JP2003306102A - Passing vehicle detection device, and rear side alarm device for drive supporting system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To prevent mistakes when detecting a passing vehicle. <P>SOLUTION: This invention is for eliminating detection of repeated patterns on a road within a passing vehicle detection area as a candidate for a passing vehicle. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an overtaking vehicle detection device for detecting an overtaking vehicle and a rear side warning device for a driving assistance system.

[0002]

2. Description of the Related Art For the purpose of improving safety while driving a vehicle, there is provided a device for avoiding an accident by evaluating the measurement result of an image input sensor that can be measured during driving and issuing a warning or driving control. Many have been proposed. For example, using a camera (vehicle-mounted camera) installed at the rear of the vehicle, a device that detects another vehicle that is approaching from the rear of the vehicle and overtaking the vehicle and issues a warning to prevent a contact accident. and so on.

An example of such a conventional technique is an overtaking vehicle detection technique using an optical flow. For example, it is shown in a reference document (Japanese Unexamined Patent Publication No. 2000-11133, moving object detection apparatus and method thereof). Such a conventional method has a problem that a repetitive pattern lined up on a straight line from the FOE to the outside in the input image is erroneously detected as an overtaking vehicle. Depending on the running speed of the host vehicle, the time interval for image input, the interval for placement of the repeat pattern, etc., the repeating pattern may appear to be approaching the own vehicle in the input image sequence. In this case, the overtaking vehicle is erroneously detected. The pattern of FIG. 1 is a repeating pattern drawn on the road surface. When such a repeating pattern is taken while the vehicle is moving forward, even though the vehicle is actually moving away from the vehicle, in the input image sequence, the movement toward the vehicle is approached. May appear to do. For example, it is assumed that the images shown in FIGS. 1 and 2 are obtained at times 1 and 2, respectively. In fact, the straight lines A, B, C and D in FIG. 1 are respectively C ′, D ′ and C ′ in FIG.
Although it corresponds to E'and F ', in appearance,
A, B, C and D are A ′, B ′, C ′ and D ′, respectively.
Seems to correspond to. In such a case, the optical flow detection unit detects an optical flow approaching the host vehicle from the repeating pattern. As a result, the overtaking vehicle is erroneously detected. Similarly, an optical flow approaching the host vehicle may be detected from the pole of the fence as shown in FIG. 3, and an overtaking vehicle may be erroneously detected.

[0004]

DISCLOSURE OF THE INVENTION The present invention provides an overtaking vehicle detection device capable of preventing an erroneous detection of an overtaking vehicle and a rear side warning device for a driving support system capable of preventing an erroneous alarm due to an erroneous detection of an overtaking vehicle. The purpose is to do.

[0005]

An overtaking vehicle detection device according to the present invention includes a photographing means for photographing an image behind a vehicle, a white line detecting means for detecting a white line from an image photographed by the photographing means, and First judging means for judging the detection area of the overtaking vehicle based on the white line detected by the white line detecting means, and a movement vector between frames in the detection area of the overtaking vehicle judged by the first judging means. Second judging means for judging an overtaking vehicle candidate based on the above, and detection of an overtaking vehicle judged by the first judging means among the overtaking vehicle candidates judged by the second judging means. And a removing means for removing an overtaking vehicle candidate by a repeated pattern in the area.

The rear side warning device of the driving support system according to the present invention includes a photographing means for photographing the image behind the vehicle, a white line detecting means for detecting a white line from the image photographed by the photographing means, and a white line detecting means. Based on the white line detected by the means, based on the first judging means for judging the detection area of the overtaking vehicle, and the movement vector between frames in the detection area of the overtaking vehicle judged by the first judging means. A second judging means for judging an overtaking vehicle candidate and an overtaking vehicle detection area judged by the first judging means among the overtaking vehicle candidates judged by the second judging means. Removing means for removing an overtaking vehicle candidate based on a repetitive pattern, and third determining means for determining a change in the steering angle of the vehicle or a lane change by a turn signal switch,
When the lane change is judged by the third judging means,
Among the overtaking vehicle candidates judged by the second judging means,
And an alarm generating unit that generates an alarm according to the position of the overtaking vehicle candidate that has not been removed by the removing unit.

The overtaking vehicle detecting device of the present invention detects the image of the rear of the vehicle by the photographing means, the white line detecting means for detecting the white line from the image photographed by the photographing means, and the white line detecting means. The first judging means for judging the detection area of the overtaking vehicle based on the white line and the movement vector between the frames in the detection area of the overtaking vehicle judged by the first judging means. Second judging means for judging a vehicle candidate, and, for each overtaking vehicle candidate judged by the second judging means, an angle of an edge in a small area around the detection position of each movement vector included in the overtaking vehicle candidate. Of the overtaking vehicle candidates are removed from the overtaking vehicle candidates determined by the second determining means, the overtaking vehicle candidates for which the amount of variance of the edge angle histogram of each overtaking vehicle candidate is smaller than a predetermined value. A movement vector amount in the same direction as the extending direction of the white line detected by the white line detecting means is calculated for each overtaking vehicle candidate determined by the first removing means and the second determining means, and this movement is performed. Second removing means for removing an overtaking vehicle candidate having a vector amount smaller than a predetermined value from the overtaking vehicle candidates determined by the second determining means;
For each of the overtaking vehicle candidates determined by the second determining means, the sum of the edge strengths in a small area around the detection position of each movement vector included in the overtaking vehicle candidates is calculated, and the sum of the edge strengths is greater than a predetermined value. And a third removing unit that removes a small overtaking vehicle candidate from the overtaking vehicle candidate determined by the second determining unit.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An overtaking vehicle detection device (rear side warning device) of a driving support system according to an embodiment of the present invention will be described below with reference to the drawings. As shown in FIG. 4, this overtaking vehicle detection device detects the overtaking vehicle 3 from the input image from the rear camera 2 that captures the rear of the vehicle 1 with a predetermined field of view. As shown in FIG.
1, white line detection unit 12, optical flow detection unit 13,
Overtaking vehicle detection unit 14, non-vehicle deletion unit 15, alarm generation unit 1
It is composed of six. This overtaking vehicle detection device is based on the fact that the overtaking vehicle 3 in the input image has a movement called FOE from the point toward the outside of the image.
Is to be detected. Further, a white line 4 is drawn on the road on which the vehicle 1 is traveling and on a straight line in the same direction as the traveling direction of the vehicle 1 to delimit the traffic section of the vehicle.

As shown in FIG. 6, the image input unit 11 captures images from the rear camera 2 for photographing the rear of the vehicle 1 at regular time intervals, and the input image is input to the white line detection unit 12. Output. As shown in FIG. 6, the white line detection unit 12 detects the white lines 4 and 4 on the left and right of the vehicle lane from the input image supplied from the image input unit 11, obtains the intersections thereof as FOE, and the optical flow detection unit 13 Is output to the overtaking vehicle detection unit 14. At this time, the boundary line between white and black at the end of the white line 4 is used as an edge (the part where the luminance change is made is an edge).

As shown in FIG. 7, the optical flow detection unit 13 obtains an optical flow (movement vector) in an area corresponding to an adjacent lane above the white line 4 in the input image supplied from the image input unit 11. . The optical flow detection unit 13 detects a motion in an image between two continuously captured images as an optical flow.

On the basis of the FOE obtained by the white line detection unit 12 and the optical flow from the optical flow detection unit 13, the overtaking vehicle detection unit 14 makes the optical flows going out of the image from the FOE close to each other. Are combined and detected as an overtaking vehicle candidate as shown in FIG. The detection result of the overtaking vehicle candidate is output to the non-vehicle deleting unit 15.

The overtaking vehicle detection unit 14, as shown in, for example, the reference (Japanese Unexamined Patent Application Publication No. 2000-11133, moving object detection apparatus and method), outputs a warning signal from the rear side of the presence of an overtaking vehicle. Can be issued. Therefore, pixels having a texture having a certain intensity or more in a designated area in one frame input from the image input unit 11 in a digitized image are associated with each other in a frame at another time point, and they are the same. It is assumed that a method of evaluating a movement vector (optical flow) between each of the points considered to be the above is used.

The non-vehicle deleting unit 15 is an overtaking vehicle detecting unit 14.
Edge angle dispersion, FO for each overtaking vehicle candidate
Three feature amounts of the projection length from E and the maximum edge strength are calculated, it is determined whether or not the vehicle is a vehicle based on these feature amounts, and when it is determined that the vehicle is not the vehicle, the overtaking vehicle detection unit 14 determines the overtaking vehicle. Is to be deleted from. When the driver (driver) turns on the blinker switch (or when the steering wheel rotates by a predetermined angle or more), the alarm generation unit 16 detects an overtaking vehicle candidate in the direction of the blinker (rotation of the steering wheel). , The driver is notified of the presence of an overtaking vehicle by a buzzer, voice, LED, or the like.

The process of deleting an overtaking vehicle candidate by the non-vehicle deleting unit 15 will be described with reference to the flowchart shown in FIG. That is, the non-vehicle deleting unit 15 is
For each overtaking vehicle candidate by the overtaking vehicle detection unit 14, the projection length Fprjl from the edge angle variances Fvar, FOE
Three feature amounts of en and maximum edge strength Femax are calculated (ST1). As a result, the non-vehicle deleting unit 15 is
Feature quantities Fvar, Fprjlen, Femax and threshold value T
var, Tprjlen, and Temax are compared with each other (ST2), and “Fvar <Tvar, Fprjlen
If any of <Tprjlen, Femax <Temax ”is satisfied, it is determined that the vehicle is not a vehicle and the overtaking vehicle candidate is deleted (ST3). The threshold value is statistically calculated.

However, in order to reduce the possibility of erroneously deleting an overtaking vehicle candidate, when the own vehicle speed is smaller than the threshold value, the overtaking vehicle candidate is continuously detected in the vicinity of the FOE. In the two cases described above, it is not deleted from the overtaking vehicle candidates even when the condition of step 2 is satisfied. This is because if the speed of the vehicle is smaller than the threshold value, that is, if the speed of the vehicle is sufficiently small, the amount of movement of the repetitive pattern in the input image series is also sufficiently small and the accuracy of optical flow detection is improved. In this case, the optical flow for approaching the host vehicle from the repeated pattern is not detected by mistake, and the false detection of the overtaking vehicle does not occur. This prevents it from being deleted from the overtaking vehicle candidates.

The repeating pattern appears from the edge of the input image, that is, from a position close to the own vehicle while the own vehicle is traveling forward. As a result, the overtaking vehicle candidate detected from the repeated pattern tends to suddenly appear near the own vehicle. On the other hand, the overtaking vehicle tends to appear near the FOE, that is, from a distance. Therefore,
When the overtaking vehicle candidates are continuously detected from the vicinity of the FOE, they are not deleted from the overtaking vehicle candidates.

The feature amount Fvar of the edge angle variance and the threshold Temax of the maximum edge strength are respectively the edge angle image θ (x, y) obtained from the input image G (x, y).
Also, it is calculated from the edge strength image E (x, y). Edge angle image θ (x, y) and edge intensity image E
(X, y) is calculated from the horizontal edge image Eh (x, y) and the vertical edge image Ev (x, y) as follows.

Θ (x, y) = arctan (Eh (x,
y) / Ev (x, y)) E (x, y) = | Eh (x, y) | + | Ev (x, y) | where Eh (x, y) and Ev (x, y) Is calculated as follows. Eh (x, y) =-G (x-1, y-1) -2 * G
(X, y-1) -G (x + 1.y-1) + G (x-1,
y + 1) + 2 * G (x, y + 1) + G (x + 1, y +
1) Ev (x, y) =-G (x-1, y-1) + G (x +
1, y-1) -2 * G (x-1, y) + 2 * G (x +
1, y) -G (x-1, y + 1) + G (x + 1, y +
1) The feature values Fvar, Fprjlen, and Femax will be described below. [Feature Fvar of Edge Angle Variance] A histogram of the edge angle θ (x, y) is obtained for a small area around the detection position of each optical flow included in the overtaking vehicle candidate. An edge angle histogram in which the horizontal axis represents the edge angle and the vertical axis represents the number of optical flows is generated. Here, the detection position of the optical flow is the position of the end point of the optical flow as shown in FIG. The small area around the detection position of the optical flow is a small rectangular area around the end point of the optical flow as shown in FIG.

FIG. 11 shows an example of the edge angle histogram. When the edge angle histogram is calculated from the repetitive patterns shown in FIGS. 1 to 3, the edge angle histogram is concentrated on one angle as shown in FIG. 11, and the variance is small. But,
Since the vehicle is composed of edges having various directions, in the case of the vehicle, the edge angle histogram as shown in FIG. 12 is obtained, and the distribution of the histogram tends to be large. Therefore, the variance Fvar of the edge angle histogram is calculated for each overtaking vehicle candidate, and when it is smaller than the threshold value Tvar, it is deleted as not a vehicle. The variance of the edge angle histogram is called the edge angle variance.

[Feature amount Fprjl of projection length from FOE
[en] Like the guardrail pole in FIG. 13, the repeating pattern is short as a straight line, and the edge angle dispersion may not be reduced due to the influence of noise or the like. In order to deal with such a case, the projection length Fprjlen from FOE obtained as follows is calculated. The detection position of each optical flow included in the overtaking vehicle candidate is projected from the FOE to a straight line Lleft parallel to the y axis as shown in FIG.
Find the minimum range of the obtained projections that includes non-zero values,
Let the length LEN-A of the range be Fprjlen. If the overtaking vehicle candidate is on the right side of the input image,
Project to Lright. Straight lines Lleft and Lright
Let t be the same distance from the FOE. Since the vehicle is high from the road surface, Fprjlen calculated for the vehicle tends to have a large value. Therefore, Fprjlen calculated for the overtaking vehicle candidate
Is smaller than the threshold value Tprjlen, the vehicle is deleted because it is not a vehicle.

[Maximum Edge Strength Femax] As shown in FIG. 15, there are cases where shadows are affected by noise and the edge angle dispersion of the repeated pattern does not become small. Therefore,
For a small area ei (i = 1, 2, ..., N) around the detection position of each optical flow included in the overtaking vehicle candidate, the sum of the edge strengths E (x, y) is calculated and ESUM (e
i) (i = 1, 2, ..., N). Here, n is the number of optical flows included in the overtaking vehicle candidate.
The maximum value of ESUM (ei) is the maximum edge strength Femax
And Vehicles tend to have high edge strength, and Femax obtained from vehicles tends to be high. So Fem calculated for the overtaking vehicle candidate
ax is compared with the threshold value Temax, and if it is smaller, it is determined that the vehicle is not a vehicle, and the corresponding overtaking vehicle candidate is deleted.

As described above, the false detection of the overtaking vehicle can be reduced, and the false alarm in the alarm generation unit 16 can be reduced. -By using the edge angle dispersion feature amount, it is possible to delete an overtaking vehicle candidate that is erroneously detected from a repetitive pattern composed of straight lines.・ F
By using the projection length feature amount from the OE, it is possible to delete the overtaking vehicle candidate that is erroneously detected from the repetitive pattern composed of short straight lines. -By using the maximum edge strength feature amount, it becomes possible to delete an overtaking vehicle candidate that has been erroneously detected from a repeating pattern with low edge strength due to illumination.

In addition, when the own vehicle speed is smaller than a threshold value (speed obtained by statistics), the overtaking vehicle candidate is FOE.
If the overtaking vehicle candidates are not deleted when they are continuously detected in the vicinity of, it is possible to reduce the possibility of accidentally deleting the overtaking vehicle candidate that is the vehicle.

[0024]

As described in detail above, according to the present invention, an overtaking vehicle detection device capable of preventing an erroneous detection of an overtaking vehicle and a rear side of a driving support system capable of preventing an erroneous alarm due to an erroneous detection of an overtaking vehicle. It is possible to provide an alarm device.

[Brief description of drawings]

FIG. 1 is a diagram showing an input image having a repeating pattern.

FIG. 2 is a diagram showing an input image having a repeating pattern.

FIG. 3 is a diagram showing an input image having a repeating pattern.

FIG. 4 is a diagram showing a traveling state of the vehicle for explaining the embodiment of the invention.

FIG. 5 is a diagram showing a schematic configuration of an overtaking vehicle detection device for explaining an embodiment of the invention.

FIG. 6 is a diagram showing an input image.

FIG. 7 is a diagram showing an example of optical flow detection.

FIG. 8 is a diagram showing an example of detecting an overtaking vehicle candidate.

FIG. 9 is a flowchart illustrating a process of deleting an overtaking vehicle candidate.

FIG. 10 is a diagram showing a small rectangular area around the end point of the optical flow.

FIG. 11 is a diagram showing an example of an edge angle histogram.

FIG. 12 is a diagram showing an example of an edge angle histogram.

FIG. 13 is a diagram showing an input image having a repeating pattern.

FIG. 14 is a diagram for explaining detection positions and projections of optical flows.

FIG. 15 is a diagram showing an input image having a repeating pattern.

[Explanation of Codes] 1 ... Own vehicle, 2 ... Rear camera, 3 ... Overtaking vehicle, 4 ... White line,
11 ... Image input unit, 12 ... White line detection unit, 13 ... Optical flow detection unit, 14 ... Overtaking vehicle detection unit, 15 ... Non-vehicle deletion unit, 16 ... Warning generation unit.

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ Identification code FI theme code (reference) G06T 1/00 G06T 1/00 330B 7/20 7/20 B 7/60 200 7/60 200J H04N 7 / 18 H04N 7/18 J // G08G 1/16 G08G 1/16 CF Term (reference) 5B057 AA16 BA02 CA08 CA12 CA17 CB08 CB12 CB16 CC01 CE11 DA02 DA07 DA12 DA15 DA16 DB02 DB09 DC02 DC08 DC09 DC23 DC32 5C054 AAFE FC12 FE19 HA30 5H180 AA01 CC04 LL04 LL06 5L096 BA04 CA02 FA03 FA06 FA35 HA04 JA11

Claims (11)

[Claims] 1. A passing means based on a white line detected by a white line detecting means for detecting a white line from a video image taken by the image taking means, and a white line detected by the white line detecting means. A first judging means for judging the detection area of the vehicle, and a second judging means for judging an overtaking vehicle candidate based on a movement vector between frames in the detection area of the overtaking vehicle judged by the first judging means. Determining means, and removing means for removing overtaking vehicle candidates by a repeating pattern in the overtaking vehicle detection area determined by the first determining means from among the overtaking vehicle candidates determined by the second determining means. An overtaking vehicle detection device comprising: 2. The overtaking vehicle detection device according to claim 1, wherein the repetitive pattern is generated by a zebra pattern on a road or a pole of a guardrail. 3. A photographing means for photographing an image behind the vehicle, a white line detecting means for detecting a white line from the image photographed by the photographing means, and an overtaking on the basis of the white line detected by the white line detecting means. A first judging means for judging the detection area of the vehicle, and a second judging means for judging an overtaking vehicle candidate based on a movement vector between frames in the detection area of the overtaking vehicle judged by the first judging means. Determining means, and removing means for removing overtaking vehicle candidates by a repeating pattern in the overtaking vehicle detection area determined by the first determining means from among the overtaking vehicle candidates determined by the second determining means. And a third judging means for judging the lane change by the change of the steering angle of the vehicle or the turn signal switch, and when the lane change is judged by the third judging means,
Among the overtaking vehicle candidates judged by the second judging means,
A rear side alarm device for a driving support system, comprising: an alarm generating unit for generating an alarm according to the position of an overtaking vehicle candidate that has not been removed by the removing unit. 4. A photographing means for photographing an image behind the vehicle, a white line detecting means for detecting a white line from the image photographed by the photographing means, and an overtaking on the basis of the white line detected by the white line detecting means. A first judging means for judging the detection area of the vehicle, and a second judging means for judging an overtaking vehicle candidate based on a movement vector between frames in the detection area of the overtaking vehicle judged by the first judging means. For each of the overtaking vehicle candidates determined by the determining means and the second determining means, a histogram of edge angles in a small area around the detection position of each movement vector included in the overtaking vehicle candidates is obtained to obtain each overtaking vehicle. A removal means for removing an overtaking vehicle candidate having a variance amount of the edge angle histogram of the vehicle candidate smaller than a predetermined value from the overtaking vehicle candidate determined by the second determining means. Overtaking vehicle detection apparatus according to claim. 5. The overtaking vehicle detection device according to claim 4, wherein the edge is a change in brightness. 6. An overtaking device for taking an image of an image behind a vehicle, a white line detecting device for detecting a white line from an image taken by the image taking device, and an overtaking on the basis of the white line detected by the white line detecting device. A first judging means for judging the detection area of the vehicle, and a second judging means for judging an overtaking vehicle candidate based on a movement vector between frames in the detection area of the overtaking vehicle judged by the first judging means. The moving vector amount in the same direction as the extending direction of the white line detected by the white line detecting unit is calculated for each of the overtaking vehicle candidates judged by the judging unit and the second judging unit. An overtaking vehicle detection device comprising: a removing unit that removes an overtaking vehicle candidate having a number smaller than a predetermined value from the overtaking vehicle candidates determined by the second determining unit. 7. An image pickup means for picking up an image behind a vehicle, a white line detection means for detecting a white line from an image picked up by the image pickup means, and an overtaking based on the white line detected by the white line detection means. A first judging means for judging the detection area of the vehicle, and a second judging means for judging an overtaking vehicle candidate based on a movement vector between frames in the detection area of the overtaking vehicle judged by the first judging means. For each of the overtaking vehicle candidates determined by the determining means and the second determining means, the sum of the edge strengths in a small region around the detection position of each movement vector included in the overtaking vehicle candidates is calculated, and the sum of the edge strengths is calculated. An overtaking vehicle detection device comprising: a removing unit that removes an overtaking vehicle candidate having a smaller than a predetermined value from the overtaking vehicle candidates determined by the second determining unit. 8. The overtaking vehicle detection device according to claim 7, wherein the edge strength is a magnitude of a luminance change. 9. An overtaking device for taking an image of a rear part of a vehicle, a white line detecting device for detecting a white line from an image taken by the imaging device, and an overtaking on the basis of the white line detected by the white line detecting device. A first judging means for judging the detection area of the vehicle, and a second judging means for judging an overtaking vehicle candidate based on a movement vector between frames in the detection area of the overtaking vehicle judged by the first judging means. For each of the overtaking vehicle candidates determined by the determining means and the second determining means, a histogram of edge angles in a small area around the detection position of each movement vector included in the overtaking vehicle candidates is obtained to obtain each overtaking vehicle. First removing means for removing an overtaking vehicle candidate having a variance amount of an edge angle histogram of vehicle candidates smaller than a predetermined value from the overtaking vehicle candidate determined by the second determining means; The moving vector amount in the same direction as the extending direction of the white line detected by the white line detecting means is calculated for each overtaking vehicle candidate determined by the determining means of the overtaking vehicle, and the moving vector amount is less than the predetermined value. Second removal of vehicle candidates from overtaking vehicle candidates determined by the second determination means
For each of the overtaking vehicle candidates determined by the removing means and the second determining means, the sum of the edge intensities in the small area around the detection position of each movement vector included in the overtaking vehicle candidates is calculated to obtain the edge intensities. An overtaking vehicle detection device comprising: a third removing unit that removes an overtaking vehicle candidate having a total sum smaller than a predetermined value from the overtaking vehicle candidates determined by the second determining unit. 10. An overtaking device for taking an image of an image behind a vehicle, a white line detecting device for detecting a white line from an image taken by the image taking device, and an overtaking on the basis of the white line detected by the white line detecting device. A first judging means for judging the detection area of the vehicle, and a second judging means for judging an overtaking vehicle candidate based on a movement vector between frames in the detection area of the overtaking vehicle judged by the first judging means. For each of the overtaking vehicle candidates determined by the determining means and the second determining means, a histogram of edge angles in a small area around the detection position of each movement vector included in the overtaking vehicle candidates is obtained to obtain each overtaking vehicle. First removing means for removing an overtaking vehicle candidate having a variance amount of an edge angle histogram of the vehicle candidate smaller than a predetermined value from the overtaking vehicle candidate determined by the second determining means; The moving vector amount in the same direction as the extending direction of the white line detected by the white line detecting means is calculated for each overtaking vehicle candidate determined by the second determining means, and the moving vector amount is smaller than a predetermined value. Second removal of overtaking vehicle candidates from overtaking vehicle candidates determined by the second determination means
For each of the overtaking vehicle candidates determined by the removing means and the second determining means, the sum of the edge intensities in the small area around the detection position of each movement vector included in the overtaking vehicle candidates is calculated to obtain the edge intensities. Third removing means for removing an overtaking vehicle candidate whose total sum is smaller than a predetermined value from the overtaking vehicle candidates judged by the second judging means; and when the speed of the vehicle is equal to or lower than a predetermined speed, An overtaking vehicle detection device comprising: a prohibiting unit that prohibits deletion of an overtaking vehicle candidate by the first to third removing units. 11. An overtaking device based on a white line detected by the white line detecting device, and a white line detecting device for detecting a white line from an image taken by the image taking device. A first judging means for judging the detection area of the vehicle, and a second judging means for judging an overtaking vehicle candidate based on a movement vector between frames in the detection area of the overtaking vehicle judged by the first judging means. For each of the overtaking vehicle candidates determined by the determining means and the second determining means, a histogram of edge angles in a small area around the detection position of each movement vector included in the overtaking vehicle candidates is obtained to obtain each overtaking vehicle. First removing means for removing an overtaking vehicle candidate having a variance amount of an edge angle histogram of the vehicle candidate smaller than a predetermined value from the overtaking vehicle candidate determined by the second determining means; The moving vector amount in the same direction as the extending direction of the white line detected by the white line detecting means is calculated for each overtaking vehicle candidate determined by the second determining means, and the moving vector amount is smaller than a predetermined value. Second removal of overtaking vehicle candidates from overtaking vehicle candidates determined by the second determination means
For each of the overtaking vehicle candidates determined by the removing means and the second determining means, the sum of the edge intensities in the small area around the detection position of each movement vector included in the overtaking vehicle candidates is calculated to obtain the edge intensities. Third removing means for removing an overtaking vehicle candidate having a total sum smaller than a predetermined value from the overtaking vehicle candidates determined by the second determining means, and the white line detecting means for forming a lane in which the vehicle is passing. When the second determining means determines that the vehicle is an overtaking vehicle continuously near the intersection of the two white lines detected by the above, the first to third removing means delete the overtaking vehicle candidates. An overtaking vehicle detection device comprising: a prohibition unit that prohibits