JPH10115519A - Apparatus for recognizing position of vehicle - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve measurement accuracy for a distance between vehicles by arranging three or more identification means at a rear face of a vehicle body of a preceding vehicle and measuring a relative position to the preceding vehicle on the basis of a binary image. SOLUTION: Three IR(infrared ray)-LEDs 2 are arranged at position of angles of an isosceles triangle at a rear face 1a of a body of a preceding vehicle 1. An IR-camera 4 for photographing the rear face of the body of the preceding vehicle 1 and an image-processing device are set at a following vehicle 3. The image-processing device converts an input image from the IR-camera 4 to a binary image of a brightness level. The IR-LEDs 2 are divided with the rear face 1a of the preceding vehicle 1 at the background, so that three bright pints A1 -A3 are given to the binary image. In order to obtain a coordinate position of each bright point A1 -A3 , a histogram in an (x) direction and a (y) direction is formed from the binary image. The coordinate position of each bright point A1 -A3 is determined from the histogram. A distance between bright points is calculated from the coordinate positions, and a distance between the vehicles is obtained from the calculated values, a focal length of a camera lens and an actual distance between bright points.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for recognizing a position of a vehicle which is used for controlling the running of a vehicle group.

[0002]

2. Description of the Related Art In order to improve road use efficiency and reduce a burden on a driver, an automatic driving system for a group of vehicles that performs a close-up run in which a plurality of succeeding vehicles are arranged in a line with a leading vehicle has been developed. . In such a traveling system, it is necessary to accurately grasp the relative position between the vehicles in order to control the approaching and following in a row, and therefore, an IR-LED (infrared LED) is provided on the rear surface of the body of the preceding vehicle. ), An IR camera for photographing the rear surface of the preceding vehicle on the succeeding vehicle, and means for converting the image into a binary image, and a recognition technology for measuring the relative position of the succeeding vehicle with respect to the preceding vehicle from the binary image. Is known ("The second world congress on i
ntelligent transport systems '95 Yokohama 'P
roceeding volumeШ P1272 ~ P1277 『Platoon system b
ased opitical inter vehicle communication ”).

[0003]

However, in this recognition technology, since there are two IR-LEDs, if one of them fails, the relative position between the vehicles cannot be detected. In the process of measuring the distance between vehicles, I
The deviation of the recognition position of the R-LED greatly affects the measurement error, and this error is proportional to the distance, so that the measurement range of the inter-vehicle distance cannot be widened. That is, there is a problem that high recognition accuracy cannot be expected.

An object of the present invention is to solve such a problem.

[0005]

In the first invention, FIG.
A high-brightness identification means a is provided on the rear surface of the vehicle body of the preceding vehicle, and a camera means b for photographing the rear surface of the vehicle body of the preceding vehicle on the following vehicle, and the relative position with respect to the preceding vehicle is measured based on the binary image. In the recognition device provided with the image processing means c, three or more identification means a are arranged on the rear surface of the vehicle body of the preceding vehicle.

[0006] In the second invention, FIG.
The image processing means c for converting the input image into a binary image having a lightness level; a means e for generating histograms in the x and y directions of bright points corresponding to each identification means of the binary image; Means f for determining the coordinate position of each bright point from the histogram, means g for calculating the distance between each bright point from these coordinate positions, the calculated value, the focal length of the camera lens and the actual distance between the bright points. And means for calculating an inter-vehicle distance for each bright spot distance, and means for calculating an average value of these as a measured value of the inter-vehicle distance.

In the third invention, the image processing means c in the first invention corresponds to means j for converting an input image into a binary image of a lightness level as shown in FIG. Means k for creating a histogram in the x direction of the bright spot
Means m for determining a bright spot position on the x coordinate from the histogram, means n for calculating the ratio of the distance between two bright points from the x coordinate positions of at least three of these points, Means p for calculating the relative yaw angle with respect to the preceding vehicle from the difference from the actual ratio.

In a fourth aspect based on the first aspect, the identification means a is arranged at each corner of an isosceles triangle or equilateral triangle on the rear surface of the vehicle body of the preceding vehicle.

According to a fifth aspect, in the first aspect, an infrared LED is provided as the identification means a and an infrared camera is provided as the camera means b.

[0010]

According to the first aspect of the present invention, a plurality of distances between bright spots (distances between the identification means on the image) are used to determine an inter-vehicle distance for each of the distances between the identification means. Can measure well. If one of the identification means fails, the remaining 2
Measurement values based on the two identification means are obtained. Since a plurality of distances between bright spots can be obtained from the three identification means, it is possible to measure a relative yaw angle between the vehicles.

In the second invention, the coordinate position of each bright spot is determined from the histograms of the binary image in the x and y directions,
The distance between each bright point is calculated from these coordinate positions. Then, the inter-vehicle distance is calculated for each inter-bright point distance from the focal length of the camera lens and the actual inter-bright point distance, and by taking an average of these distances, the measurement accuracy of the vehicle distance can be improved.

In the third invention, a bright spot position on the x coordinate is determined from a histogram of the binary image in the x direction, and a ratio of a distance between two bright spots is calculated from the x coordinate positions of at least three points. I do. Since the x-coordinate position on the image shifts in accordance with the relative yaw angle with respect to the preceding vehicle by the perspective transformation, the ratio of the distance between the two bright points on the x-coordinate and the corresponding actual ratio is calculated. From the difference, the yaw angle with respect to the preceding vehicle can be accurately measured.

According to the fourth aspect of the present invention, the calculation processing is facilitated, and the relative yaw angle between the vehicles is easily measured by positioning the symmetry axis of the isosceles triangle or the equilateral triangle at the center of the rear surface of the vehicle body. I can do it.

According to the fifth aspect, a clear binary image can be obtained day and night.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIGS. 1 to 4 illustrate an example of application to traveling control of a group of vehicles in which a plurality of succeeding vehicles approach and follow a single line to a leading vehicle. In this example, three IR-LEDs 2 (infrared LEDs) are arranged at each corner of an isosceles triangle (or an equilateral triangle) having a horizontal base in this example. The following vehicle 3 includes an IR-camera 4 for photographing the rear surface 1a of the vehicle body of the preceding vehicle 1 and its image processing device 5
Is provided. In addition, since these intermediate vehicles except the first vehicle and the last vehicle of the vehicle group become the following vehicle with respect to the preceding vehicle and the preceding vehicle with respect to the following vehicle, one IR-L is used.
The ED 2 is equipped with an IR-camera 4 and an image processing device 5.

The IR-LED 2 performed by the image processing device 5
The extraction method of FIG. 5 will be described with reference to FIG.
Are classified into two types based on the lightness level. By this binarization processing, the binary image shown in FIG. Prior IR-LED2 against the background of the vehicle body rear 1a of the vehicle 1 are separated to provide a binary image (a) on three bright points A ₁ to A ₃ (high lightness pixels).

The coordinate positions (x ₁ , y ₁ ) of each of the bright points A _{1 to} A ₃ ,
In order to obtain (x ₂ , y ₂ ) and (x ₁ , y ₂ ), a histogram in the y direction is created from the binary image (a). Since this histogram has two peaks, this image (a) is separated into two parts by the valley of the histogram. FIG. 2B shows the binary image after separation, and shows two images b.
_1, each of the x-direction histogram from b ₂ is created. Image b ₁ in the x-direction histogram peak for one, and a y-direction histogram and the x-direction histogram, a coordinate position of the bright point A ₁ of these peaks (x _1,
y ₁ ).

[0018] Since the x-direction histogram peak image b ₂ has two exits, the image b ₂ is left in the valley of the histogram 2
Separated into two. FIG. 9C shows the binary image after separation. From the histogram in the y direction and the histogram in the x direction of the image c ₁ , these peaks are represented by the coordinate position (x ₂ , y ₂ ) of the light point A _2. Is determined. Similarly, these peaks are determined as the coordinate position (x ₃ , y ₃ ) of A ₃ from the histogram of the image c _{2 in} the y direction and the histogram in the x direction.

In this way, the coordinate positions (x ₁ , y ₁ ), (x ₂ , y ₂ ) and (x ₁ , y ₂ ) of each IR-LED ₂ are extracted, but these bright points A ₁ expressed distant relationships between to a _3, is as shown in FIG. Akira points on the image distance d ₁ to d
_{3, d 1 = [(x 3} -x 2) 2 + (y 3 -y 2) 2] 1/2 ... (1) d 2 = [(x 2 -x 1) 2 + (y 2 -y ₁₎ ^2] obtained by the _{^{1/2 ... (2) d 3 =}} [(x 1 -x 3) 2 + (y 1 -y 3) 2] 1/2 ... (3).

When obtaining the inter-vehicle distance from the preceding vehicle 1,
Calculated values between these bright point distance d ₁ to d _3, the actual IR-L
From the ED distances k _{1 to} k ₃ and the lens focal length F of the IR-camera 4, the inter-vehicle distances z _{1 to} z _{3 for} each of the bright spot distances d _{1 to} d _3.
Z ₁ = F · k ₁ / d ₁ (4) z ₂ = F · k ₂ / d ₂ (5) z ₃ = F · k ₃ / d ₃ (6) An average value z = (z ₁ + z ₂ + z ₃ ) / 3 (7) is obtained, and this value z is used as a measured value of the following distance.

By giving the inter-vehicle distance z to the preceding vehicle 1 as an average value of the measured values z _{1 to} z ₃ , the measuring accuracy can be improved to at least 1.732 times. Therefore, assuming that the measurement accuracy is maintained at the same level as in the related art, the measurement range of the inter-vehicle distance can be expanded to 1.732 times. Also, if one of the three IR-LEDs 2 fails, the remaining two IR-LEDs 2
Although the measurement accuracy is reduced based on the LED 2, the function of measuring the distance between vehicles (one of the functions of recognizing the position of the vehicle) can be secured.

Referring to the flowcharts of FIGS. 7 and 8, the measurement of the following distance will be described. An image of the rear surface 1a of the preceding vehicle 1 is input by the IR-camera 4 (step 1). The input image of the IR-camera 4 is converted into the binary image of FIG. 5A, and a flag indicating the number of detections of the IR-LED 2 is set to n.
= 0 (steps 2 and 3). A histogram in the y direction is obtained from the binary image, and it is determined whether or not the histogram has a peak (steps 4 and 5).

When there are two peaks, the binary image is shown in FIG.
(B) is separated into upper and lower two images b ₁ and b ₂ , and the coordinate position (x ₁ , y ₁ ) of the bright point A ₁ is determined for the upper image b _{1 from} the histogram in the y direction and the histogram in the x direction. Is determined, and the detection number flag of the IR-LED 2 is set to n = n + 1 (steps 6 to 10). Then step 12
Although proceed to, if there are two peaks in the histogram of the binary image of FIG. 5 (a), the process proceeds from step 6 to step 11 considers the binary image and image b _2, the process proceeds to step 12 .

[0024] At step 12 a histogram of the x direction of the image b _2. Then, also for this histogram, it is determined whether or not there are two peaks. If there are two peaks, the image b ₂ is divided into two images c ₁ and c at the valley of the peak.
₂ and the coordinate position (x ₂ , y ₂ ) of the bright spot A ₂ is obtained from the histogram of the image c _{1 in} the y direction and the histogram in the x direction.
And set the detection number flag of the IR-LED 2 to n = n + 1.
(Steps 14 to 17). Then, although the processing proceeds to step 19, if there is no two peaks in the histogram of the x-direction of the image b _2, the process proceeds from step 13 to step 18 regards the image b ₂ and the image c _2, the process proceeds to step 19 .

[0025] determining x-direction histogram of the image c ₂ in step 19. Next, the histogram in the y direction and x
And a direction histogram, the coordinate position of the bright point A ₃ (x _3,
y ₃ ) is determined, and the detection number flag of the IR-LED 2 is set to n =
It is set to n + 1 (steps 21 and 22).

When n = 2, one of the IR-LEDs 2 is out of order, and based on the remaining two IR-LEDs 2, the inter-vehicle distance z derived from _{one of the} bright spot distances d _{1 to} d _3.
One of ₁ to z ₃ are the measured value z (step 22, step 23). When n = 3, distances between three bright points d _{1 to} d ₃
The inter-vehicle distance z is given as an average value z of the inter-vehicle distances z _{1 to} z ₃ derived from the above (steps 24 and 25). n = 2
When n and n are other than 3, since the light-to-light distances d _{1 to} d ₃ do not appear, the inter-vehicle distance z cannot be measured (step 2).
6).

When the three IR-LEDs 2 are connected to each other,
Since an isosceles triangle (or an equilateral triangle) is formed as described above, the coordinate position x ₁ of the bright spot A ₁ on the image as shown in FIG.
The relative yaw angle θ of the following vehicle 3 with respect to the preceding vehicle 1 is also easily measured based on the deviation △ d between the center point p = (x ₃ −x ₂ ) / 2 between the light point A ₂ -A ₃ and the bright point A ₂ -A _3. it can.

The vehicle attitude parameters are the yaw angle θ, the pitch angle φ, and the roll angle ψ, and the IR-LED 2 of the preceding vehicle 1 captured by the IR camera 4 of the rear vehicle 3.
The space coordinates u, v, w of the following equation change as follows. In addition,
The coordinate x is the horizontal position of the IR-camera 4, the coordinate y is the height position from the ground, and the coordinate z is each IR from the IR-camera 4.
-Indicates the inter-vehicle distance to LED2.

[0029]

(Equation 1)

For the purpose of measuring the yaw angle θ, assuming that the pitch angle φ and the roll angle ψ are not generated, this equation is expressed by u = x−zθ v = yw = xθ + z .

The coordinates x ₁ of the bright points A _{1 to} A ₃ in FIG.
By substituting x ₂ and x ₃ into the above equation, and further transforming them into perspective coordinates on the xy plane on the image of the lens focal length F, x (A ₁ ) = F · u (A ₁ ) / w (A ₁ ) = F · (x ₁ −zθ) / (x ₁ θ + z) x (A ₂ ) = F · u (A ₂ ) / w (A ₂ ) = F · (x ₂ −zθ) / (x ₂ θ + z) x (A ₃ ) = F · u (A ₃ ) / w (A ₃ ) = F · (x ₃ −zθ) / (x ₃ θ + z) Since the arrangement of the IR-LED 2 is an isosceles triangle (or an equilateral triangle), 2 (x ₁ −x ₂ ) = x ₃ −x ₂ holds between x ₁ , x ₂ , and x ₃ . Given x ₂ as the origin 0, a 2x ₁ = x _3. Substituting this into the equation of the coordinate x (A ₃ ) on the xy plane, x (A ₃ ) = F ・ u (A ₃ ) / w (A ₃ ) = F ・ (2x ₁ −zθ) / (2
x ₁ θ + z) becomes, d ₁ / dx _l is x (A ₃ in FIG. _{6) / x (A 1)} = (2x 1 -zθ) (x 1 θ + z) / (2x 1 θ +
z) (x ₁ −zθ).

From this equation, when the yaw angle θ is 0, x (A ₃ ) /
When x (A ₁ ) = 2, the x coordinate of A ₁ becomes the center point of the x coordinate of A ₂ −A ₃ , but does not become the center point when the yaw angle θ is other than 0. That is, when x (A ₃ ) / x (A ₁ ) = 2, a yaw angle θ corresponding to △ d in FIG. 9 is generated. In this equation, x (A ₃ ) / x ( Once A ₁ ), x ₁ and z are determined, the yaw angle θ can be easily calculated.

X (A ₃ ) / x (A ₁ ) = (2 × ₁ −zθ) (x ₁ θ + z)
/ (2x ₁ θ + z) in (x ₁ -zθ), the yaw angle theta (sin
Assuming that θ) is very small, x ₁ θ is a value close to zero, so x (A ₃ ) / x (A ₁ ) is (2x ₁ -zθ) z / z (x ₁ -zθ) = (2x ₁ -zθ) / (x ₁
−zθ). Therefore, if x (A ₃ ) / x (A ₁ ) = a, the yaw angle θ can be obtained by θ = (ax ₁ −2 × ₁ ) / z (a−1).

In the flowchart of FIG. 8, after the processing in step 26, the coordinate positions x ₁ , x ₂ , x ₃ are obtained from the histogram of the image processing in step 27, and x (A ₃ ) = x ₃ −
x ₂ , x (A ₃ ) = x ₃ −x ₂ is calculated. Then, the process proceeds to a step 28, wherein it is determined whether or not x (A ₃ ) / x (A ₃ ) = 2.
If x (A ₃ ) / x (A ₃ ) = 2, the yaw angle θ is set to 0 in step 29. On the other hand, when x (A ₃ ) / x (A ₃ ) = 2 is not satisfied, in step 30, θ = (ax ₁ −2 × ₁ ) / z
(A-1) is calculated.

The number of IR-LEDs 2 may be increased in order to further increase the measurement accuracy. IR-LE
The arrangement state of D2 is not limited to isosceles triangles or equilateral triangles, and any arrangement in any plane shape complicates the calculation process. However, when three or more IR-LEDs 2 are provided, two or more sides are used. , The measurement of the yaw angle θ is also possible.

[0036]

According to the first aspect of the invention, a high-brightness identification means is provided on the rear surface of the vehicle body of the preceding vehicle, and the camera means for photographing the rear surface of the vehicle body of the preceding vehicle on the following vehicle, and the binary image based on the binary image. In the recognition device including the image processing means for measuring the relative position with respect to the preceding vehicle, three or more identification means are arranged on the rear surface of the vehicle body of the preceding vehicle. Therefore, the inter-vehicle distance is obtained for each of the plurality of identification means. By taking the average value, the measurement accuracy of the inter-vehicle distance can be improved. Even if one of the identification means fails, the inter-vehicle distance can be measured based on the remaining two identification means, so that a high fail-safe function can be obtained. Since three or more identification means are provided, measurement of a relative yaw angle between vehicles is also possible.

According to the second invention, the image processing means in the first invention comprises: means for converting the input image into a binary image having a lightness level; and x of a bright point corresponding to each of the identification means for the binary image. Means for creating a histogram in the direction and y direction, means for determining the coordinate position of each bright spot from the histogram,
Means for calculating the distance between each bright spot from these coordinate positions;
Since there is provided a means for obtaining an inter-vehicle distance for each inter-bright distance from the calculated value, the focal length of the camera lens, and the actual inter-bright distance, and a means for calculating an average value of these as a measured value of the inter-vehicle distance The distance between vehicles is calculated for each distance between bright spots, and by taking an average value of them, a high measurement accuracy of the vehicle distance can be obtained.

According to the third invention, the image processing means in the first invention comprises: means for converting an input image into a binary image of a lightness level; and x of a bright point corresponding to each identification means of the binary image. Means for creating a histogram of directions, means for determining a bright spot position on the x coordinate from the histogram, means for calculating the ratio of the distance between two bright spots from the x coordinate positions of at least three of these, A means for calculating the relative yaw angle with respect to the preceding vehicle from the difference between the ratio and the actual ratio corresponding to the ratio is provided. Can be measured.

According to the fourth aspect, in the first aspect, the identification means is arranged at each corner of an isosceles triangle or an equilateral triangle on the rear surface of the vehicle body of the preceding vehicle, so that the calculation process is facilitated and the isosceles triangle is provided. By positioning the symmetry axis of the triangle or equilateral triangle at the center of the rear surface of the vehicle body, the relative yaw angle between the vehicles can be easily measured.

According to the fifth aspect, in the first aspect, since the infrared LED is provided as the identification means and the infrared camera is provided as the camera means, a clear binary image can be obtained regardless of day or night.

[Brief description of the drawings]

FIG. 1 is a side view of a preceding vehicle and a following vehicle.

FIG. 2 is a plan view of the same.

FIG. 3 is a rear view of the vehicle body of the preceding vehicle.

FIG. 4 is a schematic plan view of a following vehicle.

FIG. 5 is an explanatory diagram of image processing.

FIG. 6 is an explanatory diagram illustrating a distance relationship between bright points on an image.

FIG. 7 is a flowchart illustrating image processing.

FIG. 8 is a flowchart illustrating image processing.

FIG. 9 is an explanatory diagram illustrating a yaw angle measurement technique.

FIG. 10 is a configuration diagram of the present invention.

FIG. 11 is a partial configuration diagram of the present invention.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 preceding vehicle 2 IR-LED 3 succeeding vehicle 4 IR-camera 5 image processing device

Claims (5)

[Claims] A high-brightness identification means is provided on the rear surface of a vehicle body of a preceding vehicle, a camera means for photographing the rear surface of the vehicle body of the preceding vehicle on a following vehicle, and a relative position of the preceding vehicle is measured based on a binary image thereof. A recognition device comprising an image processing means for locating a vehicle, wherein three or more identification means are arranged on a rear surface of a vehicle body of a preceding vehicle. 2. The image processing means according to claim 1, wherein the input image is a lightness level 2
Means for converting into a value image, means for creating histograms in the x and y directions of bright points corresponding to each identification means of the binary image, means for determining the coordinate position of each bright point from the histogram, Means for calculating each inter-bright distance from the coordinate position of the above, means for calculating the inter-vehicle distance for each inter-bright distance from the calculated value, the focal length of the camera lens and the actual inter-bright distance, and the average of these 2. The position recognition apparatus according to claim 1, further comprising means for calculating a value as a measured value of an inter-vehicle distance. 3. An image processing means for converting an input image into a lightness level 2
Means for converting into a value image, means for creating a histogram in the x direction of bright points corresponding to each identification means of a binary image, means for determining a bright spot position on the x coordinate from the histogram, Means for calculating the ratio of the distance between two bright spots from the three x-coordinate positions, and means for determining the relative yaw angle with respect to the preceding vehicle from the difference between this ratio and the corresponding actual ratio. The position recognition device according to claim 1, wherein: 4. The position recognizing device according to claim 1, wherein the identification means is arranged at each corner of an isosceles triangle or an equilateral triangle on the rear surface of the body of the preceding vehicle. 5. The position recognition device according to claim 1, wherein an infrared LED is provided as identification means, and an infrared camera is provided as camera means.