JP3456843B2 - Front inter-vehicle distance measurement device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for photographing a moving object such as an automobile running in front with two upper and lower cameras to measure an inter-vehicle distance to the moving object.

[0002]

2. Description of the Related Art When a driver of a vehicle becomes tired while traveling on a highway, it is required to prevent a rear-end collision with a vehicle ahead. Further, when the road is congested, it is also required to automatically measure the inter-vehicle distance to the preceding vehicle and prevent a rear-end collision.

Recently, equipment such as ABS (anti-lock / brake / system) and airbag for the purpose of ensuring user's safety has been installed in automobiles. However, these equipment can be installed at the time of accident. It is only a consideration of safety and does not prevent the occurrence of an accident.

On the other hand, as a technique for preventing an accident, a driving support means using an image processing technique as a technique for substituting a human visual function is in the spotlight. Among them, sensors such as ultrasonic waves and radar have a large measurement error in measuring the distance between the vehicles ahead when the traffic is congested. Therefore, the method based on the image processing technique is effective for measuring the distance between the vehicles forward when the traffic is congested.

In the distance measurement of the front vehicle by the image processing, a stereo method is generally used in which the parallax of the front vehicle is obtained from two cameras and the distance between the front vehicles is measured by the principle of triangulation.
The patent applicant previously applied for a measurement device by a stereo method in which a camera is installed vertically as Japanese Patent Application No. 7-83816.

The stereo method of installing the camera horizontally is inferior to the stereo method of installing the camera vertically in the following points. 1. In the case of a stereo system with a camera installed horizontally, the vertical edge of the car is used as an index. If the horizontal edge is used as an index when the camera is installed horizontally, the distance in the depth direction cannot be obtained. Moreover, in comparison with the horizontal edge of the car shown by the thick line in FIG. 13 (A), the number of vertical edges shown by the heavy line in FIG. 13 (B) is not only small, but also short. The number of horizontal edges is smaller than the number of horizontal edges, and the length of the edges is short, so there is more than three times the probability that extraction will not be possible due to changes in the environment such as when the vehicle shakes laterally or when the car is in the shade. become. This is because the vertical edge length is only 1/3 or less of the horizontal edge of the rear window or the horizontal edge such as the bumper edge. Therefore, it is necessary to position the camera vertically to utilize horizontal edges.

2. When the camera is installed horizontally, the distance between vehicles can be obtained by the vertical edge, but the height is not accurate. When the camera is placed vertically, the height and distance with respect to the index are accurate. If the height can be measured accurately, even if characters or the like are written on the road surface, it is possible to exclude objects close to the road surface due to the height. That is, marks and shadows on the road surface can be removed.

3. For example, if there are two cars in front and they seem to overlap, when trying to detect the frontmost car, the two cars overlap and there is no interruption, so it is not possible to determine how far in front The distance will be calculated as it is. For this reason, when two cars appear to overlap each other, it is necessary to combine it with an algorithm for identifying which is the front car, so that the speed of detection of the front car cannot be increased. On the other hand, when the cameras are placed vertically, the horizontal edge is used for measurement, so even if cars overlap, by checking from the frontmost horizontal edge, you can measure from the nearest car. Since the algorithm for separating is unnecessary, the measurement can be performed at high speed.

As described above, the vertically arranged stereo system is superior to the conventional horizontally arranged stereo system.
Next, the vertically arranged stereo system will be described with reference to FIG. In FIG. 14, the vehicle V in front is imaged by the cameras CA and CB vertically arranged in the vehicle behind and the distance L (m) is obtained. Here, the focal lengths of the cameras CA and CB are f (m), the resolution (vertical) of each camera is F (m / pixel), the height of the camera CB is H (m), and the distance between the cameras CA and CB is The vertical distance, that is, the baseline is B (m), and the height h (m) of the vehicle V is
When one point of the horizontal edge located at is P ₂ , the optical axis of the camera CA is LA, the optical axis of the camera CB is LB, and the points P ₁ , P ₃ , and P ₄ are as shown in FIG. , Triangles P ₁ , P ₂ and P ₄ satisfy the following equation (1).

F: Fb = L: (H + B) -h (1) Further, the following formula (2) is established by the triangles P ₃ , P ₂ and P ₄ . f: Fa = L: (H−h) (2) From equation (1)

[0011]

[Equation 1]

From the equation (2), f (H−h) =
Since LFa is obtained, the distance L is obtained by the following equation (3).

[0013]

[Equation 2]

[0014]

As described above, the vertical stereo system is superior to the conventional horizontal stereo system, but the only problem is the distance between the cameras in FIG. 14, that is, the vertical arrangement. That is, the length of the baseline B, which is the distance between the cameras at the time, cannot be sufficiently secured.

As is clear from the above equation (3), the length of this baseline has a great influence on the measurement error, and for example, in order to measure the distance to the preceding vehicle 20 m ahead within the error / m, at least Baseline B requires 20 cm.

In the horizontal stereo system, this length can be secured in the upper part of the windshield, but in the vertical stereo system, there is a problem in that only a length of about 5 cm can be secured so as to sandwich the rearview mirror in the upper part of the windshield. Therefore, an object of the present invention is to provide a longitudinal stereo type front inter-vehicle distance measuring device capable of sufficiently securing the length of the baseline B.

[0017]

In order to achieve the above-mentioned object, according to the present invention, as shown in FIG. 1, a camera CA and a camera CB are arranged vertically apart from each other by a baseline B, and a step is also formed horizontally. Position X (m) apart. As a result, one camera CB can be installed, for example, above the room mirror, and the other camera CA can be installed on the dashboard. As a result, a long baseline B of 25 cm or more can be secured even in a small car without disturbing the driver.

With this arrangement, the vehicle V in front is as follows.
It is possible to obtain the distance L (m) from In FIG. 1, the above equation (1) is established by the triangles P ₁ , P ₂ , and P _{4, and the following} equation (4) is obtained from the equation (1).

F (H−h) = L · Fb−fB ... (4) Further, the following formula (5) is obtained from the triangles P ₅ , P ₂ and P ₆ . f: Fa = (L−X) :( H−h) ... (5) From the equation (5), f (H−h) = Fa (L−X). Then, f (H-h) = L in the equation (4) is obtained in the equation (6).
-Substituting Fb-fB to obtain equation (7).

L · Fb−fB = Fa (L− X) (7) Thus, the distance L to the preceding vehicle can be obtained by the following equation (8).

[0021]

[Equation 3]

In this way, the camera CA and the camera CB are
The distance from the front vehicle can be measured not only by arranging the positions in the vertical direction, but also by arranging the positions in the horizontal direction.

[0023]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 2 shows a first embodiment of the present invention.
~ It demonstrates based on FIG. FIG. 2 is an explanatory view of an installation state of two cameras based on the present invention, FIG. 3 is a first embodiment diagram of the present invention, FIG. 4 is an explanatory view of a vehicle lane detection state, and FIG. 5 is an edge operator and a scanning state. FIG. 6 is an operation explanatory diagram of the horizontal edge histogram generation unit, and FIG. 7 is an operation explanatory diagram of the histogram corresponding unit.

As shown in FIG. 2, in the present invention, when two cameras CA and CB are installed, the camera CA is installed on the dashboard DB and the other camera CB is installed on the upper part of the room mirror RM. To do. In FIG. 2, H is a steering wheel and FW is a windshield. Camera C
A and CB have the same characteristics such as focal length and angle of view, and the optical axis LA of the camera CA and the optical axis L of the camera CB are used.
Install B so that it is parallel to the ground.

In FIG. 3, reference numeral 1 is a camera unit, and 2-
0, 2-1 is a horizontal edge extraction unit, 3-0, 3-1 is a vehicle lane detection and holding unit, 4-0, 4-1 is a horizontal edge histogram generation unit, 5 is a histogram corresponding unit, 6 is distance calculation It is a department.

The camera section 1 is composed of cameras CA and CB for taking an image of a vehicle in front in order to measure the distance to the vehicle in front, and as shown in FIG.
A and CB have vertical steps.

The horizontal edge extraction unit 2-0 scans the image captured by the camera CB with the edge operator for detecting the vehicle lane or the edge operator for horizontal edge extraction. Similarly, the horizontal edge extraction unit 2-1 scans the image captured by the camera CA with an edge operator for detecting the vehicle lane or an edge operator for horizontal edge extraction.

The own vehicle lane detection / holding unit 3-0 uses the own vehicle lane detection edge operator to scan the image taken by the camera CB in the horizontal edge extraction unit 2-0 on the basis of the data scanned by the own vehicle lane, as will be described later. The image area of the lane is detected and held. Similarly, the vehicle lane detection and holding unit 3-1 includes a camera C in the horizontal edge extraction unit 2-1.
The image area of the own vehicle lane is detected based on the data obtained by scanning the photographed image of A by the edge operator for detecting the own vehicle lane, and this is held. When scanning is performed by an edge operator for horizontal edge extraction in the horizontal edge extraction units 2-0 and 2-1, the own vehicle lane detection holding unit 3-0,
Scan between the own vehicle lanes held in 3-1.

The horizontal edge histogram generator 4-0
In the horizontal edge extraction unit 3-0, the output value of the edge operator obtained by scanning the edge operator for horizontal edge extraction, that is, the edge strength is projected on the axis vertical to the image and cumulatively added. Similarly, the horizontal edge histogram generation unit 4-1 projects the output value of the edge operator obtained by scanning the edge operator for horizontal edge extraction in the horizontal edge extraction unit 3-1 onto the axis vertical to the image. , Is cumulatively added to create a histogram.

The histogram correspondence unit 5 extracts, for example, 10 histograms each having a high edge strength from each of the histograms created by the horizontal edge histogram generators 4-0 and 4-1. , If the difference is within a certain value, it is determined that there is a correspondence.

The distance calculation unit 6 calculates the inter-vehicle distance L from the corresponding positions of the histogram based on the equation (8). That is, in the formula (8), f, F, and X are known,
By calculating a and b from the corresponding histogram positions, the equation (8) is calculated to calculate the inter-vehicle distance L.

[0032] Since the height h of the image from the ground can be calculated, that is, whether or not it is on the road surface can be determined, the distance calculation is not performed on the road surface, and the inter-vehicle distance calculation is performed based on the following corresponding histogram.

Next, the horizontal edge extraction unit 2- in FIG.
The operation of 0, 2-1 to distance calculation unit 6 will be described in more detail. For example, the image shown in FIG. 4A taken by the camera CB is digitized with 256 gradations and held, and the horizontal edge extraction unit 2-0 first causes the image shown in FIG.
FIG. 4A, which is predetermined by the edge operator EO for detecting the own vehicle lane of 3 × 1 bits shown in FIG.
Scan to the left of the image center at the position of the vertical axis Y ₁ . Then, the calculated value of the edge operator EO and the position of the horizontal axis thereof are sequentially held by the own vehicle lane detection holding unit 3-0, and scanning is performed to the left end of the image.

The vehicle lane detection / holding unit 3-0 extracts the maximum value from the values of the edge operator EO thus transmitted, and checks whether or not the value is a predetermined threshold value. When it is within the threshold value, the value X ₁ in the horizontal axis direction is held. If when the maximum value is not less than the threshold value, 'performs the same scan from the position of, X ₁ of a point within the maximum value of the threshold' Y ₁ slightly above Y ₁ obtained.

Next, the edge operator EO is similarly scanned to the left from the center of the image at the position of the vertical axis Y ₂ in FIG. 4A, which is separated from Y ₁ by a predetermined vertical distance, and its maximum A value is extracted, and when it is within the threshold value, the value X ₂ in the horizontal axis direction is held, and when the maximum value is equal to or greater than the threshold value, Y ₂
Similar scanning is performed from the position of Y ₂ ′, which is slightly higher, and X ₂ ′ of the point whose maximum value is within the threshold value is obtained.

The vehicle lane detector 3-0 then obtains (Y ₁ , X ₁ ), (Y ₂ , X ₂ ), or (Y ₁ ′, X ₁ ′), (Y ₂ ′) thus obtained. , X ₂ ′) and so on Y ₁
The vehicle lane LW ₀ on the left side is calculated from the X-coordinate Y-coordinates of two points near the axis, the Y ₂ axis, or the axes, and this is held.

Similarly, for the image shown in FIG.
The edge operator EO is scanned to the right from the center of the image at the position of the predetermined vertical axis Y ₁ , and the calculated value of the edge operator EO and the position of the horizontal axis thereof are sequentially detected by the vehicle lane detecting and holding unit 3.
When it is sent to −0 and scanned to the right end, it is checked whether the maximum value is within the threshold value, and if it is within the threshold value, the horizontal axis positions X ₃ and Y ₁ of the maximum value are held. Next, the edge operator EO is scanned from the center of the image at the position of the vertical axis Y _{2 to} the right end, and it is checked whether the maximum value is within the threshold value. If it is within the threshold value, the horizontal axis positions X ₄ and Y _{2 of the} maximum value are checked. Hold. When it exceeds the threshold, the vertical position is moved by ΔY to obtain this.

Depending on the lane, own lanes LW _1-1 and L
Although there is a gap such as between W _1-2 , the maximum value does not reach the predetermined value at this time, so the vertical position is greatly moved to detect the vehicle lane. The own vehicle lane W ₁ on the right side is calculated from the two points thus obtained and held.

In the horizontal edge extraction section 2-1 and the own vehicle lane detection holding section 3-1, the same processing is performed on the image data photographed by the camera CA, and the own vehicle lane on the left side and the own vehicle lane on the right side. You get a car lane.

Next, the horizontal edge extracting section 2-0 is set to the one shown in FIG.
As shown in (A), of the image data captured by the camera CB by the edge operator E1 for horizontal edge extraction, the left vehicle lane and the right vehicle lane held by the vehicle lane detection holding unit 3-0 are displayed. Figure 5 between the vehicle lanes
The inside of the lane area AR shown in (B) is scanned.

Edge operator E1 for horizontal edge extraction
Has a 2-bit configuration of “1” and “−1” in the vertical direction and an n-bit configuration in the horizontal direction, and has a size of 2 × 11 pixels with n = 11 bits, for example. To detect horizontal edges, the width, or n
Is preferably long to some extent. In the case of n = 11, for example, as the edge strength of the upper center pixel of the edge operator E1, the sum of the products of the respective coefficients of the edge operator E1 and the image value thereunder is calculated, that is, the edge strength. As the vertical position of the scanning position of the edge operator E1, it is sequentially output to the horizontal edge histogram generation unit 4-0 for each scanning.

As a result, as shown in FIG. 6, when horizontal edges H0, H1, H2, ... Are present between the left vehicle lane LW ₀ and the right vehicle lane LW ₁ of the image data, the edges are generated. When the horizontal edge H0 is scanned by the operator E1, the edge operator E1 scans the edge position ES ₀ along with the scanning position Y _{0 of} the edge operator E ₀ ,
ES ₁ ... Is output. Then, the horizontal edge histogram generation unit 4-0 receives these outputs and calculates the cumulative value HS0 of the edge strength as HS0 = ES0 + ES1 + ..., And the cumulative value HS is set at the vertical axis position Y ₀ of the image data.
Create a histogram with 0 entered.

Similarly, when the horizontal edge extraction unit 2-0 scans the horizontal edge H ₁ , the vertical axis position Y ₁ and the edge strength thereof are sequentially output, and the horizontal histogram generation unit 4-0 receives these outputs. , A histogram is created in which the cumulative value HS1 is entered in the vertical axis position Y ₁ of the image data. In this way, the horizontal edges H0, H1, H2 ... As shown in FIG.
Corresponding to, the vertical position Y ₀ , Y ₁ , of the horizontal edge,
Accumulated value HS0 edge strength Y ₂ ···, HS1, HS
A horizontal edge histogram in which 2 ... Is entered is created. That is, the edge intensity can be projected on the axis perpendicular to the image data and the result of cumulative addition can be created.

Similar control is performed on the image data projected by the camera CA by the horizontal edge extraction section 2-1, the vehicle lane detection / holding section 3-1, and the horizontal edge histogram generation section 4-1. , A horizontal edge histogram is created.

The histogram correspondence unit 5 finds the correspondence of the histograms created by the horizontal edge histogram generation units 4-0 and 4-1 in this way. First, the histogram correspondence unit 5 extracts, for example, 10 high edge intensity histograms from the edge intensity histogram obtained by the horizontal edge histogram generation unit 4-0 based on the image captured by the camera CB arranged on the upper side. The diagram on the left side of FIG. 7 shows five selected from the lower position. Similarly, based on the image from the camera CA arranged on the lower side, in the histogram of the edge strength obtained by the horizontal edge histogram generation unit 4-1, one having a high edge strength is, for example, 1
Extract 0 pieces. The diagram on the right side of FIG. 7 shows five selected from the lower position.

Then, the histogram correspondence unit 5 makes the histograms correspond in order from the bottom to the top. For example, in FIG. 7, the histograms of the edge strengths HS0 and hs0 are made to correspond to each other, and the edge strength difference | HS0−
hs0 | is calculated, and it is checked whether this is within a certain value HT. When it is within the constant value HT, these histograms are discriminated as corresponding ones, that is, for the same horizontal edge.

Histograms determined to correspond
Is the vertical position Y of each₀, Y₀Is transmitted to the distance calculator 6.
Reached Vertical position Y₀Corresponds to b in FIG. 1 and the vertical axis position
Setting y ₀Corresponds to a in FIG.
Equation (8) is calculated to obtain the distance L to the preceding vehicle. Note that
Prior to this, in the distance calculation unit 6, as described above,
Multiplying these a and b by the resolution F, these histograms
To determine whether is on the road.

A second embodiment of the present invention will be described with reference to FIGS.
It will be explained based on. FIG. 8 is an explanatory diagram of the second embodiment of the present invention, FIG. 9 is an explanatory diagram of distance calculation when the camera is tilted, and FIG. 10 is a configuration diagram of the second embodiment of the present invention. FIG. 11 is an explanatory diagram of the position on the image of the point at infinity (vanishing point) when the optical axis of the camera is horizontal to the ground, and FIG. 12 is an explanatory diagram of the position of the vanishing point and the depression angle when the camera is tilted.

In the second embodiment of the present invention, as shown in FIG. 8, when the cameras CA and CB are installed, their optical axes LA and LB are not horizontal to the ground but are inclined by depression angles α and β. When installed, the distance L to the vehicle V ahead is obtained.

In FIG. 9, the camera C is mounted so as to be inclined by the depression angle α, and the distance to an object located at a height h above the ground is L.
Then, the following formula is established. In FIG. 9, α is the inclination (rad) of the camera C, H is the height (m) of the camera C,
f is the focal length of the camera C (m), L is the distance between the camera C and the object (m), aF is the position of the object on the image plane (m), where a is the position in the image (pixel). , F is the resolution (vertical direction) (m / pixel).

[0051]

[Equation 4]

In the upper camera CB shown in FIG. 8, as is clear from FIG. 8, H in the equation (9) is changed to B + H.
, Α is β, and a is b. That is,

[0053]

[Equation 5]

Similarly, in the camera CA arranged in the lower part shown in FIG. 8, L in the equation (9) is L-X. I.e.

[0055]

[Equation 6]

From the equations (10) and (11), Hh
If you offset L and arrange about L,

[0057]

[Equation 7]

It becomes If you proceed with further organization

[0059]

[Equation 8]

[Mathematical formula-see original document] Then, the distance L to the vehicle V ahead is obtained from the equation (12). In this way, as shown in FIG. 8, the distance L to the vehicle V ahead can be measured even if the cameras CA and CB are installed with inclinations α and β with respect to the ground, respectively. In this case, the cameras CA and CB have the same characteristics such as focal length and angle of view.

Since the distance L to the vehicle V ahead can be measured even if the cameras CA and CB are arranged so as to be inclined as described above,
The second embodiment of the present invention is configured as shown in FIG . In FIG. 10, 10 is a camera unit, 11-
Reference numerals 0 and 11-1 are camera tilt extraction units, 12-0 and 12-1.
Is a horizontal edge extraction unit, 13-0 and 13-1 are own vehicle lane holding units, 14-0 and 14-1 are horizontal edge histogram generation units, 15 is a histogram correspondence unit, and 16 is a distance calculation unit.

The camera section 10 is composed of cameras CA and CB for taking an image of a vehicle in front of the vehicle in order to measure the distance to the vehicle in front, and as shown in FIG.
A and CB are installed vertically at vertical positions and vertically apart from each other by X, and the camera CA is installed at a depression angle α and the camera CB is installed at a depression angle β and inclined with respect to the horizontal direction. There is.

The camera inclination extracting unit 11-0 is arranged to detect the camera CB.
It calculates the depression angle β of the vehicle and the own vehicle lane. First, the method of calculating the depression angle will be described. As shown in FIG. 11, when the optical axis of the camera is horizontal to the ground, mathematically obtaining the intersection D ₀ of the lane based on the image obtained by taking a parallel line such as a lane from the center position of the lane Located in. Since the lanes are parallel lines, they cross at the point at infinity. The intersection on the image by the extension line of the left vehicle lane and the right vehicle lane is called a vanishing point.

Therefore, the camera tilt extraction unit 11-0 digitizes the image photographed by the camera CB and displays it in the vehicle lane detection edge operator as shown in FIG. 4B, for example in FIG. 5A. Using the edge operator as shown, scan lines S ₀ , S ₁ , S as shown in FIG.
_2, the horizontal scanning for the white line search by S ₃ · · ·.
The search for the white line is performed in the lower half area of the image, and the brightness is + in the horizontal direction at a plurality of points such as the scanning lines S ₀ , S ₁ , and S _3.
The point with the maximum value and the point next to it becomes the position of the left or right white line. Therefore, the scanning line S ₀ can be scanned in the horizontal direction from right to left, or by scanning from the center to the left or right, or the like.
Request W right end position P ₀ of the left end position P ₀ and white line LW _{1 0} '.

In this way, the scanning line S₁For the white line
LW₀Left end position P of₁And white line W₁Right end position P₁'But
Required, scan line S₃Also for white line LW₀Leftmost position
Setting P ₃And white line W₁Right end position P₃′ Is required. Shi
Scanning line S₂For the white line LW₀Left end position P of₂
Detected, but scan line S₂For the rightmost lane
White line LW because it exists in the blank area₁The rightmost position of
I can't come.

P ₀ , P ₁ ...
Alternatively, when P ₀ ′, P ₁ ′, etc. are used, that is, when two or more points of the white line on the left and the points on the right of the center are extracted, linear approximation is performed by the least squares method, respectively. 12
As shown in (A), find the intersection D ₁ of each straight line,
The inclination β of the camera is obtained from the length d in the vertical direction between the obtained intersection D ₁ and the center horizontal line of the image.

That is, in FIG. 12B, the camera CB
F (m) for the focal length and F (m / pixel for the resolution
l), when the inclination angle is β (rad), and when the optical axis LB is inclined by the optical axis Lb and β when the ground is horizontal and the image shift on the focal plane is d (pixel), To establish.

[0068]

[Equation 9]

In this way, the camera tilt extraction unit 11-0
Then, the depression angle β is calculated by the equation (13), and this is sent to the distance calculation unit 16. At this time, the white lines LW ₀ and LW ₁ calculated as described above are held as the own vehicle lane in the own vehicle lane holding unit 13-0, and the area between the white lines LW ₀ and LW ₁ is stored in the horizontal edge extraction unit 12-. At 0, the area is scanned by the edge operator E1 shown in FIG. 5A to extract the horizontal edge.

The camera tilt extraction unit 11-1 detects the depression angle α of the camera CA in the same manner as the camera tilt extraction unit 11-0, and sends this to the distance calculation unit 16. In order to hold the own vehicle lane obtained when calculating the depression angle α in the own vehicle lane holding unit 13-1 and to extract the horizontal edge in the horizontal edge extraction unit 12-1,
The area is scanned by the edge operator E1 shown in FIG.

The horizontal edge extraction unit 12-0 is connected to the camera CB.
The image captured by is scanned by the edge operator E1 for horizontal edge extraction as shown in FIG. 5 (A) in the area held by the vehicle lane holding unit 13-0.
In this case, for example, n = 11.

The horizontal edge extraction section 12-1 is connected to the camera CA.
The image captured by is scanned by the edge operator E1 for horizontal edge extraction in the area held by the vehicle lane holder 13-1.

Horizontal edge histogram generator 14-0
Is the horizontal edge histogram generation unit 4- shown in FIG.
It operates similarly to 0. The horizontal edge histogram generation unit 14-1 operates similarly to the horizontal edge histogram generation unit 4-1.

The histogram corresponding section 15 operates similarly to the histogram corresponding section 5. The distance calculation unit 16 receives the depression angle β transmitted from the camera tilt extraction unit 11-0, the depression angle α transmitted from the camera tilt extraction unit 11-1, the focal length f of the cameras CA and CB, the resolution F, and the base. Line B, horizontal step X of cameras CA and CB,
Based on the image positions a, b, etc., the above formula (12) is calculated to calculate the distance L to the vehicle ahead.

That is, when the driver of the car first drives the car and becomes ready to take a picture of the vehicle lane,
The vehicle lane is extracted, and the depression angle α of each camera CA, CB,
β is calculated, and then the horizontal edge extraction unit 12-0 to the distance calculation unit 16 are operated, and the distance L to the vehicle ahead is calculated as described above.
Can be obtained.

According to the front inter-vehicle distance measuring apparatus of the present invention, it is possible to solve the problem of mounting the camera while utilizing the characteristics of the vertical stereo system. By accessing the histogram according to the vertical axis, that is, in the primary direction, desired data can be easily obtained and calculated. In addition, even if the camera tilt changes due to vehicle vibration during operation, the tilt angle is measured and the inter-vehicle distance is measured next to correct this camera tilt change to accurately measure the inter-vehicle distance. You can

[0077]

According to the present invention described in claim 1,
Not only are the cameras vertically offset,
When installing the camera in a car, one unit should be installed near the rearview mirror and the other unit should be placed on the dashboard, for example, to increase the length of the baseline. As a result, it has become possible to measure the inter-vehicle distance very accurately.

According to the present invention described in claim 2, since the optical axis of each camera is arranged to be horizontal to the ground,
The calculation formula for obtaining the inter-vehicle distance is simplified, and as a result, the hardware configuration can be simplified.

According to the present invention as set forth in claim 3, the inter-vehicle distance can be calculated even if the optical axis of each camera is slanted. Therefore, in actual operation, for example, due to vibration, the mounting angle of the camera Even if is changed, the inter-vehicle distance can be calculated according to the changed angle, so that the inter-vehicle distance can be accurately measured without being affected by the change in the tilt of the camera.

According to the present invention described in claim 4, since the tilt of each camera can be calculated by obtaining the position of the vanishing point from each image taken by each camera, the method is very simple. Since the tilt of each camera can be obtained, even if the tilt of the camera changes, the correction can be performed in a short time.

[Brief description of drawings]

FIG. 1 shows a principle diagram of the present invention.

FIG. 2 is an explanatory diagram illustrating an installation state of two cameras according to the present invention.

FIG. 3 is a configuration diagram of a first embodiment of the present invention.

FIG. 4 is an explanatory diagram of a vehicle lane detection state.

FIG. 5 is an explanatory diagram of an edge operator and its scanning state.

FIG. 6 is an operation explanatory diagram of a horizontal edge histogram generation unit.

FIG. 7 is an operation explanatory diagram of a histogram corresponding unit.

FIG. 8 is an explanatory diagram of the second embodiment of the present invention.

FIG. 9 is an explanatory diagram of distance calculation when the camera is tilted.

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

FIG. 11 is an explanatory diagram of a default image position of an infinite point when the optical axis of the camera is horizontal to the ground.

FIG. 12 is a diagram for explaining the position and depression angle of the point at infinity when the camera is tilted.

FIG. 13 is an explanatory diagram of a vehicle edge.

FIG. 14 is an explanatory diagram of a conventional example.

[Explanation of symbols]

1 camera section 2-0, 2-1 Horizontal edge extraction unit 3-0, 3-1 Own vehicle lane detection holding unit 4-0, 4-1 Horizontal edge histogram generation unit 5 Histogram corresponding part 6 Distance calculator 10 camera section 11-0, 11-1 Camera tilt extraction unit 12-0, 12-1 Horizontal edge extraction unit 13-0, 13-1 Own vehicle lane holder 14-0, 14-1 Horizontal edge histogram generation unit 15-Histogram corresponding part 16 Distance calculator

─────────────────────────────────────────────────── ─── Continued Front Page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G01C 3/00-3/32 G01B 11/00-11/30 102 G06T 1/00-9/40 B60R 21 / 00-21/00 630

Claims (4)

(57) [Claims] 1. A front edge distance measuring device having a plurality of cameras for taking images of a front vehicle, which are provided at different heights, and horizontal edge extracting means for extracting a horizontal edge from an image captured by the cameras, A horizontal edge histogram generating means for generating a cumulative value of the intensity of the extracted horizontal edges as a histogram projected on the vertical axis of the captured image, and an error obtained by the two horizontal edge histogram generating means.
Edge hist with high edge strength
Totograms are each extracted, and each extracted hist
Corresponding edges from bottom to top in the gram
The strength difference is calculated, and when this is within a certain value, these hist
Grams can be identified as for the same horizontal edge.
According to, the histogram correspondence means that corresponds the viewpoints of the vertical axes of both cameras from the created histogram, the difference between the position on the vertical axis of the corresponding viewpoints transmitted from this histogram correspondence means, and the vertical position of the camera A front inter-vehicle distance measuring device, comprising: a calculating means for calculating a distance from a front vehicle based on a step which is a difference between the baseline and the horizontal position of the camera. 2. The front inter-vehicle distance measuring device according to claim 1, wherein the optical axes of the cameras are arranged so as to be parallel to the ground. 3. A front vehicle distance measuring device having a plurality of cameras for taking images of a front vehicle, which are provided at different heights, and a camera inclination extraction for extracting a tilt angle of the camera from a captured image of the camera. Means, a horizontal edge extraction means for extracting horizontal edges from the captured image of the camera, and a horizontal edge histogram generation means for creating a cumulative value of the intensities of the extracted horizontal edges as a histogram projected on the vertical axis of the captured image, The error obtained from the two horizontal edge histogram generating means.
Edge hist with high edge strength
Totograms are each extracted, and each extracted hist
Corresponding edges from bottom to top in the gram
The strength difference is calculated, and when this is within a certain value, these hist
Grams can be identified as for the same horizontal edge.
According to, the histogram corresponding means corresponding to the viewpoints of the vertical axes of both cameras from the created histogram, the camera inclination transmitted from the camera inclination extracting means, and the corresponding histogram transmitted from the histogram corresponding means, respectively. The position on the vertical axis of the camera and the baseline, which is the difference in the vertical position of the camera, and the step that is the difference in the horizontal position of the camera. Characteristic forward distance measurement device. 4. The camera tilt extraction means obtains the positions of vanishing points of the left and right lanes from the image taken by the camera, and calculates the tilt of the camera from the difference between the center horizontal line of the image and the vanishing points. The front inter-vehicle distance measuring device according to claim 3.