JPH08147599A - Obstruction approaching state detector - Google Patents

Abstract

PURPOSE: To surely detect the approaching state of an obstruction, to simplify hardware constitution and processing algorithm, to lower a cost and to perform a high-speed processing. CONSTITUTION: This obstruction approaching state detector is provided with a two-dimensional CCD camera 1 for continuously picking up images inside a prescribed visual field and an image processor 2 for judging whether or not the obstruction is in the approaching state by processing video signals for expressing the images of the camera 1 at the two sequential earlier and later points of time. The image processor 2 sets plural horizontal scanning lines including a processing object reference line as a processing object range for the images of the camera 1, uses the video signals on the processing object reference line at the earlier point of time and the video signals on the horizontal scanning lines within the processing object range at the later point of time corresponding to them, obtains motion vectors in the horizontal direction between the two sequential earlier and later points of time for plural points on the processing object reference line at the earlier point of time, detects the obstruction based on the motion vectors and judges whether or not the obstruction is in the approaching state.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an obstacle approaching state detecting device, and more specifically, it is mounted on an automobile, for example, and detects obstacles such as other vehicles based on a video signal of a two-dimensional image pickup device such as a television camera. The present invention relates to a device that detects whether or not a vehicle is approaching.

[0002]

2. Description of the Related Art As an obstacle detection device of this kind, "Preliminary Collection 924 of Academic Society of Automotive Engineers of Japan" (1992)
(October, 2013) As described in "Development of ranging algorithm using stereo image processing", pages 153 to 156 (hereinafter referred to as known document 1), stereo image processing using two TV cameras and A combination of monocular image processing for extracting an optical flow is known. In this apparatus, two television cameras are used to measure the distance between points within the entire visual field by known stereo image processing at regular intervals. Then, during the distance measurement by this stereo image processing, one horizontal scanning line including the point at infinity of the television camera is obtained from two temporally continuous images obtained by one television camera by a known density gradient method. A horizontal one-dimensional optical flow (movement vector) is obtained for each of the above points, and this is used to estimate distance information during distance measurement by stereo image processing.

Also, "Preliminary Proceedings of the Autumn 1993 Autumn Research Presentation" at Toyota Central Research Institute (issued November 26, 1993)
As described in “Extraction of Moving Objects by Moving Image Processing” on pages 31 to 32 (hereinafter referred to as known document 2), one television camera is used and two images that are temporally consecutive are By using a known corresponding point search method similar to that for stereo image processing, a two-dimensional optical flow is extracted over the entire small area within the field of view of the television camera, and the flow due to the movement of the vehicle is canceled to It is known to extract a moving object from the environment.

Further, as described in JP-A-5-112224 (hereinafter referred to as known document 3), 1
The area of an obstacle is calculated based on the video signals from the two TV cameras, and whether the obstacle is approaching due to the change in area between two consecutive images over time It is known to judge whether or not the user is moving away.

[0005]

In the case of the device described in the known document 1, it takes a long time to perform one distance measurement only by the stereo image processing, and the horizontal direction is set for each point on a specific horizontal scanning line. Since accurate distance information cannot be obtained only by obtaining the one-dimensional optical flow of, stereo image processing and monocular image processing are used together. Since two television cameras are used and stereo image processing and monocular image processing are used together, there is a problem that the hardware configuration and processing algorithm are complicated and the cost is high.

In the case of the device described in the known document 2,
Since the optical flow is extracted over the entire screen, it takes time to process, complicated hardware is required for speeding up, and there is a problem of high cost. Moreover, it can only detect whether or not there is a moving object, and cannot judge whether the moving object is approaching or moving away. In order to judge this, it is necessary to use stereo image processing together, as described in the known document 1.

In the case of the device described in the known document 3,
One television camera can be used to determine if an obstacle is approaching or moving away. However, there is a problem in that it is necessary to perform processing for area calculation over the entire area or a wide area of the screen, which requires time, complicated hardware is required for speeding up, and cost is high.

The object of the present invention is to solve the above problems,
An object of the present invention is to provide an economical obstacle approaching state detection device capable of surely detecting the approaching state of an obstacle, having a simple hardware configuration and a processing algorithm, and thus being inexpensive and capable of high-speed processing.

[0009]

An obstacle approaching state detection device according to the present invention is one two-dimensional image pickup device for continuously picking up an image in a predetermined visual field at a fixed time interval, and two consecutive front and rear image pickup devices. The image processing apparatus includes an image processing apparatus that determines whether or not an obstacle is in an approaching state by processing a video signal representing an image of the two-dimensional imaging apparatus at a time point, and the image processing apparatus is the two-dimensional imaging apparatus. For the image, a plurality of horizontal scanning lines including a predetermined processing target reference line are set as a processing target range, and the video signal on the processing target reference line at the previous time point and the processing target at the subsequent time point corresponding thereto Using the video signal on the horizontal scanning line within the range, before and after consecutive 2 points at a plurality of points on the reference line to be processed at the previous time point.
Characterized in that a horizontal movement vector between two time points is obtained, an obstacle is detected based on the movement vector, and whether or not the obstacle is in an approaching state is determined. Is.

For example, the image processing apparatus extracts, on the reference line to be processed at the previous time point, a portion in which the magnitude of the movement vector is continuously changed to be an obstacle detection portion, and detects the obstacle. The magnitude in the first lateral direction of the movement vector at the end on the first side of the left and right sides of the portion, and the above-mentioned magnitude of the movement vector at the end on the second side of the left and right sides of the obstacle detection portion. The difference with the size of 1 in the lateral direction is obtained, and when the difference is a positive value, it is determined that the obstacle is in the approaching state.

[0011]

Since only the video signals on some horizontal scanning lines within the processing target range including the processing target reference line need to be processed, the processing does not take time even if the hardware configuration is not complicated, High-speed processing is possible. Then, the image processing apparatus determines whether or not the obstacle is in the approaching state, based on the horizontal movement vector between two consecutive time points before and after the plurality of points on the processing target reference line at the previous time point. Since the determination is made, it is possible to surely detect whether or not the obstacle is in the approaching state, although the video signals of a few horizontal scanning lines are only processed.

The image processing apparatus extracts, on the reference line to be processed at the previous time point, a portion in which the magnitude of the movement vector continuously changes, and sets it as an obstacle detection portion. Of the magnitude of the movement vector in the first lateral direction at the end on the first side and the magnitude of the movement vector in the first lateral direction at the end on the second side of the left and right sides of the obstacle detection portion. If the difference is obtained and it is determined that the obstacle is approaching when the difference is a positive value, the obstacle is easily and surely detected, and the obstacle is approaching. Whether or not there is can be easily and surely determined.

[0013]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 schematically shows an example of the structure of an obstacle approaching state detection device mounted on a vehicle. This device detects an obstacle such as another vehicle in front of or behind the vehicle, and detects whether or not it is in an approaching state with respect to the host vehicle.

In FIG. 1, the obstacle approaching state detecting device comprises a two-dimensional CCD camera (1) and an image processing device (2).

The camera (1) constitutes a two-dimensional image pickup device for continuously picking up images in a predetermined visual field such as the front and side of the vehicle at regular time intervals.

The image processing device (2) judges whether or not the obstacle is in the approaching state by processing the video signals representing the images of the camera (1) at two consecutive points before and after. The image processing device (2) is composed of a microcomputer, and includes a processing target range setting unit (3), a motion detection processing unit (4), and an approach obstacle presence / absence determination unit (5).

The processing target range setting unit (3) is for setting a plurality of horizontal scanning lines including a predetermined processing target reference line as a processing target range for the image of the camera (1). The reference line to be processed is set to include one or several horizontal scanning lines. The processing target range is set so as to include a processing target reference line and a plurality of upper and lower horizontal scanning lines centering on the reference target line, for example, about 20 horizontal scanning lines. The processing target reference line and the processing target range are set based on the initial image captured by the camera (1) when the apparatus is activated. After this initial setting, the processing target reference line and the processing target range may not be changed, or these settings may be changed as necessary. In this embodiment, the camera (1)
The vertical coordinate of the point at infinity is obtained from the initial image captured in step 1 by a known method, and one horizontal scanning line corresponding to this vertical coordinate is set as the processing target reference line. About horizontal scanning lines are set as the processing target range and are not changed thereafter.

The motion detection processing section (4) uses the video signal on the processing target reference line at the previous time point and the video signal on the horizontal scanning line in the processing target range at the subsequent time point corresponding thereto, This is for obtaining a horizontal movement vector between two consecutive time points before and after the plurality of points on the processing target reference line at the time point. A known corresponding point search method such as a block matching method using a grayscale image or a characteristic image is used to search for horizontal scanning lines within the processing target range and to extract movement vectors corresponding to the processing target reference line at the previous time. Can be done by

The approaching obstacle presence / absence determining section (5) is for detecting an obstacle based on the movement vector and determining whether or not the obstacle is in an approaching state. In this embodiment, on the processing target reference line at the previous time point, a portion in which the magnitude of the movement vector continuously changes is extracted as an obstacle detection portion. When an obstacle in front of the vehicle is targeted, the movement vector of the background portion appears smaller than that of the obstacle portion, and the boundary portion becomes intermittent. Further, since the size of the limit of the movement vector of the background portion is substantially determined according to the speed of the host vehicle, it is possible to extract a portion exceeding the limit as an obstacle portion, that is, an obstacle detection portion. Further, when targeting an obstacle behind and behind the vehicle, since all the background movement vectors are vectors in the direction of leaving, it is possible to separate this and extract only the obstacle part as the obstacle detection part. it can. The magnitude in the first lateral direction of the movement vector at the first side end of the left and right sides of the obstacle detection portion and the movement vector at the second side end of the left and right sides of the obstacle detection portion The difference from the size of the first side direction is determined, and when the difference is a positive value, it is determined that the obstacle is approaching.

Next, with reference to FIGS. 2 and 3, an example of the operation of the vehicle approaching state detection device will be described.
FIG. 2 is a flowchart showing the operation of the device. FIG. 3 shows an example of the initial image of the camera (1).

In FIG. 2, when the apparatus is activated, first, the camera (1) captures an image of its field of view (step 1), and an initial image as shown in FIG. 3 is obtained. The target reference line and the processing target range are set (step 2). As described above, the pixel of infinity point (P) is obtained from the initial image by a known method,
One horizontal scanning line corresponding to the vertical coordinate (Y-axis coordinate) is set as the processing target reference line (L), and about 20 horizontal scanning lines above and below the center scanning line are set as the processing target range ( It is set as S).
When the setting of the processing target range is completed, the video signal on the processing target reference line of the initial image and a plurality of upper and lower scanning lines centering on the processing target reference line are stored as the video signal at the previous time point (step 3). Next, when a certain time (for example, 100 ms) has passed from the initial image capturing in step 1, the camera (1) captures the next image (step 4) and the horizontal scanning line within the processing target range of the image. The upper video signal is stored as a video signal at a later time (step 5). Next, the motion detection process (step 6) and the determination and output of the presence or absence of the approaching obstacle (step 7) are performed. As described above, the motion detection processing in step 6 is publicly known based on the stored video signal on the processing target reference line at the previous time point and the stored video signal on the horizontal scanning line in the processing target range at the subsequent time point. Using the corresponding point search method of, the horizontal scanning line corresponding to the processing target reference line at the previous time point is extracted within the processing target range at the later time point, and a plurality of horizontal scanning lines on the processing target reference line at the previous time point are extracted. A horizontal movement vector between two consecutive time points before and after the point is obtained. Extraction of the horizontal scanning line at the later time point corresponding to the reference line to be processed at the earlier time point is performed as follows. That is, first, with respect to the processing target reference line at a later time point, a known point is used to search for a corresponding point with the processing target reference line at a previous time point. Then, if a correspondence relation is found between them, the processing target reference line at the later time point is set as the horizontal scanning line corresponding to the processing target reference line at the earlier time point. If no correspondence is found between the processing target reference line at the earlier time point and the processing target reference line at the later time point, within the processing target range at the later time point, the vertical direction from the processing target reference line The horizontal scanning line at each position is sequentially searched for corresponding points with respect to the reference line to be processed at the previous time point while shifting the position one by one, and when the corresponding relationship is found, the horizontal line is detected. The scan line is a horizontal scan line corresponding to the reference line to be processed at the previous time. Then, the movement vector of each point is obtained from the horizontal displacement of the corresponding point of the processing target reference line at the previous time point and the corresponding horizontal scanning line at the subsequent time point. The determination and output of the presence or absence of the approach obstacle in step 7 are performed as follows. That is, first, as described above, based on the movement vector of each point on the processing target reference line at the previous time point, the background portion is removed,
A portion in which the magnitude of the movement vector continuously changes is extracted and used as an obstacle detection portion. If there is no obstacle detection part, that is, if no obstacle is detected, it is determined that there is no approach obstacle, and a signal indicating no approach obstacle is output. If there is an obstacle detection part, the first movement vector at the end on the first side of the left and right sides of the obstacle detection part
And the magnitude of the movement vector in the first lateral direction of the movement vector at the end on the second side of the left and right sides of the obstacle detection portion is determined. When this difference is a positive value First, it is determined that the obstacle is approaching. Note that either the left side or the right side may be the first side, but in the following description, the left side is the first side. If it is determined that the obstacle is in the approaching state, a signal indicating that the approaching obstacle is present is output, and otherwise, a signal indicating that the approaching obstacle is not present is output. When the process of step 7 is completed, the video signal on the reference line to be processed and the upper and lower scanning lines centered on the reference line to be processed of the video signal at the later time stored is the video signal at the previous time. Is stored as (step 8). Then, when the predetermined time has elapsed from the imaging in step 4, the process returns to step 4 and the above operation is repeated. When the operation of the device is stopped, the process is completed, and when the device is activated again, the step 1
The following operations are performed.

Next, with reference to FIGS. 4 to 12, the operation of the above-described vehicle approaching state detection device for determining the presence or absence of an approaching obstacle will be described in more detail. 4 to 12 show the state of the obstacle and the movement vector at that time at two consecutive points before and after. (A) of FIG. 4 to FIG.
Fig. 7 shows television images of an obstacle at two points before and after consecutive times. In (a) of each of these figures, the broken line shows the image of the obstacle at the previous time point and the solid line shows the image of the obstacle at the later time point. 4 to 12 (b) show movement vectors of a plurality of points of the obstacle between two consecutive time points before and after. In (b) of each of these figures, the horizontal axis represents the horizontal position of the point on the processing target reference line at the previous time point, and the vertical axis represents the direction and magnitude of the movement vector at each point. That is, the upper side of these vertical axes represents the leftward movement vector, and the lower side thereof represents the rightward movement vector. In the following description, the movement vector at the left end of the obstacle detection portion (hereinafter simply referred to as the left end vector) is Va, and the movement vector at the right end (hereinafter simply referred to as the right end vector) is Vb.
And the absolute value of the magnitude of the left end vector Va is a (≧
0), the absolute value of the magnitude of the right end vector Vb is b (≧ 0)
And In addition, in (b) of FIGS. 4 to 12, only the movement vector of the obstacle detection portion is shown excluding the background portion.

FIG. 4 shows the case where the relative positional relationship between the host vehicle and the obstacle does not change between the previous time and the later time, that is, the obstacle does not approach or move away from the host vehicle. It represents. In this case, since the magnitudes of the movement vectors of the respective points of the obstacle on the processing target reference line at the previous time point are all 0, it is determined that there is no obstacle detection portion and there is no approaching obstacle.

FIG. 5 shows a case where the obstacle moves in parallel to the left side between the previous time point and the later time point without approaching or moving away from the host vehicle. In this case, both the left end vector Va and the right end vector Vb of the obstacle detection portion are directed to the left side, and the absolute value (| a |) of the left end vector Va and the absolute value (| b |) of the right end vector Vb. ) Are equal to each other. Therefore, the value (a-b) of the difference (a-b) between the left-side size (a) of the left-end vector Va and the left-side size (b) of the right-end vector Vb becomes 0, and it is determined that there is no approach obstacle. It

FIG. 6 shows a case where the obstacle moves in parallel to the right side between the previous time point and the later time point without approaching or moving away from the own vehicle. In this case, the left end vector Va and the right end vector Vb of the obstacle detection portion are both directed to the right, and the absolute value (| a |) of the left end vector Va and the absolute value (| b |) of the right end vector Vb. ) Are equal to each other. Therefore, the difference (ba) between the left-sided magnitude (-a) of the left-end vector Va and the left-sided magnitude (-b) of the right-end vector Vb becomes 0, and there is no approach obstacle. To be judged.

FIG. 7 shows a case in which the obstacle moves away from the host vehicle in the left direction between the previous time point and the later time point. In this case, the left end vector Va and the right end vector Vb of the obstacle detection portion are both directed to the left side, and the absolute value (| a |) of the left end vector Va is the absolute value (| b |) of the right end vector Vb. ) Less than. Therefore, the difference (a-) between the left-sided magnitude (a) of the left-end vector Va and the left-sided magnitude (b) of the right-end vector Vb.
b) has a negative value, and it is determined that there is no approach obstacle.

FIG. 8 shows a case where the obstacle moves away from the host vehicle in the right direction between the previous time point and the later time point. In this case, the left end vector Va and the right end vector Vb of the obstacle detection portion are both directed to the right side, and the absolute value (| a |) of the left end vector Va is the absolute value (| b |) of the right end vector Vb. ) Greater than. Therefore, the difference (ba) between the left-sided magnitude (-a) of the left-end vector Va and the left-sided magnitude (-b) of the right-end vector Vb becomes a negative value, and there is no approach obstacle. To be judged.

FIG. 9 shows a case in which an obstacle moves straight away from the host vehicle between the previous time and the later time. In this case, the left end vector Va of the obstacle detection portion is directed to the right side, and the right end vector Vb is directed to the left side. Therefore, the left-sided magnitude of the left end vector Va (-a) and the left-sided magnitude of the right end vector Vb (b)
The difference (-a-b) with is a negative value, and it is determined that there is no approach obstacle.

FIG. 10 shows a case in which an obstacle approaches the host vehicle in the leftward direction between the previous time point and the later time point. In this case, the left end vector Va and the right end vector Vb of the obstacle detection portion are both directed to the left side, and the absolute value (| a |) of the left end vector Va is the absolute value (| b |) of the right end vector Vb. ) Greater than. Therefore, the difference (a) between the left-sided magnitude (a) of the left-end vector Va and the left-sided magnitude (b) of the right-end vector Vb.
-B) has a positive value, and it is determined that there is an approach obstacle.

FIG. 11 shows a case where the obstacle approaches the vehicle in the right direction between the previous time point and the later time point. In this case, the left end vector Va and the right end vector Vb of the obstacle detection portion are both directed rightward, and the absolute value (| a |) of the left end vector Va is the absolute value (| b |) of the right end vector Vb. ) Less than. Therefore, the leftward magnitude Va of the left-end vector Va (-a)
And the difference (ba) between the right end vector Vb and the leftward size (-b) has a positive value, and it is determined that there is an approach obstacle.

FIG. 12 shows a case where an obstacle approaches the host vehicle almost straight between the previous time and the later time. In this case, the left end vector Va of the obstacle detection portion is directed to the left side, and the right end vector Vb is directed to the right side. Therefore, the left-sided magnitude of the left-end vector Va (a) and the left-sided magnitude of the right-end vector Vb (-b)
The difference (a + b) between and becomes a positive value, and it is determined that there is an approach obstacle. When the obstacle approaches the host vehicle almost straight as shown in FIG. 12, the left end vector Va is directed to the left side and the right end vector Vb is directed to the right side, so that the obstacle is directed almost straight to the host vehicle. You can also judge that you are approaching.

In the above-mentioned obstacle approaching state detecting device, since it is necessary to process only the video signals on some horizontal scanning lines within the processing target range including the processing target reference line, the hardware configuration is not complicated. However, the processing does not take time, and high-speed processing at 100 ms intervals is possible. And
Whether or not the image processing device (2) is in the approaching state, based on the horizontal movement vector between two points before and after consecutive points on the processing target reference line at the previous point Therefore, it is possible to surely detect whether or not the obstacle is in the approaching state, although the video signals of a few horizontal scanning lines are only processed. Moreover, since a plurality of horizontal scanning lines including a predetermined processing target reference line are set as the processing target range for the image of the two-dimensional imaging device, even when the camera (1) vertically vibrates, Within the processing target range, the horizontal scan line corresponding to the processing target reference line at the previous time can be extracted, the vertical movement vector component of the camera (1) can be canceled, and the processing target reference at the previous time can be canceled. A horizontal movement vector can be obtained for each point on the line. Further, the image processing device (2), on the above-mentioned processing target reference line at the previous time point, extracts a portion in which the magnitude of the movement vector continuously changes, and sets it as an obstacle detection portion. The first of the left and right sides
Of the magnitude of the movement vector in the first lateral direction at the end of the side (the left side in the above embodiment) and the second side (the right side in the above embodiment) of the left and right sides of the obstacle detection portion. The difference between the movement vector at the end and the magnitude in the first lateral direction is obtained, and when the difference is a positive value, it is determined that the obstacle is in the approaching state. In addition, it is possible to reliably and easily detect whether or not the obstacle is in the approaching state.

[0034]

As described above, according to the obstacle approaching state detecting device of the present invention, the approaching state of the obstacle can be detected with certainty, and the video signal on some scanning lines within a predetermined range. Since only the processing is required, the processing does not take time, high-speed processing is possible, the performance of the apparatus is improved, and the cost is reduced without complicating the hardware configuration.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram of an obstacle approaching state detection device showing an embodiment of the present invention.

FIG. 2 is a flowchart showing an example of the operation of the obstacle approaching state detection device of FIG.

FIG. 3 is an explanatory diagram showing an example of an initial television image.

FIG. 4 is an explanatory diagram showing a movement vector of each point on a television image and a processing target reference line.

FIG. 5 is an explanatory diagram illustrating another television image and a movement vector of each point on a reference line to be processed.

FIG. 6 is an explanatory diagram showing movement vectors of other points on a television image and a reference line to be processed.

FIG. 7 is an explanatory diagram showing still another television image and movement vectors of respective points on a processing target reference line.

[Fig. 8] Fig. 8 is an explanatory diagram illustrating movement vectors of other points on a television image and a processing target reference line.

[Fig. 9] Fig. 9 is an explanatory diagram illustrating still another television image and a movement vector of each point on a reference line to be processed.

FIG. 10 is an explanatory diagram illustrating movement vectors of other points on a television image and a reference line to be processed.

FIG. 11 is an explanatory diagram illustrating movement vectors of other points on a television image and a reference line to be processed.

[Fig. 12] Fig. 12 is an explanatory diagram illustrating movement vectors of still another television image and points on the processing target reference line.

[Explanation of symbols]

(1) Two-dimensional CCD camera (two-dimensional imaging device) (2) Image processing device (3) Processing target range setting unit (4) Motion detection processing unit (5) Approach obstacle presence / absence determination unit (L) Processing target reference line (S) Processing range Va, Vb movement vector

 ─────────────────────────────────────────────────── ─── Continued Front Page (72) Inventor Yoshihiro Goto 3-5-8 Minamisenba, Chuo-ku, Osaka Koyo Seiko Co., Ltd. (72) Inventor Tomoho Yoshida 3-5-8 Minamisenba, Chuo-ku, Osaka Koyo Within Seiko Co., Ltd. (72) Inventor Yoshifumi Obata 3-5-8 Minamisenba, Chuo-ku, Osaka City Koyo Seiko Co., Ltd.

Claims (2)

[Claims] 1. A two-dimensional image pickup device for continuously picking up an image in a predetermined visual field at fixed time intervals, and a video signal representing images of the two-dimensional image pickup device at two consecutive points before and after. The image processing device for determining whether or not the obstacle is in the approaching state is provided, and the image processing device,
A plurality of horizontal scanning lines including a predetermined processing target reference line is set as a processing target range, and a video signal on the processing target reference line at a previous time point and a horizontal line in the processing target range corresponding to the video signal at a subsequent time point are set. Using a video signal on the scanning line, a horizontal movement vector between two consecutive time points before and after the plurality of points on the processing target reference line at the previous time point is obtained, and an obstacle is detected based on the movement vector. And an obstacle approaching state detecting device for detecting whether the obstacle is approaching or not. 2. The image processing apparatus extracts, on the reference line to be processed at a previous time point, a portion in which the magnitude of the movement vector continuously changes, and sets it as an obstacle detection portion, and detects the obstacle. The magnitude in the first lateral direction of the movement vector at the end on the first side of the left and right sides of the portion, and the above-mentioned magnitude of the movement vector at the end on the second side of the left and right sides of the obstacle detection portion. The obstacle approaching state according to claim 1, wherein the obstacle approaching state is determined when a difference from the lateral size of 1 is obtained, and the obstacle is approaching when the difference is a positive value. Detection device.