JPH07104160B2 - Inter-vehicle distance detection method - Google Patents

Description

The present invention relates to an inter-vehicle distance measuring method for measuring a distance to a preceding vehicle, and more particularly, to an apparent vehicle width of the preceding vehicle obtained from image data, The present invention relates to an inter-vehicle distance detection method that knows the direction of change in inter-vehicle distance from changes and issues an alarm when an abnormal approach occurs.

[Conventional technology]

As a distance measuring method for measuring a distance to an object, a method of emitting a ultrasonic wave or a laser beam to catch a reflected wave from the object and measuring the distance by this is known. According to these, the absolute distance to the object can be measured without contact.

[Problems to be solved by the invention]

However, the distance measuring method using ultrasonic waves or laser light has the following problems.

． What can be measured by these methods is basically the distance from the observation point to a point on the object on which ultrasonic waves or laser light is projected. Therefore, if these methods are used to measure the inter-vehicle distance to the preceding vehicle during traveling, it is difficult to always reliably measure the distance due to shaking and relative position change accompanying traveling.

． When trying to measure the distance to a distant object, it is necessary to increase the output of ultrasonic waves or laser light,
It causes cost or space problems.

． In the case of ultrasonic waves, the beam cannot be concentrated due to the directivity, and therefore it is difficult to measure a long distance. Although laser light can be sharply directed, there is another problem that the beam is dangerous when it enters the human eye.

[Means for solving problems]

The present invention solves the problems of the prior art as described above.
An inter-vehicle distance measuring method for detecting a change in relative distance to a moving object is provided with the following means.

A horizontal component extracting means (step 1) for spatially differentiating an image of the object to extract a horizontal component and a binarizing means means for binarizing the image in which only the horizontal component remains (step S2). ), And a significant width detecting means (step S4,5) for determining a significant width in the vertical direction in the binarized image and detecting a significant width in the horizontal direction in the vertical significant region, The amount of time change of the width of the, and the increase or decrease direction of the time change is detected to determine the approach or separation of the object (step S6),
Abnormal proximity determination means for detecting an abnormal approach when the temporal variation exceeds a predetermined value in an increasing direction.

[Action]

The object is imaged and the image data is processed to obtain the width (apparent vehicle width) of the object on the screen. In this case, by utilizing the fact that the rear view of the vehicle has a horizontal edge component, the horizontal feature is extracted by image processing to determine the vehicle width on the screen. Such measurement is performed in real time, and the direction of increase or decrease in the temporal change of the vehicle width on the screen is obtained, and a display indicating whether the preceding vehicle is moving away or approaching is displayed. Further, it detects that the amount of change exceeds a predetermined value in the increasing direction, and issues an alarm or display indicating an abnormal approach.

Example of Invention

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention. In FIG. 1, the image pickup apparatus 1 is for picking up an image of an object, especially a preceding vehicle, and can be a conventionally known ITV camera using a camera tube or a CCD camera using a CCD element, for example. The imaging device 1 is installed at a position in the vehicle 2 that does not hinder the driving of the vehicle 2 and at an appropriate location where a sufficient forward field of view can be secured. A / D converter 3
Is an analog digital (A /
D) Convert and input to the processing device 4.

The processing device 4 has a memory 5, and the digitized video signal from the A / D converter 3 is stored in the memory 5. The processing device 4 further has a computer 6, calculates the vehicle width of the preceding vehicle from the data stored in the memory 5 every predetermined time (sampling time), and compares it with the vehicle width at the time of the previous measurement,
The vehicle width increases (+) by obtaining the vehicle width change direction and the vehicle speed change
A signal indicating a direction or a decrease (-) direction and an abnormal approach warning means indicating that the predetermined threshold value is exceeded in the increasing direction are output.

The inter-vehicle distance meter 7 is based on the calculation result of the computer 6,
The + direction signal indicates that the preceding vehicle is approaching, and the-direction signal indicates that the preceding vehicle is approaching, and the abnormal approach warning signal is displayed. When it occurs, the message "Please slow down" is displayed and the buzzer sounds to give an alarm.

FIG. 2 is a diagram for explaining the algorithm of the processing of the computer 6 in the processing device 4. In the method of the present invention,
Conventional technology for accelerating processing by utilizing the fact that the image of the rear view of the vehicle has a horizontal edge component when straight ahead and for observing a point on the body of the preceding vehicle to measure the distance Unlike the above, since the preceding vehicle is grasped as a certain area on the screen and processed, a reliable (result of less detection error due to noise) processing result is obtained.

FIG. 2a shows the image data, and for ease of understanding,
Only vehicle 11 is shown. An image processing technique is applied to the image data to extract only the vehicle 11 as shown in FIG. 2B, and the size of the vehicle 11 is determined at a predetermined sampling time interval Δt (A-B in the vertical direction and C in the horizontal direction). Then, the inter-vehicle distance is calculated. Here, it is assumed that the image data is a grayscale (black and white) image, and each pixel has brightness information. That is, the A / D converter outputs digital image data having gradation to the processing device 4.

FIG. 2c is a flowchart showing the flow of processing. The processing will be described below in order based on these figures. 3A to 3E show the processing results of each stage.

． Detection of Horizontal Component (Step S1: Horizontal Detecting Unit) Spatial differential processing for obtaining a change value of vertical brightness is performed on the image data of FIG. 2A to obtain a differential image. This image represents the horizontal component of the vehicle image.

.2 binarization (step S2: binarization means) The differential pixel (hereinafter simply referred to as pixel) of each point in the image obtained in step S1 is set to "0" by a certain threshold value n.
(White) ”and“ 1 (black) ”are binarized to remove noise. For example, when each pixel value <n; each pixel value is set to 0 (white).

When each pixel value ≧ n; each pixel value is set to 1 (black).

FIG. 3a illustrates an image obtained by binarizing in this manner after detecting the horizontal component.

.A, B determination (step S3: Significant width detection means) For the image obtained in FIG. 3a, the value (0 or 1) of each pixel is integrated in the horizontal direction, and as shown in FIG. 3b. A simple horizontal histogram. On the other hand, a certain threshold value m is set, the image is vertically scanned from the upper and lower ends, and the histogram is compared with the threshold value m. The first points satisfying the histogram value ≥m are A and B, respectively. . When A and B are obtained in this manner, the processing area is determined with A and B as the upper and lower limits, as shown in FIG. 3c.

.C, D determination (step S4: Significant width detecting means) The values (0 or 1) of each pixel are vertically integrated in the processing area shown in FIG. 3c, and then shown in FIG. 3d. Create a vertical histogram like this: Next, find the average value of the vertical histogram, find the widest range that satisfies "histogram value ≥ average value", and set the points at both ends as C and D, respectively. Since C and D are determined by obtaining the average value in this way, even if there is noise as indicated by a in the histogram in the vertical direction, it will not be detected.

FIG. 3e is sized with A, B and C, D thus determined. Fig. 3 shows an extracted image of a vehicle. In FIG. 3, the distance between C and D indicates the apparent vehicle width W on the screen.

． Calculation of relative inter-vehicle distance (step S5: Abnormal approach determination means) If the vehicle width of the preceding vehicle at time t is W (t), it can be expressed as W (t) = D-C.

Furthermore, the vehicle width before the sampling time Δt is W
If (W-t) -W (t- [Delta] t) <0, it is determined that the preceding vehicle is moving away.

B. When W (t) -W (t-Δt)> 0, it is determined that the preceding vehicle is approaching.

C. When a certain threshold value n (n> 0) is set and W (t) -W (t-Δt)> n, it is determined that the state is an abnormal approaching state.

Depending on the judgment result of a to c, as described above, a display indicating whether the preceding vehicle is approaching or moving away is displayed in the inter-vehicle distance meter, and an abnormal approach is warned and a buzzer sounds,
The display of "decelerate" can be performed.

FIG. 4 shows an example of image processing according to the present invention, and a shows original image data. That is, the image data output from the A / D converter 3 to the processing device 4 is shown.
However, this example emphasizes the contour line of the image data and is different from the actual image data. this is,
Although the original image data is an image with gradation, the contour line is emphasized because it is necessary to express it as a diagram. Fourth
FIG. B shows an image in which the horizontal component has been detected and binarized, and corresponds to the image data for which the processing (step S2) in FIG. 3a has been completed. FIG. 4c shows an extracted image of the vehicle,
This corresponds to the image data after the processing (steps S4 and S5) in FIG. As you can see in Figure 4,
The image data of the vehicle under the natural environment is clearly extracted, which shows the effectiveness of the present invention.

Further, as an application example of the present invention, when the vehicle width of the preceding vehicle is known, the absolute inter-vehicle distance can be calculated as follows.

It is assumed that the image pickup apparatus 1 shown in FIG. 1 is located at a height H from the ground and is attached with the optical axis of the image pickup lens parallel to the ground. As shown in FIG. 5, the center of the lens having the focal length f is set at the origin of the three-dimensional coordinate system, and the point P (x, y, z) on the object (vehicle) in the three-dimensional space is center-projected. If the image created on the image plane is p (φ, n), the following relationship holds.

φ = f · (x / z) n = f · (y / z) Coordinates on the image plane corresponding to points C and D in FIG.
Let (φ ₁ , n ₁ ) and (φ ₂ , n ₂ ), respectively, and have the corresponding coordinates in three-dimensional space as (X ₁ , Y ₁ , Z ₁ ), (X ₂ , Y ₂ , Z ₂ ), respectively. Then, assuming that the road surface is flat and is a straight road,
The following relationship holds.

n ₁ = n ₂ Y ₁ = Y ₂ Also, the following relationship exists between the vehicle width (X ² −X ¹ ) and the inter-vehicle distance Z.

X ² −X ¹ = (Z / f) · (φ ² −φ ¹ ) In this case, since the vehicle width (X ² −X ¹ ) is known, the inter-vehicle distance Z is calculated as follows.

Z = f · (X ² −X ¹ ) (φ ² −φ ¹ ) As described above, according to the present embodiment, the distance to the target object in real time (inter-vehicle distance) using only one imaging device. distance)
Can be measured. In this case, the processing speed can be increased by utilizing the fact that the rear view of the vehicle has a horizontal edge component when going straight. In the method of the present invention, unlike the prior art in which the distance between a certain point on the preceding vehicle and the observation point is measured, the vehicle is grasped and processed as an image of a certain area on the screen, and therefore a highly reliable measurement result is obtained. Can be obtained. As an application of the present invention, if the vehicle width of the preceding vehicle is known, it is possible to calculate the absolute inter-vehicle distance.

〔The invention's effect〕

Since the present invention is configured and functions as described above, according to this, since the significant width detecting means operates as described above,
The noise can be removed well. As a result, the inter-vehicle distance can be satisfactorily measured even in an environment where it is very difficult to perform the inter-vehicle distance between relatively moving objects in one moving body in terms of image processing. In addition, since the image processing utilizing the fact that the rear view of the vehicle ahead has many horizontal edge components is performed, this excellent noise removal and the extraction processing of the vehicle ahead can be performed very rapidly.

Moreover, even if the portion other than the vehicle is imaged as described above, the vehicle-to-vehicle distance is measured after removing unnecessary noise portions, so that the vehicle-to-vehicle distance can be measured with outstanding accuracy. A measurement can be made.

In this way, an extremely excellent inter-vehicle distance detection method capable of accurately calculating the inter-vehicle distance by effectively removing noise and effectively extracting only the vehicle portion for which the inter-vehicle distance is to be calculated is provided. can do.

[Brief description of drawings]

FIG. 1 is a diagram showing an embodiment of the present invention, FIGS. 2a to 2c are diagrams showing an algorithm of a computer process in a processing apparatus, and FIGS. 3a to 3e are diagrams for explaining processes of respective steps in the present invention. FIGS. 4A to 4C are diagrams showing an example of image processing according to the present invention, and FIG. 5 is a diagram showing a coordinate system at the time of measuring an absolute inter-vehicle distance. 1 ... Imaging device for capturing an image of an object, 2 ... Vehicle,
3 ... A / D converter, 4 ... Processor equipped with each functional means, 7 ... Inter-vehicle distance meter, 11 ... Vehicle.

Claims (1)

[Claims] 1. A vehicle-to-vehicle distance measuring method for detecting a change in relative distance to a moving object, which is horizontal by performing a spatial differentiation process for obtaining a change value of brightness in a vertical direction with respect to an image of the object. A horizontal component extracting means for extracting a directional component, a binarizing means for binarizing an image in which only the horizontal component remains, and a horizontal histogram value of the binarized image is a threshold value. The significant width in the vertical direction is determined by obtaining two points exceeding m, and the significant width in the horizontal direction is detected by obtaining the widest width in which the histogram value in the vertical direction in the vertical significant region is not less than the average value. Means for detecting the width of the horizontal significant region, and the direction of increase or decrease of the time change to determine the approach or separation of the object, and Target An inter-vehicle distance detection method comprising: an abnormal approach determination means for detecting an abnormal approach when the amount of change exceeds a predetermined value in an increasing direction.