JPH0843083A - Preceding vehicle detection device and access alarming device - Google Patents

Abstract

PURPOSE:To eliminate erroneous detection and to speedily detect a preceding vehicle. CONSTITUTION:A horizontal edge image (a) is created by performing filter treatment to a forward road image (b). A horizontal edge intensity where the change in the density value of an original image from brightness to darkness direction from the lower portion of the horizontal edge image is reflected is evaluated by Ne(y)/Nw(y) function, a horizontal edge which is regarded as the lower-edge candidate of a preceding vehicle according to whether a function value smaller than a threshold Rt exists or not is extracted, and an upper direction is retrieved from a position coordinate yo of the horizontal edge and a coordinate position ys where the absolute value of the intensity of the horizontal edge is minimized for the first time is detected. Then, it is judged whether the detected edge belongs to the preceding vehicle or not according to the ratio of a horizontal edge width at the upper part of the coordinate position ys of the edge to the road width at the coordinate ys, thus surely detecting the preceding vehicle with less erroneous detection.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a preceding vehicle detecting device for detecting a preceding vehicle by image processing, and particularly detecting a preceding vehicle on an expressway and measuring an inter-vehicle distance using the preceding vehicle to cause an abnormal approach to the preceding vehicle. The present invention relates to an approach warning device that warns when a user does.

[0002]

2. Description of the Related Art A conventional preceding vehicle detection device is disclosed in, for example, Japanese Patent Application Laid-Open No. 3-273500. This device recognizes the vehicle traveling area from the front image obtained by the image pickup means, and detects the horizontal edge in the traveling area by searching for the horizontal edge by utilizing the horizontal edge component of the vehicle. In addition, the inter-vehicle distance is calculated from the video camera with the known attachment position and the horizontal edge position of the vehicle bottom end in the image, and the risk is judged between the preceding vehicle and the preceding vehicle based on the own vehicle speed, and it is judged as dangerous. In this case, the system alerts the driver with an alarm.

[0003]

However, in such a conventional preceding vehicle detection device, the detection of the preceding vehicle is performed based on the detection of the horizontal edge. There are many targets whose edges are detected, such as shadows of overpasses, white lines for checking the distance between vehicles, dark / light-dark such as road joints, or light-dark-light density differences, etc. In addition, since it is distributed in a direction substantially perpendicular to the traveling direction of the road, there are many false detections.

Although it is stated that the inter-vehicle distance is detected by the position in the image of the lowermost horizontal edge of the vehicle, the vehicle has a shadow 253 on the road surface as shown in FIG.
Since the vehicle shadow has a density gradation, a plurality of horizontal edges due to the shadow 253 exist in the vicinity of the lower portion of the vehicle as shown in (b), and even if the horizontal edge is detected,
It is difficult to identify the bottom horizontal edge of the vehicle. Therefore, as shown in FIG. 17, if the inter-vehicle distance is calculated from the lower horizontal edge near the vehicle, a distance value shorter than the actual inter-vehicle distance is output. For example, when applied to an approach warning device, a warning is frequently issued. This may give the driver a feeling of strangeness.

As one countermeasure against the above-mentioned problem, horizontal edges such as bridge shadows, road signs, and road joints are distributed over the entire width of the road, and in the case of a road shoulder stationary object such as a utility pole, the maximum of the stationary object is located on the horizontal edge of the shadow. Since the upper edge of the lower part is distributed from outside the driving lane, there is a method of preventing erroneous detection using, for example, that the width of the horizontal edge detected above the lower end of the vehicle is within a predetermined range. Conceivable. However, in this case, when converting the horizontal edge width to the actual vehicle width, the road width at the vehicle bottom coordinate is necessary,
Since the coordinates of the lowermost end of the horizontal edge near the vehicle cannot be the coordinates of the lower end of the vehicle, there is a problem that a new means for specifying the lower end of the vehicle is required.

Further, since the density value in the lane has a low density value that can be recognized as a vehicle shadow in the original image around the lower horizontal edge of the image, the density value prevents erroneous detection due to road signs, seams, and the like. Considering the algorithm to do, the efficiency is very low because it is not known where to look in the lane. Therefore, in this case, there is a problem that it is not specified whether a sufficiently low density value is obtained as a shadow. In view of the above problems, an object of the present invention is to provide a more accurate preceding vehicle detection device that eliminates erroneous detection, and an approach warning device using the same.

[0007]

Therefore, according to the present invention as set forth in claim 1, as shown in FIG. 1, an image pickup means 1 which is mounted on a vehicle and picks up an image of a front road, and an original image of the front road is filtered. A horizontal edge detecting means 2 for detecting a horizontal edge to generate an edge image, a storage means 3 for holding the horizontal edge image and the original image, and a vehicle running lane detecting means 4
And an evaluation means 5 for evaluating the horizontal edge strength reflecting the change of the density value of the original image in the vehicle traveling lane by a predetermined function R (y) of the image coordinate y in the vertical direction, and the evaluation result. Based on the above, the preceding vehicle candidate detecting means 6 for detecting the horizontal edge position yo that can be regarded as a preceding vehicle candidate in the vehicle traveling lane, and the coordinate ys at which the edge strength becomes the minimum for the first time at the upper part of the screen starting from the yo. The preceding vehicle lower end position detecting means 7 and the density determining means 8 for determining whether or not the original image density value on the coordinate ys is within a predetermined range in the vehicle traveling lane are provided, and the original image density value is predetermined. The detection signal of the preceding vehicle is output when it is within the range.

According to a second aspect of the present invention, an image pickup means 1 mounted on a vehicle for picking up an image of a front road, and a horizontal edge detecting means for detecting a horizontal edge by subjecting an original image of the front road to a horizontal edge and generating an edge image. 2 and the horizontal edge image,
The storage means 3 for holding the original image, the own vehicle traveling lane detecting means 4, and the horizontal edge strength reflecting the change of the density value of the original image in the own vehicle traveling lane are predetermined at the vertical image coordinate y. Evaluating means 5 for evaluating with the function R (y) of
Based on the evaluation result, a preceding vehicle candidate detecting means 6 for detecting a horizontal edge position yo that can be regarded as a preceding vehicle candidate in the vehicle traveling lane, and a coordinate ys at which the edge strength becomes minimum for the first time at the upper part of the screen starting from the yo. Detecting means 7 for detecting the lower end position of the preceding vehicle, and the coordinate ys in the traveling lane of the own vehicle.
A density determination means 8 for determining whether the upper original image density value is within a predetermined range, and a width determination means 9 for determining the presence or absence of a horizontal edge having a width within a predetermined range above the coordinate ys are provided. The detection signal of the preceding vehicle is output when the original image density value is within the predetermined range and the width of the horizontal edge is within the predetermined range.

According to a third aspect of the present invention, the above function R (y)
Is the ratio of the number of pixels corresponding to the road width determined by y to the number of horizontal edge points having a predetermined range of intensity within the road width. The invention according to claim 4 is the above-mentioned function R
(Y) is the horizontal edge strength on the search line obtained by converting the search line parallel to the lane marker into the image coordinate system. In the invention according to claim 5, the detection of the edge strength minimum value by the preceding vehicle lower end position detection means is based on the difference and product between the strength value of a certain horizontal edge and the horizontal edge strength values before and after the y coordinate of the horizontal edge. I decided to do it. According to a sixth aspect of the present invention, the edge strength minimum value is detected by the preceding vehicle lower end position detecting means by the function R (y).
The value is detected by determining whether or not the value is less than or equal to a predetermined value. According to the invention of claim 7, the detection of the edge position ys by the preceding vehicle lower end position detecting means stores the past preceding vehicle candidate yo and the position of the coordinate ys with respect to yo in the memory table, and the memory is detected by the detected yo. The detection is performed by referring to the table.

According to an eighth aspect of the present invention, the vehicle lane detecting means 5 detects a white line candidate point on the original image,
The white line determined by the candidate points is approximated by an approximate curve described by a predetermined parameter to detect the vehicle traveling lane. In the invention according to claim 9, the descriptive expression of the approximate curve is at least a (corresponding to the distance from the white line),
b (corresponding to road curvature), c (corresponding to road yaw angle), d
It is assumed to include five parameters (pitch angle corresponding to the angle formed by the road and the camera) and e (corresponding to the road width). In the invention according to claim 10, the descriptive expression of the approximate curve is a quadratic expression. In the invention according to claim 11, the quadratic expression is x = (a + ie) (yd) + b /
(Y−d) + c (where i is an integer). In the invention described in claim 12, the approximation is performed by the least square method.

According to a thirteenth aspect of the present invention, as shown in FIG. 2, an image pickup means 1 attached to the front of the vehicle for picking up an image of the road ahead.
And a horizontal edge detecting means 2 for generating an edge image by detecting horizontal edges by filtering the original image of the road ahead.
And a storage means 3 for holding the original image and the horizontal edge image.
, The vehicle running lane detecting means 4, and the horizontal edge strength reflecting the change of the density value of the original image on the vehicle running lane, the predetermined function R of the vertical image coordinate y.
(Y) Evaluation means 5 for evaluating, a preceding vehicle candidate detecting means 6 for detecting a horizontal edge position yo that can be regarded as a preceding vehicle candidate in the vehicle traveling lane based on the evaluation result, and an upper part of the screen with the yo as a starting point. At first, the leading vehicle lower end position detecting means 7 for detecting the coordinate ys at which the edge strength becomes the minimum, and the density for judging whether or not the original image density value on the coordinate ys is within a predetermined range in the own vehicle traveling lane. The determining means 8, the distance calculating means 10 for calculating the approximate inter-vehicle distance based on the coordinate ys, the distance measuring means 11 mounted on the vehicle for measuring the inter-vehicle distance, and the own vehicle speed The speed detecting means 12 for detecting, the above-mentioned approximate inter-vehicle distance value, the measured value of the distance measuring means,
The alarm means 13 is provided for judging the approaching situation to the preceding vehicle based on the own vehicle speed and issuing an alarm.

[0012]

According to the first aspect of the invention, the original image of the road ahead is subjected to the filter processing to create the horizontal edge image. Then, the horizontal edge intensity reflecting that the density value of the original image changes from the bottom to the light → dark direction of the horizontal edge image is determined by a predetermined function R.
The horizontal edge that can be regarded as the lower end candidate of the preceding vehicle is extracted based on the evaluation result in (y), and the absolute value of the horizontal edge strength is determined to be minimum for the first time after searching further upward from the position coordinate yo of the horizontal edge. The coordinate position ys is detected.
Then, it is determined whether or not the detected edge belongs to the preceding vehicle based on the presence / absence of a pixel having a predetermined density value or less within a predetermined range on the original image from the coordinate position ys of the edge. As a result, the number of erroneous detections is small, and the preceding vehicle is reliably detected.

According to the invention of claim 2, in addition to the invention of claim 1, the detected edge is determined by the presence or absence of a horizontal edge within a predetermined range on the original image from the position coordinate ys of the detected horizontal edge. It is repeatedly judged whether it belongs to the preceding vehicle. The leading vehicle can be detected more reliably. In the invention of claim 3, the function R (y) is set as a ratio between the number of pixels corresponding to the road width and the number of horizontal edge points having the intensity within a predetermined range within the road width. This prevents false detections due to overpasses and road joints. Claim 4
In the invention, the function R (y) is set by the horizontal edge strength on the search line obtained by converting the search line parallel to the lane marker into the image coordinate system. Thus, the horizontal edge can be searched from the horizontal edge and the vertical direction.

According to the fifth aspect of the invention, the preceding vehicle lower end position detecting means detects the minimum edge strength value by the difference and product of the horizontal edge strength value before and after a certain y coordinate and the strength value at the coordinate. As a result, the detection value is highly reliable and the detection accuracy of the preceding vehicle amount is improved. In the invention of claim 6, the preceding vehicle lower end position detecting means detects the edge strength minimum value by comparing a predetermined value with the value of the function R (y). This reduces the load on the image processing apparatus. In the invention of claim 7, the past detection values yo and ys are stored in the table memory, and the coordinate ys is detected by referring to the detected yo. This improves the detection speed of ys. In the eighth aspect of the present invention, the vehicle traveling lane is detected by approximating the white line candidate points on the original image with an approximate curve described by a predetermined parameter.
This improves the detection speed.

According to a ninth aspect of the invention, the descriptive expression for the approximate curve is at least a (corresponding to the distance from the white line) and b.
(Corresponding to road curvature), c (corresponding to road yaw angle), d
The position of each lane marker can be recognized because it includes five parameters (pitch angle corresponding to the angle formed by the road and the camera) and e (corresponding to the road width). Claim 10
In the invention, the approximation curve is constructed by using a quadratic equation. Fast approximation is possible. In the eleventh aspect of the present invention, the approximated curve is x = (a + ie) (yd) + b / (yd) + c.
(However, i is an integer.) High-speed and error-free approximation is possible. According to the twelfth aspect of the present invention, the approximation curve is approximated by the least square method. The approximation error is the minimum.
In the invention of claim 13, the preceding vehicle is detected, and the approximate inter-vehicle distance is calculated from the position on the image. An accurate measurement value is extracted from the distance measuring means based on the approximate inter-vehicle distance, and the approach situation to the preceding vehicle is judged based on the own vehicle speed to activate the alarm device. This enables accurate position measurement.

[0016]

EXAMPLE FIG. 3 shows the structure of this example. First, a video camera 21 is provided in the upper part of the front window of the vehicle 20 toward the front. The video camera 21 captures the road ahead of the vehicle 20, and the road surface image is input to the controller 22. Here, the road surface image is subjected to image processing to detect the vehicle traveling lane. Then, the preceding vehicle is further detected in the own vehicle traveling lane, and the approximate inter-vehicle distance is calculated.

A laser radar range finder 23 provided at the front end of the vehicle 20 and a vehicle speed sensor 24 provided at the front wheels are further connected to the controller 21, and the detected value is input. The laser radar rangefinder 23 has three heads,
Each of them is installed with a predetermined directivity angle in three directions of left (L), center (C), and right (R). Each head emits a beam in each direction to detect the distance of an object including a preceding vehicle.

The controller 21 selects and adopts a detection value as the inter-vehicle distance from the three distance detection values by the laser radar range finder 23 based on the above-mentioned approximate inter-vehicle distance. Then, the approach situation to the preceding vehicle is determined based on the inter-vehicle distance and the input value of the vehicle speed sensor 24. In response to the judgment result, the display device 25 connected to the judgment device displays a corresponding alarm and warns the driver.

Next, the detection of the preceding vehicle in this embodiment will be described. FIG. 4 shows a road surface image including a preceding vehicle and image characteristics thereof. In FIG. 4, a vehicle shadow having a density gradation exists at the bottom of the vehicle. The horizontal edge strength caused by the shadow of the vehicle and reflecting the change of the density value thereof changes according to the change of the y coordinate, and becomes a minimum value at the bottom of the vehicle showing the low density value. Therefore, on the horizontal edge image, the position where the horizontal edge intensity becomes minimum for the first time from the bottom of the screen can be regarded as the lowermost end portion of the vehicle, and is the position where the density value is the lowest on the original image.

In the traveling lane, the maximum value of the length of the horizontal edge near the vehicle is the length reflecting the vehicle width,
For example, in the case of a light vehicle, if the road width is 3.5 m assuming a highway with respect to the vehicle width of 1.4 m, the shadow is shaded by the horizontal edge length and the road width ratio Ne / Nw where the horizontal edge is located. I know if it is a thing.

Therefore, on the horizontal edge image, the edge strength from the bottom of the horizontal edge image is Ne of y on the image coordinate system.
It is possible to evaluate (y) / Nw (y) and detect a horizontal edge that is considered as a vehicle candidate on the traveling lane based on the evaluation result. Then, the preceding vehicle can be detected by detecting the coordinate where R (y) becomes the minimum at the upper part of the screen with the detected y coordinate as the starting point.

FIG. 5 shows the flow of processing in the controller 21. That is, first in step 100,
Three distance measurement values LL, LC, L from the radar range finder 23
Take in R. In step 101, a road surface image signal is captured from the video camera 21. In step 102, the vehicle speed is fetched from the speed sensor 24. Then step 103
In the above, the above detection values are stored in the RAM memory together with the image signal. The RAM memory holds each signal in synchronization with the scanning of the image signal until the next frame comes.

In step 104, the image signal held in the RAM memory is subjected to image processing to detect a horizontal edge. Then, in step 105, the horizontal edge image is stored in the RAM memory. This is also updated for each frame similarly to the above. Note that steps 103 to 10 above
5 can be performed simultaneously by using an image processing device having a pipeline structure.

Subsequently, at step 106, lane marker detection is performed on the road surface image, and a No number is given to each lane marker as shown in FIG. 6 to recognize each coordinate position. For this purpose, for example, the lane marker in the road surface image can be approximated and recognized by a white line model formula including parameters associated with the vehicle position and azimuth, such as the following formula. xi = (a + ie) (y−d) + b / (y−d) + c where a: corresponds to the distance from the white line on the left side b: corresponds to road curvature c: corresponds to diagonal road yaw angle d: pitch angle , Equivalent to the angle formed by the road and the video camera e: Equivalent to the road width, i: Integer, lane marker N counting from the left side
o is the number. That is, some white line candidates are detected on the original image, and the parameters a to e are obtained by curve fitting the above white line model, for example, by the least squares method, and the running lane is recognized by the number of lane marker Nos. To be done.

The following processing will be described below assuming that the coordinate position of the detected lane marker is a white line model display. In step 107, the lane marker of the own vehicle is detected by the parameters a to e of the white line model, and the lane marker No.
Is required to recognize the vehicle driving lane. Step 1
At 08, on the horizontal edge image held in the RAM memory, a horizontal edge having a constant intensity from the bottom of the screen and having a ratio with the road width within a predetermined range is detected as a vehicle lower end candidate.

First, the arrow A shown in FIG.
As described above, the traveling lane is searched from the bottom of the screen along the y axis to detect the number of each horizontal edge point and the number of pixels of the road width at the y coordinate of the horizontal edge. The y-axis coordinate of the horizontal edge is detected by the following formula. if R (y)> Rt then yo = y Here, R (y) is [Ne / Nw] × 100%, and N
e is the number of horizontal edge points, Nw is the number of pixels of the road width at the horizontal edge, and Rt is a threshold value that defines the lower limit strength of the horizontal edge. When the horizontal edge is detected, the process proceeds to the next step. If it cannot be detected, the process returns to step 100.

When the vehicle lower end candidate is detected, the following equation is calculated in step 109, and as shown in FIG. 7A, the coordinates at which R (y) becomes the minimum for the first time from the upper side of the screen from the coordinate yo. The value is detected, and the detected value ys is recognized as the vehicle lower end coordinate.

[Equation 1] Here, Δy is a constant that limits the search range of the lower end position of the preceding vehicle in the y direction. The subscript j indicates the immediately preceding / posterior relationship of the three y coordinates, and -1 is one pixel below on the screen and +1.
Indicates one pixel above the screen. Figure 8 shows y
The detection process of the s coordinate is shown. That is, when the sign changes in the direction indicated by the arrow C based on the front-back difference of the evaluation function R (y), it is recognized as the minimum value and the coordinate ys is detected.

In step 110, it is further checked on the original image whether or not there are pixels having a predetermined density or less within the area defined by the ys coordinates and the road width. If so, go to the next step. If it does not exist, the process returns to step 100. As a result, it is not the vehicle shadow (a), (b) in FIG.
Inter-vehicle distance confirmation white line 113, road joint 114
It is possible to prevent erroneous detection such as. In step 111, the arrow B is displayed on the horizontal edge image as shown in FIG.
Within the vehicle lane area above ys in the direction indicated by
An evaluation function G (y) is created with a ratio Ne / Nw (ys) of the number Ne of points of each horizontal edge and the number Nw (ys) of road width pixels in the ys coordinate. The evaluation function G (y) can be created, for example, by counting the number of pixels of the corresponding road width for each y coordinate of the horizontal edge.

In step 112, it is checked from G (y) above whether there is a horizontal edge satisfying the following expression. 1.4 / 3.5 ≦ G (y) ≦ 2.5 / 3.5 Note that 2.5 is the vehicle width of a large truck. Then, the coordinates xc0, xc1 and y of both ends of the horizontal edge satisfying the above expression
own vehicle running lane marker coordinate x0 (ys), x
From 1 (ys), it is detected by the following equation whether the horizontal edge is within the lane. [X0 (ys) <xc0] and [x1 (ys)> xc
1] If all of the above conditions are satisfied, proceed to the next step. If not satisfied, return to step 100. As a result, ys can be recognized as the lower end of the vehicle, and FIG.
11 is the same shadow as the overpass bridge, and FIG.
The shadow 112 of the roadside stationary object such as is distinguished from the shadow of the vehicle.

When the preceding vehicle is detected by the above processing, the approximate inter-vehicle distance and the approaching condition with the host vehicle are subsequently detected. In step 113, ignoring the road slope,
Using the pitch angle d obtained by the approximation of the white line model and the above ys, the approximate distance to the preceding vehicle is calculated by the following equation. L_IMG = F · H0 / (ys-d) Here, F is a constant corresponding to the focal length, and H0 is the mounting height of the video camera. RAM in step 114
The three measured values of the laser radar are read from the memory, and the absolute values ΔLL, ΔLC, and ΔLR of the differences from the above-mentioned approximate distance are obtained. In step 115, the minimum value within the above ΔLL to ΔLR is obtained, and the radar distance value that brings the minimum value is adopted as the distance to the preceding vehicle.

In step 116, the detected value of the vehicle speed is read from the RAM memory, the inter-vehicle distance obtained above is differentially calculated with respect to time, and whether the vehicle speed is too close by the relative speed and the vehicle speed. It is determined whether or not it is determined that the distance is too close, and the process returns to step 100.
If it is determined that they are too close to each other, an alarm command is output to the display device 25. After that, the process returns to step 100, and the above flow is repeated to continuously detect. This embodiment is configured as described above, and as a result, it is possible to prevent erroneous detection due to the white line for checking the inter-vehicle distance that is not the shadow of the vehicle and horizontal edges such as road joints, and the shadow of the overpass and the stationary object on the shoulder of the preceding vehicle. It is distinguished from the shadow of the quantity, its erroneous detection is prevented, and the preceding vehicle can be surely detected.

FIG. 12 shows a second embodiment of the present invention.
The modification of the detection of the vehicle lower end shown in the first embodiment is shown.
That is, the threshold value Rt is set in advance, and the minimum value is detected by comparing the function R (y) with it. Then, the coordinates of the minimum value are detected as the lower end position of the preceding vehicle.
Other configurations are similar to those of the first embodiment.

Next, as a third embodiment of the present invention, the first
The modification of the detection of the lower end of the vehicle shown in the embodiment of FIG. In this embodiment, as shown in FIG. 13, five search lines parallel to the lane marker detected on the original image are created, and horizontal edge intensity distribution on the search line is created for each line as shown in FIG. . Threshold value eh for that intensity
_T is provided, a line having more peaks is selected, and under the condition that a certain number or more exist,
In the example of No. 4, the average value of the peak positions y2 to y4 is yo =
By taking (y2 + y3 + y4) / 3, the preceding vehicle lower end candidates yo and ys are calculated. Other configurations are similar to those of the first embodiment.

Next, a further simplified detection method will be described as a fourth embodiment. That is, the vehicle lower end candidate coordinates yo detected in the past, the vehicle lower end coordinates ys, and the difference Δ between them.
The y is stored in the table memory as shown in FIG. 15, and the lower end position of the preceding vehicle is detected by referring to the table thereafter. In the above-described embodiment, the example in which the approach warning device is applied has been mainly described, but the present invention is not limited to this, and may be applied to, for example, an autonomous vehicle that automatically follows a preceding vehicle.

[0035]

As described above, the horizontal edge strength reflecting the change of the density value of the original image from the bottom of the horizontal edge image to the light → dark direction is evaluated by a predetermined function R (y), and the evaluation is performed. Based on the result, the horizontal edge that can be regarded as the lower end candidate of the preceding vehicle is extracted, and the coordinate position where the absolute value of the intensity of the horizontal edge becomes minimum for the first time is detected from the position coordinates where the horizontal edge is extracted. Then, the lower end position coordinates of the preceding vehicle are detected depending on the presence or absence of a pixel having a predetermined density value or less within a predetermined range from the coordinate position of the horizontal edge. When the presence of the preceding vehicle is determined by the presence of the horizontal edge within the predetermined range above the image from the detected lower end position coordinates of the preceding vehicle, the preceding vehicle is detected more reliably. Further, when the function R (y) is set as a ratio of the number of pixels corresponding to the road width to the number of horizontal edge points having the intensity within a predetermined range within the road width, erroneous detection due to an overpass, a road joint, or the like will occur. Can be prevented. Furthermore, when the function R (y) is set by the average value of horizontal edge intensities on the search line obtained by converting the search line parallel to the lane marker into the image coordinate system, the reliability of the detected value is improved. As a result, it is possible to detect a preceding vehicle with high reliability, and the inter-vehicle distance can be calculated more accurately using images.

[Brief description of drawings]

FIG. 1 is a correspondence diagram of a claim.

FIG. 2 is a correspondence diagram of claims.

FIG. 3 is a diagram showing a configuration of a first exemplary embodiment.

FIG. 4 is a diagram showing a road surface image including a preceding vehicle, its density, and horizontal edge strength.

FIG. 5 is a flowchart showing a flow of processing in a controller.

FIG. 6 is an explanatory diagram of a white line model.

FIG. 7 is a diagram showing a relationship between a horizontal edge and ay coordinate in a horizontal edge image and an original image thereof.

FIG. 8 is a diagram showing a process of detecting a horizontal edge of a vehicle lower shadow portion.

FIG. 9 is an image having horizontal edges.

FIG. 10 is an image having a shadow.

FIG. 11 is an image having a shadow.

FIG. 12 is a diagram showing detection of a horizontal edge of a vehicle lower shadow portion.

FIG. 13 is a road surface image having a search line.

FIG. 14 is a diagram showing horizontal edge strength detected on a search line.

FIG. 15 is a diagram showing an example of a table memory.

FIG. 16 is a diagram showing an original image and its horizontal edge image.

FIG. 17 is a diagram showing changes in an original image and its horizontal edge strength.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Imaging means 2 Horizontal edge detection means 3 Storage means 4 Own vehicle traveling lane detection means 5 Evaluation means 6 Preceding vehicle candidate detection means 7 Preceding vehicle lower end position detection means 8 Density determination means 9 Width determination means 10 Distance calculation means 11 Speed means 12 Distance measuring means 13 Warning means 20 Own vehicle 21 Video camera 22 Controller 23 Laser radar 24 Speed sensor 25 Warning indicator

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G06T 9/20 G08G 1/16 A

Claims (13)

[Claims] 1. An image pickup means mounted on a vehicle for picking up an image of a front road, a horizontal edge detecting means for detecting a horizontal edge by subjecting an original image of the front road to a horizontal edge and generating an edge image, the horizontal edge image, A storage means for holding the original image, a vehicle running lane detection means, and a horizontal edge strength reflecting a change in the density value of the original image in the vehicle running lane is a predetermined function of the image coordinate y in the vertical direction. Evaluating means for evaluating with R (y), preceding vehicle candidate detecting means for detecting a horizontal edge position yo that can be regarded as a preceding vehicle candidate in the vehicle traveling lane based on the evaluation result, and upward of the screen starting from the yo. For the first time, the leading vehicle lower end position detecting means for detecting the coordinate ys at which the edge strength becomes the minimum, and the darkness for judging whether or not the original image density value on the coordinate ys is within a predetermined range in the vehicle traveling lane. A preceding vehicle detection device for outputting a detection signal of a preceding vehicle when the original image density value is within a predetermined range. 2. An image pickup means mounted on a vehicle for picking up an image of a road ahead, a horizontal edge detecting means for detecting a horizontal edge by applying a filtering process to an original image of the road ahead and generating an edge image, the horizontal edge image, A storage means for holding the original image, a vehicle running lane detection means, and a horizontal edge strength reflecting a change in the density value of the original image in the vehicle running lane is a predetermined function of the image coordinate y in the vertical direction. Evaluating means for evaluating with R (y), preceding vehicle candidate detecting means for detecting a horizontal edge position yo that can be regarded as a preceding vehicle candidate in the vehicle traveling lane based on the evaluation result, and upward of the screen starting from the yo. For the first time, the leading vehicle lower end position detecting means for detecting the coordinate ys at which the edge strength becomes the minimum, and the darkness for judging whether or not the original image density value on the coordinate ys is within a predetermined range in the vehicle traveling lane. A degree determining unit and a width determining unit above the coordinate ys for determining the presence or absence of a horizontal edge having a width within a predetermined range, and the original image density value is within the predetermined range,
A preceding vehicle detection device, which outputs a preceding vehicle detection signal when the width of the horizontal edge is within a predetermined range. 3. The function R (y) is a ratio between the number of pixels corresponding to a road width determined by y and the number of horizontal edge points having an intensity in a predetermined range within the road width. The preceding vehicle detection device according to 1 or 2. 4. The preceding vehicle inspection according to claim 1 or 2, wherein the function R (y) is a horizontal edge strength on a search line obtained by converting a search line parallel to the lane marker into an image coordinate system. Output device. 5. The edge strength minimum value detected by the preceding vehicle lower end position detecting means is performed by the difference and product between the strength value of a certain horizontal edge and the horizontal edge strength values before and after the y coordinate of the horizontal edge. Claim 1 or 2 characterized
The preceding vehicle detection device described. 6. The minimum edge strength value detected by the preceding vehicle lower end position detecting means is detected by determining whether or not the value of the function R (y) is less than or equal to a predetermined value. The preceding vehicle detection device according to claim 1, 2, 3, or 4. 7. The detection of the edge position ys by the preceding vehicle lower end position detecting means stores a past coordinate position ys for the preceding vehicle candidates yo and yo in a memory table, and refers to the memory table by the detected yo. The preceding vehicle detection device according to claim 1 or 2, wherein the detection is performed by performing the detection. 8. The own vehicle traveling lane detecting means detects a white line candidate point on the original image, approximates the white line determined by the candidate point with an approximate curve described by a predetermined parameter, and obtains the own vehicle. The preceding vehicle detection device according to claim 1 or 2, which detects a traveling lane. 9. The descriptive expression of the approximate curve is at least a
(Corresponding to the distance from the white line), b (corresponding to the road curvature), c
9. The preceding vehicle detection according to claim 8, further comprising five parameters (equivalent to road yaw angle), d (pitch angle corresponding to angle formed by road and camera), and e (equivalent to road width). apparatus. 10. The preceding vehicle detection device according to claim 8, wherein the descriptive expression of the approximate curve is a quadratic expression. 11. The quadratic formula is x = (a + ie) (y
11. The preceding vehicle detection device according to claim 10, wherein -d) + b / (yd) + c (where i is an integer). 12. The preceding vehicle detection device according to claim 8, wherein the approximation is performed by a least square method. 13. An image pickup means mounted on a vehicle for picking up an image of a front road, a horizontal edge detection means for filtering a front road original image to detect a horizontal edge and generating an edge image, the horizontal edge image, A storage means for holding the original image, a vehicle running lane detection means, and a horizontal edge strength reflecting a change in the density value of the original image in the vehicle running lane is a predetermined function of the image coordinate y in the vertical direction. R (y)
And the horizontal edge position yo that can be regarded as a preceding vehicle candidate in the own vehicle traveling lane based on the evaluation result.
A preceding vehicle candidate detecting means for detecting the above, a preceding vehicle lower end position detecting means for detecting a coordinate ys at which the edge strength becomes minimum for the first time in the upper part of the screen starting from the yo, and an original on the coordinate ys in the vehicle running lane. Density determination means for determining whether or not the image density value is within a predetermined range; distance calculation means for calculating a rough inter-vehicle distance to the preceding vehicle based on the coordinates ys;
Distance measuring means mounted on the vehicle for measuring the distance between the preceding vehicle and the preceding vehicle, speed detecting means for detecting the own vehicle speed, the approximate vehicle distance value, the measured value of the distance measuring means, and approaching the preceding vehicle according to the own vehicle speed. An approach warning device comprising: a warning means for judging a situation and issuing a warning.