), and the own vehicle traveling route is estimated based on the vehicle speed V and the yaw rate γ of the own vehicle 1. Then, when the control unit 5 determines, for example, that the own vehicle traveling path crosses the warning determination line on either the left or right within the set distance in front of the own vehicle (for example, 10 to 16 [m]). It is determined that the own vehicle 1 is likely to deviate from the current own vehicle traveling lane, and a lane departure warning is issued.

Next, the white line recognition process performed in the stereo image recognition device 4 described above will be described according to the flowchart of the white line recognition routine shown in FIG.

This routine is repeatedly executed every set time, and when the routine starts, the stereo image recognition device 4 first sets the left and right white line search areas Adl and Adr on the reference image in step S101. In these white line search areas Adl and Adr, for example, the left and right white line approximation lines Ll and Lr recognized in the previous frame are offset by preset offset amounts ΔE on the inside and outside in the vehicle width direction in the real space. It is set by converting the region formed by (see FIG. 8) into the coordinates on the reference image.

In the following step S102, the stereo image recognition device 4 checks whether or not the search procedure for the white line start point (and the white line end point) has been performed for all the search lines l in the white line search areas Adl and Adr.

Then, when it is determined in step S102 that the search process for the white line start point has been performed for all the search lines l, the stereo image recognition device 4 proceeds to step S106.

On the other hand, if it is determined in step S102 that the search process for the white line start point has not been performed for all the search lines l, the stereo image recognition device 4 proceeds to step S103 and unprocessed in the white line search areas Adl and Adr. The new search line l of is extracted.

Then, when the process proceeds from step S103 to step S104, the stereo image recognition device 4 performs a search process for the white line start point (and the white line end point) with respect to the extracted search line l.

That is, the stereo image recognition device 4 sequentially searches the pixels on the search line l from the inside to the outside in the vehicle width direction, and the brightness of the pixels outside in the vehicle width direction is relative to the brightness of the inner pixels. The brightness of the edge points that are high and the differential value of the brightness indicating the amount of change is equal to or higher than the setting threshold value (dBth) on the + side, and the brightness of the pixels outside the vehicle width direction are relatively low with respect to the brightness of the inner pixels. And, the pair with the edge point where the differential value of the luminance indicating the change amount is equal to or less than the set threshold value (−dBth) on the − side is extracted as the white line start point and the white line end point.

Then, when the process proceeds from step S104 to step S105, the stereo image recognition device 4 verifies the detected white line start point (and white line end point).

The verification of the white line start point is executed, for example, according to the flowchart of the white line start point verification subroutine shown in FIG. 10. When the subroutine is started, the stereo image recognition device 4 first unverifies in step S201. Check if there is a white line start point (and white line end point).

Then, when it is determined in step S201 that there is no unverified white line start point, the stereo image recognition device 4 exits the subroutine.

On the other hand, if it is determined in step S201 that there is an unverified white line start point, the stereo image recognition device 4 proceeds to step S202 and proceeds to step S202, where the unverified new white line start point (more specifically, the white line start point). The pair with the white line end point) is extracted.

Then, when the process proceeds from step S202 to step S203, the stereo image recognition device 4 calculates the evaluation value S0 of the white line start point based on the above equations (1) to (8).

Then, when the process proceeds to step S204, the stereo image recognition device 4 checks whether or not the evaluation value S0 calculated in step S203 is less than the preset threshold value Sth (for example, Sth = 0.5).

Then, when it is determined in step S204 that the evaluation value S0 is less than the threshold value Sth, the stereo image recognition device 4 returns to step S201 as it is.

On the other hand, when it is determined in step S204 that the evaluation value S0 is equal to or higher than the threshold value Sth, the stereo image recognition device 4 causes the currently selected white line start point (and white line end point) to be due to an image of light or the like. It is determined that there is a high possibility that the image was detected, and after excluding the white line start point (and the white line end point), the process returns to step S201.

When such a white line start point search subroutine ends, the stereo image recognition device 4 returns from step S105 to step S102 in the main routine of FIG.

Further, when the process proceeds from step S102 to step S106, the stereo image recognition device 4 calculates the white line approximation line based on the white line start point (verified white line start point) detected by each search line l, and then exits the routine.

In the calculation process of the white line approximation line, the stereo image recognition device 4 first extracts a predetermined white line start point detected on each search line l as a white line candidate point Pd. That is, for example, when the white line start point of 1 is detected on the search line l, the stereo image recognition device 4 extracts the white line start point as the white line candidate point Pd and 2 on the same search line l. When the above white line start point is detected, the white line start point located on the outermost side in the vehicle width direction is extracted as the white line candidate point Pd.

Further, the stereo image recognition device 4 performs, for example, the following processing in order to recognize the white line behind the own vehicle 1 by using the white line candidate point Pold detected before the previous frame.

That is, the stereo image recognition device 4 is, for example, based on the vehicle speed V of the own vehicle 1 and the yaw angle θ obtained from the yaw rate γ from the relationship shown in FIG. 7A, per frame of the captured image (the captured image is The movement amounts Δx and Δz of the own vehicle 1 in t) until one frame is updated are calculated using the following equations (9) and (10).
Δx = V × sinθ… (9)
Δz = V × cosθ… (10)

Then, as shown in the following equations (11) and (12), the stereo image recognition device 4 has its own vehicle movement amount Δx, Δz with respect to the white line candidate points Pold (Xold, Yold) detected before the previous frame. After subtracting, the coordinates of the white line candidate point Ppre existing within a predetermined distance behind the own vehicle 1 are calculated by performing coordinate conversion to the vehicle fixed coordinate system (X, Z) in the current frame (FIG. 7). See (b)).
Xpre = (Xold ・ Δx) × cosθ- (Zold ・ Δz) × sinθ… (11)
Zpre = (Xold ・ Δx) × sinθ + (Zold ・ Δz) × cosθ… (12)

Then, the stereo image recognition device 4 is based on a point sequence consisting of the white line candidate point Pd detected in the current frame and the white line candidate point Ppre obtained by performing coordinate conversion of the white line candidate point Pold before the previous frame. The white line approximation lines Ll and Lr are defined by the approximation equations shown in the following equations (13) and (14) using the least squares method (see FIG. 8).
Xl = Al ・ Z ² ＋ Bl ・ Z ＋ Cl… (13)
Xr = Ar ・ Z ² ＋ Br ・ Z ＋ Cr… (14)

In the equations (13) and (14), the secondary coefficient A (Al, Ar) is a parameter indicating the curvature component of the white line, and the linear coefficient B (Bl, Br) is the lane yaw angle component (own vehicle 1). C (Cl, Cr) is a parameter indicating the lane position (horizontal position of the white line with respect to the own vehicle 1).

According to such an embodiment, an edge point where the luminance changes predeterminedly is detected on the search line l, an edge point where the luminance changes from dark to bright is extracted as a white line start point, and the luminance changes from bright to dark. The edge point that changes to is extracted as the white line end point, and the luminance difference between the white line estimation section estimated based on the white line start point and the white line end point and the road surface other than the white line estimation section, and the white line estimation section and the road surface By verifying whether the white line start point and white line end point are edge points corresponding to the actual white line based on the length of the transition section in which the luminance transitions between the two, the light projected on the road surface Appropriate white line recognition can be performed by eliminating the influence of the image.

The present invention is not limited to the embodiments described above, and various modifications and changes can be made, which are also within the technical scope of the present invention.

For example, in the above-described embodiment, an example of calculating the evaluation value S0 using all of the evaluation values S1 to S7 has been described, but the present invention is not limited to this, and the evaluation values S1 to S7 are not limited thereto. It is also possible to calculate the evaluation value S0 using any one or two or more. That is, for example, it is possible to calculate the evaluation value S0 using only the evaluation values S2, S5, and S6 relating to the brightness difference between the white line estimation section and the road surface other than the white line estimation section, or the white line estimation section and the road surface. It is also possible to calculate the evaluation value S0 using the evaluation values S3 and S4 relating to the length of the transition section with. Further, it is also possible to calculate the evaluation value S0 by using other combinations of the evaluation values S1 to S7. In this case, it goes without saying that the coefficients k1 to k7 of each evaluation value S1 to S7 are set by using a statistical method such as "t-test" according to the evaluation value to be adopted.

1 ... Vehicle (own vehicle)
2 ... Driving support device 3 ... Stereo camera (imaging means)
4 ... Stereo image recognition device (edge point detection means, edge point verification means)
5 ... Control unit 11 ... Vehicle speed sensor 12 ... Yaw rate sensor 13 ... Main switch 14 ... Steering angle sensor 15 ... Accelerator opening sensor 16 ... Throttle valve 17 ... Active booster

Claims (6)

An imaging means that captures the environment outside the vehicle in front of the vehicle,
An edge point whose luminance changes predeterminedly is detected on a search line extending in the horizontal direction of the image captured by the imaging means, and the edge point whose luminance changes from dark to bright is extracted as a white line start point and is also extracted. An edge point detecting means for extracting the edge point whose brightness changes from light to dark as a white line end point, and
The brightness difference between the white line estimation section estimated based on the white line start point and the white line end point and the road surface other than the white line estimation section, and the transition in which the brightness transitions between the white line estimation section and the road surface. It is provided with an edge point verification means for verifying whether or not the white line start point and the white line end point are edge points corresponding to the actual white line based on the length of the section .
The edge point verification means evaluates that the longer the distance from the own vehicle to the white line start point, the higher the possibility that the white line start point and the white line end point are edge points corresponding to the actual white line. A characteristic white line recognition device for vehicles. An imaging means that captures the environment outside the vehicle in front of the vehicle,
An edge point whose luminance changes predeterminedly is detected on a search line extending in the horizontal direction of the image captured by the imaging means, and the edge point whose luminance changes from dark to bright is extracted as a white line start point and is also extracted. An edge point detecting means for extracting the edge point whose brightness changes from light to dark as a white line end point, and
The white line start point and the white line end point correspond to the actual white line based on the luminance difference between the white line estimation section estimated based on the white line start point and the white line end point and the road surface other than the white line estimation section. It is equipped with an edge point verification means for verifying whether or not it is an edge point .
The edge point verification means evaluates that the longer the distance from the own vehicle to the white line start point, the higher the possibility that the white line start point and the white line end point are edge points corresponding to the actual white line. A characteristic white line recognition device for vehicles. An imaging means that captures the environment outside the vehicle in front of the vehicle,
An edge point whose luminance changes predeterminedly is detected on a search line extending in the horizontal direction of the image captured by the imaging means, and the edge point whose luminance changes from dark to bright is extracted as a white line start point and is also extracted. An edge point detecting means for extracting the edge point whose brightness changes from light to dark as a white line end point, and
The white line start point and the white line based on the length of the transition section in which the brightness transitions between the white line estimation section estimated based on the white line start point and the white line end point and the road surface other than the white line estimation section. It is equipped with an edge point verification means for verifying whether or not the end point is an edge point corresponding to the actual white line .
The edge point verification means evaluates that the longer the distance from the own vehicle to the white line start point, the higher the possibility that the white line start point and the white line end point are edge points corresponding to the actual white line. A characteristic white line recognition device for vehicles. The edge point verification means evaluates that the smaller the luminance difference between the white line estimation section and the road surface, the lower the possibility that the white line start point and the white line end point are edge points corresponding to the actual white line. The white line recognition device for a vehicle according to claim 1 or 2, characterized in that. The edge point verification means evaluates that the longer the transition section between the white line estimation section and the road surface, the lower the possibility that the white line start point and the white line end point are edge points corresponding to the actual white line. The white line recognition device for a vehicle according to claim 1 or 3, characterized in that. The edge point verification means evaluates that the higher the luminance value of the white line estimation section, the higher the possibility that the white line start point and the white line end point are edge points corresponding to the actual white line. The vehicle white line recognition device according to any one of claims 1 to 5.

Images