JP3351340B2 - Optical flow type backward information detection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to optical flow type rear information for detecting a vehicle approaching from behind (particularly from the rear side) using a movement vector (optical flow) on an image captured by a camera or the like. It relates to a detection device.

[0002]

2. Description of the Related Art Conventionally, there has been known a technique of calculating a movement vector at the same point in two images captured at different times by a camera or the like and detecting the presence of a moving object approaching from the calculated movement vector. , Especially in the automotive sector,
An image of the rear of the vehicle is captured by a camera or the like, and the rear side (for example,
There is known an optical flow type rear information detection device that detects the presence of another vehicle approaching from the rear of an adjacent lane).

[0003] This optical flow type rear information detecting device is an automatic cruise control device which has been developed recently.
That is, it is promising as an elemental technology of a device for determining the position and orientation of a vehicle with respect to a road on which the vehicle is traveling and the relative relationship with other vehicles and performing automatic driving control of the vehicle based on the information. A backward information detecting device has been proposed.

For example, the technique disclosed in Japanese Patent Application Laid-Open No. 8-83345 (first prior art) ensures that even when a moving object having a large occupied area is present in an image, an optical flow caused by the moving object can be reliably performed. This is a technique relating to a method of extracting and further determining whether or not the moving object is approaching based on the extracted optical flow. The technique disclosed in Japanese Patent Application Laid-Open No. Hei 9-18883 (second related art) includes a plurality of cameras or a rotatable camera to increase the monitoring area behind the vehicle to enhance safety. Things.

[0005]

As described above, various researches and developments have been conventionally made on the optical flow type rear information detecting device, and most of them mainly focus on grasping the movement of an object by the optical flow. At present, the purpose is that the optical flow itself has what information and how to utilize it has not yet been studied. For example, in the above-described first related art, when it is determined that a moving object is approaching, an alarm signal is output. However, here, the optical flow simply determines whether the moving object is approaching. It is merely used as a criterion for determining whether a moving object is approaching or not.

When attempting to change lanes to an adjacent lane, the driver checks the following vehicle traveling from behind the adjacent lane and judges whether there is any danger before changing lanes. At this time, the driver measures the inter-vehicle distance or the relative speed between the following vehicle and the own vehicle with a distance meter or a speedometer, and does not judge whether or not it is dangerous. The risk of the lane change is determined from the size and movement of the following vehicle that is seen by the driver through the rearview mirror or the like.

[0007] The size and movement of the following vehicle seen by the driver are also reflected in the optical flow, and when the following vehicle approaches, the number of optical flows corresponding to the following vehicle increases, and The size of the optical flow (the length of the vector) also increases. Therefore, if attention is paid to this point, it can be considered that it is possible to determine the degree of danger to a following vehicle approaching from behind through a change in the optical flow.

In the second prior art, the size of an optical flow in the divergent direction corresponding to the following vehicle is weighted, and if the weighted value exceeds a certain threshold value, it is determined that the flow is dangerous. Has become. However, in the second prior art, it is not clear how the optical flow and the degree of risk correspond to each other, and there is no description of how to assign weights. That is, it lacks the specific validity of associating the value obtained by weighting the size of the optical flow with the risk determination. In addition, it disappears or occurs due to the subtle change in the nature of the optical flow, that is, the imaging angle. No consideration is given to the fact that it does not change in nature.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-described problems, and focuses on usefulness as information relating to the danger of an optical flow, and is capable of always performing a stable danger determination. It is an object to provide a rear information detecting device.

[0010]

Therefore, in the optical flow type rear information detecting apparatus according to the present invention, the optical flow is calculated by the optical flow calculating means from a plurality of images behind the host vehicle which are imaged at a predetermined cycle by the imaging means. Te, but detects an object approaching the host vehicle, then the risk of magnitude of the average value of the optical flows corresponding to the object approaching the own vehicle by calculating means of the object approaching the
The risk is determined as a representative value indicating the degree, and the degree of risk of approaching the object is determined by the degree of risk determination means based on the average value of the magnitude of the optical flow.

[0011] Thus, the information on the risk possessed by the optical flow is effectively used, and a stable risk determination is always performed.

[0012]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the drawings and will be described embodiments of the present invention, FIGS. 1 to 8 are optical flow equation backward information detecting apparatus as an embodiment of the present onset Ming (hereinafter, simply backward information Detection device). As shown in FIG.
At a predetermined position (for example, above or below the rear window)
C as an imaging means for imaging the road condition behind the host vehicle
A CD camera 2 is provided, and image information (analog image signal) obtained by the CCD camera 2 is input to an ECU 10 provided inside the vehicle 1.

The ECU 10 comprises an A / D converter 3, a first frame memory 4A, a second frame memory 4B, an optical flow calculation means (DSP) 5, a calculation means 6, and a risk determination means 7, and includes a CCD. The analog image signal input from the camera 2 is first converted by the A / D converter 3 into a digital image signal.

The digital image signals processed by the A / D converter 3 are alternately and periodically stored in a first frame memory 4A and a second frame memory 4B. The storage period of the digital image signal is determined by an image processing capability such as calculation of an optical flow (hereinafter simply abbreviated as a flow), which will be described later. In this case, for example, the storage period is 5 for every 10 frames (1/30 second for one frame). It is assumed that a digital image signal is input to each of the frame memories 4A and 4B based on a phase difference between frames.

The optical flow calculation means 5 calculates an optical flow based on digital image signals stored in the frame memories 4A and 4B and having different imaging times. This optical flow calculation means 5
The following describes the optical flow calculation processing in. First, the first frame memory 4A
Image (first frame) at time t stored in
, An image of interest (pixel) is extracted from the image data, and subsequently, the time t + Δt [Δt = 5/30] stored in the second frame memory 4B
Second (for five frames)], a target image (pixel) of the captured image (second frame) is extracted. In setting the pixel of interest, a pixel having a certain threshold value or more may be selected, or a pixel having a higher luminance than the average of the luminance of surrounding pixels may be selected on the captured image.

Here, as shown in FIGS. 2A and 2B, it is assumed that an image of the following vehicle 101 approaching on an adjacent lane (hereinafter, referred to as an adjacent lane) from behind at different times. Assuming that the image 100A shown in FIG. 2A is a captured image at time t and the image 100B shown in FIG. 2B is a captured image at time t + Δt, for example, pixels P _A1 and P _{A2 of} interest from the respective images 100A and 100B. , P
_B1 and _PB2 are extracted. Here, for convenience of explanation,
Although the number of pixels of interest in each image is two, more pixels of interest are actually extracted.

Then, each pixel of interest P _A1 , P _A2 ,
P _B1, P _B2 to set the window _{S A1, S A2, S B1} , S B2 , the two screens 100A, window S _A1, S between 100B
_A2, to verify the identity of the S _B1, S _B2, 2 screens 100A,
The pixel of interest corresponding to 100B is obtained. As a method of verifying the window identity, a known method can be used.
As described in JP-A-83345, it can be verified by a correlation method or a gradient method.

When a corresponding pixel of interest between the two screens 100A and 100B is obtained, a vector connecting these pixels of interest becomes an optical flow. Here, since the windows S _A1 and S _B1 are the same and the windows S _A2 and S _B2 are the same, the pixels of _interest correspond to P _A1 and P _B1, and P _A2 corresponds to P _B2. The optical flows f ₁ and f ₂ are obtained by connecting them.

In this way, by verifying the correspondence also with respect to other corresponding pixels of interest, as shown in FIG. 2C, a plurality of optical flows (see FIG. 2) corresponding to the following vehicle 101 traveling behind. Optical flow group) F
Is obtained. In addition to the following vehicle 101,
Although optical flows corresponding to the surrounding scenery such as the road sign 102 can be obtained, since these move relatively away from the own vehicle (to converge in the direction of the so-called infinity point), the following flows are generated. The direction of the vector is opposite to that of the corresponding optical flow group F. Therefore, the optical flow calculation means 5 sets a threshold value in the direction of the vector, and calculates only the optical flow in the direction approaching the host vehicle.

By the way, how the optical flow F obtained here corresponds to the actual positional relationship between the vehicle and the following vehicle 101 will be described. The geometrical relationship shown in FIG. Can be represented by FIG.
It is a representation of the geometric relationship in the horizontal plane. The distance to the adjacent lane is W, the distance from the host vehicle 1 to the following vehicle 101 is Z, and the lateral distance on the screen of the CCD camera 2 is X, the focal length of the CCD camera 2 is D. Here, the following vehicle 101 between N frames (N / 30 seconds)
Is the relative movement distance of the CCD camera 2 on the screen, and dx is the horizontal movement distance of the CCD camera 2 on the screen.
x is expressed as follows. dx = W × D × (1 / (Z−dz) −1 / Z) (1) Further, the relative speed of the following vehicle 101A with respect to the own vehicle is represented by V
Assuming that s, the relative movement distance dz of the following vehicle 101 between N frames is (Vs / 30) × N, so the horizontal movement distance dx on the screen can be expressed as follows. dx = W × D × (30 / (30 × Z−Vs × N) −1 / Z) (2) In the frame memory, the horizontal movement distance dx on the screen is expressed in pixel units. The horizontal movement distance dPx (pixel) is
Assuming that the horizontal effective size of the CCD camera 2 is L1 and the number of horizontal pixels is n1, the following equation can be used. dPx (pixel) = dx × (n1 / L1) = W × D × (30 / (30 × Z−Vs × N) −1 / Z) × (n1 / L1) (3) Similarly, the vertical movement distance dPy (pixel) on the frame memory
Can be expressed by the following equation, where H is the installation height of the CCD camera 2, L2 is the vertical effective size of the CCD camera 2, and n2 is the number of vertical pixels. dPy (pixel) = H × D × (30 / (30 × Z−Vs / N) −1 / Z) × (n2 / L2) (4) Here, on the frame memory Using the horizontal movement distance dPx (pixel) and the vertical movement distance dPy (pixel), the magnitude dPa of the optical flow is as follows: dPa = (dPx ² + dPy ² ) ^1/2 (5) The horizontal movement distance dPx (pixel) and the vertical movement distance dPy (pixe) shown in Expressions (3) and (4) can be expressed.
l), the relative speed Vs, and the inter-vehicle distance Z, the larger the relative speed Vs and the shorter the inter-vehicle distance Z, the larger the optical flow size (hereinafter simply referred to as flow size) dP
It can be seen that a increases. That is, if the relative speed V
If s is constant, it can be determined that as the magnitude dPa of the flow increases, the inter-vehicle distance Z between the own vehicle 1 and the following vehicle 101 becomes smaller.

The calculation of the flow size dPa is performed by the calculating means 6, which calculates the size of all the optical flows (optical flow group) F corresponding to the following vehicle 101 traveling behind. DPa is calculated. The operation means 6
Calculates the total number M of optical flows (hereinafter, simply referred to as the total number of flows) corresponding to the following vehicle 101, that is, the total number of pixels of interest corresponding to the following vehicle 101. It should be noted that the total number M of the flows is as follows: A is the front area of the detected vehicle, Ac is the total number of pixels of the CCD camera 2, and CC is
Assuming that the horizontal angle of view of the D camera 2 is θ1 and the vertical angle of view of the CCD camera 2 is θ2, the following equation can be used. M = A × Ac / (4 × tan (θ1 / 2) × tan (θ2 / 2)) / Z ² (6) Further, the calculating means 6 calculates the following vehicle calculated earlier. The sum of the magnitudes dPa of the M flows corresponding to 101, that is, the total amount of optical flows (hereinafter, simply referred to as the total amount of flows) ΣdPa corresponding to the following vehicle 101 is calculated. Then, an average value dPave. (DPave. = ΣdPa / M) of the flow size dPa is calculated from the calculated total number M of flows and the total amount of flows ΣdPa.

The average value dPave. Of the magnitude of the flow calculated by the calculation means 6 is calculated by the danger determination means 7 so that the danger of the following vehicle 101 with respect to the own vehicle, that is, the own vehicle travels with the following vehicle 101. This is used as a reference for determining the degree of risk when changing lanes to the adjacent lane. That is, the risk determination means 7 determines the average value dP of the flow size.
When ave. exceeds a predetermined value, the lane change to the adjacent lane is determined to be dangerous and the alarm display monitor 8 is activated. The alarm display monitor 8 is configured to indicate the degree of danger based on the average value dPave. Of the magnitude of the flow in a stepwise manner. Warning.

Here, the average value dPave.
Before describing the significance of applying to the risk determination criterion, it is necessary to first describe a deceleration function generally used as risk information. Normally, the safe traveling distance Ds when two vehicles run in a row in front and back
Is defined as follows, assuming that the preceding vehicle applies the brake. Ds = Tm in ^{Vr × Tm + Vr 2/2} / Ar -Vf 2/2 / Af ······· (7) This equation is the time headway, Vf, the vehicle respectively Vr (i.e., preceding vehicle ), Speed of the following vehicle, Af
, Ar indicate the deceleration of the host vehicle (that is, the preceding vehicle) and the deceleration of the following vehicle, respectively.

Here, in normal driving on an expressway, it is said that the inter-vehicle time Tm is 1-2 seconds, the maximum deceleration that a general driver can step on is 6.0 m / s ² , and the inter-vehicle time Tm is 1 .
Assuming that the deceleration Af of the own vehicle is 6.0 m / s ² for 3 seconds, the deceleration Ar of the following vehicle can be expressed by the following equation. ^{Ar = (Vr 2/2)} / (Ds -1.3 × Vr + Vf 2/12) ····· (8) above equation, the vehicle traveling at the vehicle speed Vf is the maximum deceleration which is normally considered When the vehicle decelerates suddenly at (6.0 m / s ² ), the deceleration Ar required for the following vehicle traveling at the vehicle speed Vr, which is necessary to maintain a safe inter-vehicle distance Ds of 1.3 seconds between vehicles.
And the deceleration Ar is a deceleration function generally used as the degree of risk.

As can be seen from the equation (8), the danger expressed by the deceleration function Ar becomes smaller as the inter-vehicle distance Ds becomes shorter and the speed difference between the vehicles (ie, Vr-Vf) becomes larger. , Which is adapted to the driver's feeling in actual driving. Therefore, it is considered significant to set parameters for determining the degree of risk having characteristics similar to the degree of risk represented by the deceleration function.

By the way, the magnitude dPa of the flow represented by the equations (3) to (5) also increases as the relative speed Vs increases and the inter-vehicle distance Z decreases as described above. , The characteristic represented by the deceleration function Ar coincides with the characteristic. However, since the pixel of interest on the image, which is a reference for calculating the optical flow, is set based on the luminance difference, the pixel may suddenly disappear due to light reflection or the like. In such a case, since the optical flow also disappears, it is not possible to continuously evaluate the flow size dPa and determine the degree of risk only by focusing on any one optical flow.

On the other hand, it is conceivable to apply the total number M of flows and the sum ΣdPa of flows as parameters for determining the degree of risk. However, the total number M of flows is
As shown in the equation (6), at the point where the inter-vehicle distance Z becomes larger as the inter-vehicle distance Z becomes shorter, the characteristic coincides with the risk represented by the deceleration function Ar, but is affected by the front area A of the detected vehicle.
It is independent of the relative speed. For this reason, even if the total number M of flows increases, it cannot be distinguished whether it is because the inter-vehicle distance is short or because the detected vehicle is large. Also, in actual running, the total number of occurrences of optical flow is not only the front area of the detected vehicle,
It is also affected by the shape and color of the vehicle, and as described above, each optical flow is lost or generated due to the degree of light reflection, etc. Can not.

Similarly, for the total amount of flow ΣdPa,
As can be seen from the equations (3) to (6), the larger the relative speed Vs and the shorter the inter-vehicle distance Z, the greater the danger is represented by the deceleration function Ar. It is the same as the total number of flows M in that it is affected by the front area A of the detected vehicle, and it is not possible to make an accurate risk degree determination based on the relative speed Vs and the inter-vehicle distance Z.

On the other hand, the average value dP of the flow size
In ave., the influence of the front surface area A of the detected vehicle and the shape and color of the vehicle is eliminated by dividing the total flow 総 dPa by the total number M of flows. Further, since the average of the magnitudes dPa of a plurality of flows, even if one of the pixels of interest disappears from the screen, the average value d of the magnitudes of the flows d
The continuity of Pave. Is not lost or greatly changed.

Then, as shown in FIG. 4, the characteristic of the average value dPave. Of the magnitude of the flow with respect to the inter-vehicle distance Z coincides with the characteristic of the degree of risk represented by the deceleration function Ar. Here, the average value dPave.
Is the size d of the flow calculated from the equations (3) to (5).
The value of each variable in the equations (3) and (4) uses the following values. W: 3.5m, H: 1.4
m, D: 0.0015 m, N: 5 frames, n1: 25
6 pixels, n2: 240 pixels, L1: 0.00654
m, L2: 0.00489 m, Vs: 40 km / h. In calculating the degree of risk Ar represented by the deceleration function, the following values are used for the variables in equation (8). Vf
: 80 km / h, Vr: 120 km / h.

That is, the average value dPave.
The deceleration function Ar
As in the case of the danger expressed by the following equation, the danger determination is adapted to the driver's feeling in actual running, that is, the danger increases as the inter-vehicle distance Z decreases and the relative speed Vs increases. It has become. Here, FIG. 5 and FIG. 6 show the results of experiments in which the optical flow type rear information detecting device is actually mounted on a vehicle.

First, FIG. 5 shows a comparison between the theoretical value of the total number M of flows and the experimental value. The solid line shows the theoretical value of the total number M of flows represented by the equation (6). Vs
Is an experimental value when the speed is set to 40 km / h, and ▲ is an experimental value when the relative speed Vs is set to 5 km / h. FIG.
As shown in the figure, the shorter the inter-vehicle distance Z, the more the total number of flows M tends to increase. However, the total number M does not reach the theoretical value.
The smaller the relative speed Vs, the more anxious. Therefore, the experimental results also show that the total number M of flows is inappropriate as a risk. Although not shown in the experimental data, the same can be said for the total amount of flow ΣdPa.

FIG. 6 shows the average value d of the flow size.
This is a comparison between the theoretical value of Pave. And the experimental value, and the solid line represents the magnitude dPav of the flow represented by the equations (3) to (5).
e is the theoretical value, and □ is the average value dPave. of the magnitude of the flow obtained in the experiment. As shown in FIG. 6, even in a high risk situation, that is, a situation where the relative speed is large and the inter-vehicle distance is short, the average value dPave. Of the flow size is stably at the theoretical value, and the flow size is large. It can also be confirmed from this experimental result that the average value dPave.

Since the optical flow type backward information detecting device according to one embodiment of the present invention is configured as described above, the risk determination process is performed in a flow as shown in FIGS. 7 and 8, for example. As shown in FIG.
The rear image taken by the D camera 2 at time t is represented by A
A / D conversion by the / D converter 3 and the first frame memory 4A
(Step S100). Next, the rear image captured after the time Δt [for example, 5/30 seconds (for five frames)] is stored in the second frame memory 4B (step S200).

The optical flow calculation means 5 extracts a pixel of interest from each of the images stored in each of the frame memories 4A and 4B based on the luminance difference, and obtains a correspondence of the pixel of interest between the two images (step S300). .
Then, an optical flow as a vector connecting the corresponding pixels of interest is obtained. At this time, the image captured by the CCD camera 2 includes the surrounding scenery such as a road sign in addition to the following vehicle and the like. However, the optical flow calculation means 5 generates a vector of the corresponding pixel of interest between the two screens. A threshold is set in the direction, and only an optical flow corresponding to a moving object approaching the own vehicle such as a following vehicle is calculated (step S400).

The calculating means 6 calculates the degree of risk, that is, the average value dPave. Of the magnitude of the optical flow based on the optical flow corresponding to the following vehicle calculated by the optical flow calculating means 5 (step S500).
In calculating the risk (average value) dPave.
As shown in (1), first, the size dPa of each optical flow corresponding to the following vehicle is calculated (step S510), and the total number M of optical flows corresponding to the following vehicle is calculated (step S520). Next, the sum of the magnitudes dPa of the M flows corresponding to the following vehicles, that is, the total flow amount ΣdPa, is calculated (step S53).
0), total number M of calculated flows, total amount of flowsΣdPa
From the average value dPave. (DPav) of the flow size dPa
e. = ΣdPa / M) is calculated as the risk (step S530).

The risk determining means 7 calculates the risk calculated by the calculating means 6 (the average value dPav of the flow size).
e.) is compared with a predetermined value set in advance (step S60).
0) If the degree of danger exceeds a predetermined value, the lane change to the adjacent lane is determined to be danger and an alarm is displayed on the alarm display monitor 8. (Step S700) As described above, in the optical flow-type rear information detection device, the risk determination when the lane is changed to the adjacent lane is performed based on the optical flow size dPa, but the optical flow size dPa itself is used. Is not affected by the size of the following vehicle as in the total flow M or the total flow フ ロ ー dPa, and the optical flow size dP
Since the characteristics of the relative speed Vs and the inter-vehicle distance Z between a and the danger degree obtained by the deceleration function Ar match, the danger degree is uniform regardless of the size of the detected vehicle and adapted to the driver's feeling. A determination can be made.

Further, since the risk is determined not based on the magnitude dPa of one optical flow but on the basis of the average value dPave. Of the magnitudes of all optical flows corresponding to the following vehicles, Even if the optical flow disappears due to a change in the degree of light reflection, or the optical flow increases due to the approach of a following vehicle, a stable risk determination can always be performed without being affected by those. It is possible.

Particularly, in a situation where the risk is high, that is, when the inter-vehicle distance Z between the following vehicle and the own vehicle is very small and the relative speed Vs is large, the information on the risk is also stable. Is strongly required. In such a case, CC
Since the image of the following vehicle on the D camera 2 also becomes large, the total number of optical flows corresponding to the following vehicle as a sample of the averaging process also increases, and the magnitude dPa of the optical flow increases as the relative speed Vs increases. As a result, the calculation accuracy of the average value dPave. Of the optical flow size is improved, and more stable risk determination is possible.

It should be noted that the present invention is not limited to the above-described embodiment, and it is needless to say that various modifications can be made without departing from the spirit of the present invention.

[0041]

As described above in detail, according to the optical flow type rear information detecting apparatus of the present invention, the degree of danger is determined based on the average value of the magnitude of the optical flow. Uniform and stable danger for moving objects approaching from the rear without being affected by the loss or occurrence of optical flows due to imaging conditions such as changes and the size of the total number of optical flows due to the size of moving objects Can be determined.

Further, since the magnitude of the optical flow and the characteristics of the relative speed and the inter-vehicle distance between the degree of danger by the deceleration function are generally the same, it is possible to judge the degree of danger adapted to the driver's feeling.

[Brief description of the drawings]

FIG. 1 is a block diagram schematically showing a configuration of an optical flow type backward information detecting device as one embodiment of the present invention.

FIGS. 2A and 2B are diagrams for describing a method of calculating an optical flow according to the optical flow backward information detection device as one embodiment of the present invention, wherein FIG. 2A is a diagram illustrating a captured image at time t, and FIG. () Is a diagram showing a captured image at time t + Δt, and (c) is a diagram showing an optical flow obtained from the captured images of (a) and (b).

FIG. 3 is a diagram showing the actual positional relationship between the host vehicle and the following vehicle and the geometric correspondence between the optical flow and the optical flow type rear information detecting apparatus according to the embodiment of the present invention;

FIG. 4 is a diagram illustrating a relationship between an average value of an optical flow magnitude and an inter-vehicle distance according to an optical flow-type rear information detecting device according to an embodiment of the present invention, and a relationship between a risk degree and an inter-vehicle distance by a deceleration function. On a single graph,
It is a figure which shows those coincidences.

FIG. 5 is a diagram illustrating a correspondence between an experimental value and a theoretical value of the total number of optical flows.

FIG. 6 is a diagram showing the correspondence between the experimental value and the theoretical value of the average value of the optical flow magnitude according to the optical flow backward information detecting device as one embodiment of the present invention.

FIG. 7 is a flowchart for explaining an overall flow of a risk determination process of the optical flow type backward information detection device as one embodiment of the present invention.

FIG. 8 is a flowchart for explaining a flow of a process of calculating an average value of an optical flow size in the optical flow backward information detecting device as one embodiment of the present invention.

[Explanation of symbols]

Reference Signs List 2 CCD camera (imaging means) 3 A / D converter 4A First frame memory 4B Second frame memory 5 Optical flow calculation means (DSP) 6 Calculation means 7 Risk determination means 8 Alarm display monitor 10 ECU 100A, 100B Captured image P _A1 , P _A2 , P _B1 , P _B2 Pixels of interest S _A1 , S _A2 , S _B1 , S _B2 Window f ₁ , f ₂ Optical flow

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-7-50769 (JP, A) JP-A-9-86314 (JP, A) JP-A-9-18863 (JP, A) JP-A-5-8663 312819 (JP, A) JP-A-6-292203 (JP, A) JP-A-2-241855 (JP, A) JP-A-5-314262 (JP, A) (58) Fields investigated (Int. Cl. ^7, DB name) G06T 1/00 330 B60R 21/00 621 G06T 7/20

Claims (1)

(57) [Claims] 1. An image pickup means for picking up an image of the rear side of a host vehicle at a predetermined period, and an optical flow calculation means for calculating an optical flow from a plurality of images obtained by the image pickup means, wherein the optical flow calculation means calculates the optical flow. An optical flow-type rear information detecting device that detects a moving object approaching the host vehicle from the optical flow, an arithmetic unit that calculates an average value of an optical flow size corresponding to the moving object; has been optical flow the magnitude of the average value as a representative value that indicates the risk of approach of the moving object
An optical flow type backward information detecting device, comprising: a risk determining means for determining a risk of approaching the object based on the magnitude of the average value .