JP6693314B2 - Vehicle approaching object detection device - Google Patents

Description

  The present invention relates to an approaching object detection device for a vehicle that detects an object approaching the vehicle based on an image captured by an imaging device (camera) fixed to the vehicle.

  An approaching object detection device for a vehicle (hereinafter, also referred to as a “conventional device”) that detects an object (approaching object) approaching the vehicle based on an image captured by an imaging device arranged at the front end of the vehicle. It is described in Patent Document 1. A conventional device is a group of vectors (that is, an optical flow vector, hereinafter simply referred to as “flow”) that represents a moving direction and a moving speed of a subject on an image in the same time interval based on two images captured at a predetermined time interval. Also called "vector".).

  The conventional device determines that the subject is an approaching object if the flow vector based on the subject included in the image obtained by photographing the left side region of the vehicle (the left side image) has a rightward horizontal component. Similarly, in the conventional device, if the flow vector based on the subject included in the image obtained by photographing the right side region of the vehicle (the right side image) has a leftward horizontal component, the subject is regarded as an approaching object. judge. In other words, the conventional apparatus determines that a subject corresponding to a flow vector having a horizontal component (horizontal direction horizontal component) toward an infinite point (FOE: Focus Of Expansion) indicating a straight direction in an image is an approaching object. To do.

  By the way, the flow vector acquired when the vehicle is traveling is different from the flow vector acquired when the vehicle is stopped. For example, when the vehicle is going straight, the horizontal component of the flow vector based on the subject included in the left side image increases to the left. Similarly, the horizontal component of the flow vector based on the subject included in the right-side image increases to the right. In other words, as the vehicle travels straight, the horizontal component of the flow vector in the direction opposite to the point at infinity increases.

  When the vehicle is traveling straight, the amount of change in the horizontal component of the flow vector (the amount of change in straight traveling) increases as the traveling speed of the vehicle increases. In addition, the straight-ahead change amount decreases as the distance between the vehicle and the subject increases. Even when the vehicle is traveling straight ahead, if the speed at which the approaching object approaches the vehicle is relatively high, the flow vector corresponding to the approaching object has a horizontal component in the central direction.

  On the other hand, when the direction of the vehicle changes (that is, when the vehicle runs while turning the steered wheels leftward or rightward), the approaching object may not be correctly determined. For example, when the vehicle is turning to the right (that is, when the vehicle changes direction as a result of the vehicle moving forward while turning the steered wheels to the right), each flow vector is compared to when the vehicle is not turning. The horizontal component of becomes larger to the left.

  The amount of change in the horizontal component of the flow vector (turning change amount) when the vehicle is turning increases as the turning speed (so-called yaw rate) of the vehicle increases. Note that the turning change amount does not change depending on the distance between the vehicle and the subject.

  For example, although the flow vector based on a certain subject included in the right side image originally does not have a component in the left direction (that is, a horizontal component in the center direction), the amount of change in turning caused by the right turn of the vehicle. Therefore, the flow vector based on the subject may have a leftward component. In this case, there is a possibility that “erroneous detection” may occur in which the conventional apparatus determines that the subject is an approaching object although the subject is not an approaching object.

  Therefore, the conventional device has a flow vector based on a subject (stationary object) that is far from the vehicle and stationary with respect to the road surface, a flow vector based on a stationary object that is close to the vehicle, a traveling speed of the vehicle, and , A correction flow vector (scalar value) based on the turning speed of the vehicle is acquired. The conventional apparatus removes the straight travel change amount and the turning change amount included in the horizontal component of each flow vector by adding the correction vector to the horizontal component of each flow vector. Therefore, according to the conventional device, it is possible to avoid the occurrence of a phenomenon (that is, erroneous detection) in which a subject (for example, a stationary object) that is not an approaching object is determined to be an approaching object.

JP, 2014-44521, A

  However, in order to acquire the correction flow vector, the conventional device needs to receive signals from a sensor (vehicle speed sensor) that acquires the traveling speed of the vehicle and a sensor (angular speed sensor) that acquires the turning speed. In addition, the conventional device needs to estimate the range (variation range) that can be taken by the straight line change amount and the turning change amount of the flow vector corresponding to the stationary object based on the acquired traveling speed and turning speed. Therefore, the calculation load when the conventional device sequentially detects approaching objects may become excessive.

  The present invention is an invention that addresses the above-mentioned problems, and one of its objects is to detect an approaching object of a vehicle that does not require a vehicle speed sensor and an angular velocity sensor and can detect an object approaching the vehicle by a simple calculation. It is to provide a device.

  An approaching object detection device for a vehicle (hereinafter, also referred to as “device of the present invention”) according to the present invention for achieving the above object includes an imaging device, a vector acquisition unit, a horizontal direction component value acquisition unit, and a demarcation threshold acquisition unit. , And an approaching object detection unit.

The imaging device (camera 20) is
An image including an object fixed to the vehicle body of the vehicle (10) and approaching the vehicle is captured.

The vector acquisition unit,
A first image acquired at a first time point by the imaging device,
Based on a second image acquired by the imaging device at a second time point after a predetermined time (imaging period Δt) from the first time point,
A plurality of values representing the starting point at the first time point, the moving amount and the moving direction from the first time point to the second time point of an arbitrary object photographed on the left side of the infinite point on the image plane of the first image The left optical flow vector of is acquired (step 710 of FIG. 7).

In addition, the vector acquisition unit, based on the first image and the second image,
A right optical flow vector representing the starting point at the first time point, the moving amount and the moving direction from the first time point to the second time point of an arbitrary object photographed on the right side of the point at infinity of the first image. Is acquired (step 710 in FIG. 7).

The horizontal component value acquisition unit,
If the magnitude of the horizontal component of the vector is zero, it becomes zero,
Positive if the vector has a horizontal component to the right,
Negative if the vector has a horizontal component to the left,
A horizontal component value that increases as the magnitude of the horizontal component of the vector increases is obtained for each of the left-side optical flow vector and the right-side optical flow vector (step 715 in FIG. 7).

The demarcation threshold acquisition unit,
A left side evaluation value (VL), which is the number of left side optical flow vectors in which the horizontal direction component value is larger than a specific value (Vc),
A right side evaluation value (VR), which is the number of right side optical flow vectors in which the horizontal direction component value is smaller than the specific value,
Is obtained for any particular value,
The specific value that minimizes the total evaluation value (VS), which is the sum of the left evaluation value and the right evaluation value, is acquired as the demarcation threshold value (Dth) (step 740 in FIG. 7).

The approaching object detection unit,
A subject corresponding to the left optical flow vector whose horizontal component value is larger than the demarcation threshold, and
A subject corresponding to the right optical flow vector whose horizontal component value is smaller than the demarcation threshold is detected as an approaching object (step 745 in FIG. 7).

  The horizontal component value is a value indicating the magnitude and direction of the horizontal component. The horizontal component value of the left optical flow vector is plotted on the left side of the origin of the two-dimensional coordinate plane as the vertical axis value, and the horizontal component value of the right optical flow vector is plotted on the right side of the origin as the vertical axis value. Then, a horizontal flow distribution map can be obtained (see FIGS. 4B, 5B, and 6B).

  If the vehicle is not turning, the points of the flow vector corresponding to the approaching object appearing to the left of the point at infinity in the image are plotted in the second quadrant of the horizontal flow distribution map. The points of the flow vector corresponding to the approaching object appearing to the right of the infinity point in the image are plotted in the fourth quadrant of the horizontal flow distribution map (see FIGS. 4B and 5B).

  However, when the vehicle turns, the horizontal component value uniformly increases or decreases. Therefore, there is a possibility that the points plotted in the second quadrant appear in the third quadrant and / or the points plotted in the fourth quadrant appear in the first quadrant (see FIG. 6 ( See B)).

  In other words, if the vehicle is not turning, and the point of the subject corresponding to the left optical flow vector is above the horizontal axis of the horizontal flow distribution map, then the subject corresponding to that point is determined to be an approaching object. it can. However, when the vehicle turns, a point corresponding to the approaching object may appear below the horizontal axis.

  Similarly, if the vehicle is not turning, if the point of the subject corresponding to the right side optical flow vector is below the horizontal axis of the horizontal flow distribution map, it can be determined that the subject corresponding to that point is an approaching object. .. However, when the vehicle turns, a point corresponding to the approaching object may appear above the horizontal axis.

  Therefore, in the device of the present invention, in the horizontal flow distribution map, if the point corresponding to the left optical flow vector is above the straight line (boundary line) parallel to the horizontal axis corresponding to the demarcation threshold, the subject corresponding to that point Is determined to be an approaching object. Similarly, if the point corresponding to the right side optical flow vector is below the demarcation line, the subject corresponding to that point is determined to be an approaching object. In other words, the demarcation line can be regarded as a straight line obtained by moving the horizontal axis to offset the increase / decrease in the horizontal component value due to the turning of the vehicle.

  The demarcation threshold is acquired based on the assumption that the area in which a stationary object is captured is larger than the area in which an approaching object is captured in the image captured by the imaging device. Specifically, the demarcation threshold is set so that the number of subjects (the number of points on the horizontal flow distribution map) determined to be approaching objects is the smallest. In other words, the demarcation threshold is set so that the number of subjects that are determined to be approaching objects is the smallest despite being moving objects other than stationary objects and approaching objects.

  According to the device of the present invention, an object approaching the vehicle can be detected with relatively high accuracy by a simple calculation without the need for the vehicle speed sensor and the angular velocity sensor.

  In the above description, in order to help understanding of the present invention, the names and / or reference numerals used in the embodiments are added in parentheses to the configurations of the invention corresponding to the embodiments described later. However, each component of the present invention is not limited to the embodiment defined by the name and / or code. Other objects, other features and attendant advantages of the present invention will be easily understood from the description of the embodiments of the present invention described with reference to the following drawings.

FIG. 1 is a schematic diagram of a vehicle (present vehicle) to which an approaching object detection device for vehicles (present detection device) according to an embodiment of the present invention is applied. It is an example of the image and the optical flow vector acquired by this detection device. It is a figure showing a mode that this vehicle approaches a Tojiji road. (A) is a diagram showing a horizontal direction component of a flow vector when the present vehicle is stopped, and (B) is a horizontal flow distribution diagram in this case. (A) is a figure showing the horizontal direction component of a flow vector when this vehicle enters straight road while going straight, and (B) is a horizontal flow distribution chart in this case. (A) is a diagram showing a horizontal component of a flow vector when the vehicle enters a T-shaped road while turning, and (B) is a horizontal flow distribution diagram in this case. 6 is a flowchart showing an approaching object detection processing routine executed by the detection device.

(Constitution)
Hereinafter, an approaching object detection device for a vehicle (hereinafter, also referred to as “present detection device”) according to an embodiment of the present invention will be described with reference to the drawings. The detection device is applied to a vehicle 10 whose schematic configuration is shown in FIG. The detection device includes a camera 20 and an ECU 30.

  The camera 20 is fixed to the center of the front end of the vehicle body of the vehicle 10. The horizontal angle of view (field of view) of the camera 20 is approximately 150 degrees from near the left horizontal direction to near the right horizontal direction, including the front of the vehicle 10 in the traveling direction.

  Specifically, the horizontal angle of view of the camera 20 is equal to the angle formed by the straight line Le and the straight line Re shown in FIG. The angle formed by the center axis Cs of the vehicle 10 in the front-rear direction and the straight line Le is equal to the angle formed by the center axis Cs and the straight line Re. The camera 20 outputs a signal representing the captured image to the ECU 30.

  The ECU 30 is an electronic circuit including a well-known microcomputer, and includes a CPU, a ROM, a RAM, an interface, and the like. The CPU reads data, performs numerical calculations, outputs calculation results, and the like by sequentially executing predetermined programs (routines). The ROM stores a program executed by the CPU, a look-up table (map), and the like. The RAM temporarily stores data.

  The ECU 30 is connected to the display device 41. The display device 41 is provided in a center console (not shown) provided in the vehicle interior of the vehicle 10. When the approaching object detection process described below is executed, the display device 41 continuously displays the images captured by the camera 20. That is, a moving image representing the front of the vehicle 10 is displayed on the display device 41.

  The display device 41 includes an operation switch (not shown). The driver of the vehicle 10 can select one of the on state and the off state of the approaching object detection process by operating the operation switch. In addition, the display device 41 includes a speaker (not shown).

(Outline of approaching object detection process-acquisition of flow vector)
When the approaching object detection process is in the ON state, the ECU 30 corresponds to the “object approaching the vehicle 10 (approaching object)” captured by the camera 20 based on the horizontal component of the optical flow vector (flow vector). "Subject" is detected. First, a method of acquiring a flow vector will be described.

  The ECU 30 causes the camera 20 to shoot an image each time a predetermined shooting cycle Δt (fixed value) elapses. The ECU 30 acquires a flow vector based on the image (first image) captured by the camera 20 and the image (second image) captured after the capturing period Δt after the capturing time of the first image.

  In other words, the first image is an image taken by the camera 20 before the second image by the shooting cycle Δt. The second image is the latest image taken by the camera 20. The time when the first image is captured is also referred to as “first time point” for convenience. The time when the second image is captured is also referred to as “second time point” for convenience.

  The ECU 30 uses a starting point at a first time point of an arbitrary subject (specifically, a part of the subject) photographed in both the first image and the second image and a movement amount from the first time point to the second time point. And a flow vector, which is a vector indicating the moving direction, is acquired for each of the plurality of subjects. That is, the ECU 30 acquires a plurality of flow vectors.

  More specifically, the ECU 30 divides each of the first image and the second image into rectangles of a predetermined size. That is, the ECU 30 handles each of the first image and the second image as a set of rectangles. The ECU 30 searches at which position in the second image each of the rectangles forming the first image appears. When the search is successful, a flow vector having the position of the rectangle (movement source) in the first image as the starting point and the position of the rectangle (moving destination) in the second image as the ending point is obtained. The ECU 30 executes this process for each of the rectangles that form the first image. As a result, multiple (very many) flow vectors are obtained.

  An example of a flow vector is indicated by the black arrow in FIG. In the example of FIG. 2, the vehicle 10 is moving forward (straight ahead) from the first time point to the second time point. The image shown in FIG. 2 is the first image. Each flow vector is acquired based on this first image and the “second image captured after the first image by the image capturing period Δt”. Therefore, each of the flow vectors represents the moving direction and the moving distance on the image plane in each shooting cycle Δt of the subject corresponding to the flow vector.

  By the way, in FIG. 2, the point at which each of the straight lines passing through the start point and the end point of each flow vector corresponding to a stationary subject (specifically, a building, a road surface, etc.) intersects is the “point of infinity FOE”. Also called. The point at infinity FOE can also be regarded as a point indicating the straight traveling direction of the vehicle 10 on the image plane shown in FIG.

  On the other hand, a straight line passing through the start point and the end point of the flow vector corresponding to the moving subject (specifically, a vehicle other than the vehicle 10) does not pass through the infinite point FOE. This is because, in addition to the movement of the subject on the image plane caused by the forward movement of the vehicle 10, the “movement of the subject on the image plane” due to the movement of the subject on the road surface occurs.

  In other words, the flow vector shown in FIG. 2 is a vector representing the movement of the subject on the image plane caused by the vehicle 10 moving forward, and the movement of the subject on the image plane caused by the movement of the subject. It can be understood as the sum of the vector and. For the sake of convenience, the former vector is also referred to as a “vehicle moving vector”, and the latter vector is also referred to as a “subject moving vector”. The subject movement vector corresponding to the stationary subject is a zero vector.

  When the ECU 30 divides each of the first image and the second image to obtain a set of rectangles in order to obtain the flow vector, a rectangle obtained by dividing the part on the left side of the infinite point FOE of the image is obtained. The number is equal to the number of rectangles obtained by dividing the part on the right side of the infinite point FOE of the image.

  In addition, the flow vector whose start point is located on the left side of the infinite point FOE of the image is also referred to as a “left side optical flow vector” for convenience. Similarly, a flow vector whose start point is on the right side of the infinite point FOE of the image is also referred to as a “right side optical flow vector” for convenience. Therefore, the number of right side optical flow vectors and the number of left side optical flow vectors are equal to each other.

(Detection of approaching object focusing on horizontal component)
Next, a method of extracting an approaching object based on the horizontal component of the flow vector will be described by taking the case where the vehicle 10 enters a T-shaped road as shown in FIG. 3 as an example. The other vehicle 51 and the other vehicle 52 are vehicles other than the traveling vehicle 10. The other vehicle 51 and the other vehicle 52 are approaching objects.

  When the vehicle 10 is stopped at the position Ps, the camera 20 captures an image as shown in FIG. To facilitate understanding, in this example, the ECU 30 divides the image captured by the camera 20 into rectangles of 10 horizontal × 8 vertical, and acquires a flow vector. In FIG. 4A, the horizontal component of the acquired flow vector is represented by an arrow.

  In the example of FIG. 4A, since the vehicle 10 is stopped, the own vehicle movement vector is a zero vector. Therefore, only the horizontal component of the flow vector corresponding to the other vehicle 51 and the other vehicle 52, which are moving objects, is indicated by an arrow in FIG. 4 (A). In other words, the flow vector shown in FIG. 4A is the horizontal component of the subject movement vector.

  FIG. 4B shows a distribution of horizontal components of the flow vector shown in FIG. In the distribution chart of FIG. 4B (hereinafter, also referred to as “horizontal flow distribution chart”), the horizontal axis represents the horizontal position of the start point position of the flow vector. The horizontal position of the point FOE at infinity corresponds to the position where the value on the horizontal axis of the horizontal flow distribution chart is “0”. The value on the horizontal axis of the horizontal flow distribution map increases as the starting point position of the flow vector moves to the right.

  The vertical axis of the horizontal flow distribution diagram represents the direction and magnitude of the horizontal component of the flow vector. When the magnitude of the horizontal component of the flow vector is "0", the value on the vertical axis of the horizontal flow distribution chart becomes "0". When the flow vector has a horizontal component in the right direction, the value on the vertical axis of the horizontal flow distribution chart becomes a positive value. When the flow vector has a horizontal component in the left direction, the value on the vertical axis becomes a negative value. The larger the horizontal component of the flow vector, the larger the value on the vertical axis. The value on the vertical axis of the horizontal flow distribution diagram is also referred to as "horizontal component value" for convenience.

  The values on the horizontal axis and the values on the vertical axis of the horizontal flow distribution chart are represented by the number of pixels (the number of pixels) of the camera 20. That is, the unit of the horizontal axis and the vertical axis is the number of pixels. For example, if the horizontal component of the flow vector is 100 pixels to the left (that is, if a certain object on the first image has moved to the left by 100 pixels on the second image, it appears). ), The value on the vertical axis of the horizontal flow distribution chart is -100 [pixels].

  In the example of FIG. 4A, since the start position of the flow vector corresponding to the other vehicle 51 is on the left side of the point FOE at infinity and has the horizontal component in the right direction, the corresponding points (five) are It is plotted in the second quadrant of the horizontal flow distribution map. On the other hand, since the start position of the flow vector corresponding to the other vehicle 52 is on the right side of the point FOE at infinity and has a horizontal component in the left direction, the corresponding points (3) are the fourth in the horizontal flow distribution diagram. It is plotted in the quadrant.

  Since the flow vectors that do not correspond to any of the other vehicles 51 and 52 (that is, the flow vectors corresponding to the stationary object) are all zero vectors, the points corresponding to these zero vectors on the horizontal axis of the horizontal flow distribution map. Is plotted.

  As described above, the other vehicle 51 and the other vehicle 52 are approaching objects. As can be understood from the examples of FIGS. 4A and 4B, the flow vector corresponding to the approaching object has a horizontal component (center-direction horizontal component) in the direction toward the infinite point FOE. Specifically, the flow vector corresponding to “another vehicle 51 on the left of the infinite point FOE” has a rightward horizontal component. On the other hand, the flow vector corresponding to “another vehicle 52 to the right of the point of infinity FOE” has a leftward horizontal component. In other words, when the vehicle 10 is stopped, the ECU 30 can extract, as an approaching object, the subject corresponding to the flow vector having the horizontal component in the direction toward the infinite point FOE.

  The example of FIG. 5 (A) represents the horizontal direction component of the flow vector acquired when the vehicle 10 advances straight ahead and enters an intersection as indicated by an arrow Aa in FIG. FIG. 5B is a horizontal flow distribution diagram corresponding to the example of FIG. In FIG. 5A, the own vehicle movement vector generated by the straight traveling of the vehicle 10 is superimposed on the subject movement vector shown in FIG. 4A.

  As can be understood from FIG. 5B, the vehicle movement vector has a leftward horizontal component on the left side of the infinite point FOE and a rightward horizontal component on the right side of the infinite point FOE. have. In addition, the magnitude of the own vehicle movement vector becomes larger as it moves away from the point FOE at infinity in the horizontal direction.

  In addition, the magnitude of the own vehicle movement vector increases as the vehicle speed of the vehicle 10 increases. As can be understood from FIGS. 5A and 5B, even if the own vehicle movement vector is superimposed on the subject movement vector, the flow vector corresponding to the other vehicle 51 has a horizontal component in the right direction, The flow vector corresponding to 52 has a horizontal component to the left. In other words, even when the vehicle 10 is traveling straight ahead, if the vehicle speed of the vehicle 10 is not excessive, the ECU 30 corresponds to the flow vector having the horizontal component in the direction toward the infinite point FOE. The subject to be extracted can be extracted as an approaching object.

  The example of FIG. 6A represents the horizontal direction component of the flow vector acquired when the vehicle 10 enters an intersection while making a turn as shown by an arrow Ab in FIG. FIG. 6B is a horizontal flow distribution diagram corresponding to the example of FIG. 6A. In FIG. 6 (A), the vector generated by the turning of the vehicle 10 (hereinafter also referred to as “own vehicle turning vector”) is the own vehicle movement vector and the subject movement vector shown in FIG. 5 (A). Are overlaid.

  The direction of the own vehicle turning vector is opposite to the direction in which the vehicle 10 turns. The own vehicle turning vector has the same direction and direction regardless of the position on the image plane. In the example of FIG. 6A, since the vehicle 10 is turning to the right, the direction of the own vehicle turning vector generated as a result of the turning is leftward.

  As can be seen from FIG. 6B, the point of the flow vector corresponding to the other vehicle 51 appearing in the second quadrant in FIG. 5B has moved to the third quadrant. Similarly, some of the flow vector points appearing in the first quadrant in FIG. 5B have moved to the fourth quadrant.

  In other words, as a result of the vehicle 10 turning, the flow vector corresponding to the approaching object does not have a horizontal component in the direction toward the infinite point FOE. That is, the ECU 30 cannot determine whether the subject corresponding to the flow vector is an approaching object based on whether the flow vector has a horizontal component in the direction toward the infinite point FOE.

  However, as understood from FIG. 6B, even if the vehicle 10 is turning, the flow vector corresponding to the other vehicle 51 is the flow vector corresponding to the stationary object on the left side of the infinite point FOE. The horizontal component to the right is large in comparison. Similarly, the flow vector corresponding to the other vehicle 52 has a larger leftward horizontal component than the flow vector corresponding to the stationary object on the right side of the infinite point FOE.

  Therefore, in FIG. 6B, the ECU 30 extracts a subject corresponding to the flow vector of a point above the demarcation line Ld and to the left of the infinite point FOE as an approaching object. Similarly, the ECU 30 extracts a subject corresponding to the flow vector of a point below the demarcation line Ld and to the right of the infinite point FOE as an approaching object.

  The demarcation line Ld is a straight line parallel to the horizontal axis corresponding to the magnitude of the own vehicle turning vector in the horizontal flow distribution map. When the vehicle 10 is turning to the right, the demarcation line Ld appears below the horizontal axis of the horizontal flow distribution map. When the vehicle 10 is turning to the left, the demarcation line Ld appears above the horizontal axis.

  The magnitude of the demarcation threshold Dth, which is the distance (real number with a sign) between the demarcation line Ld and the horizontal axis of the horizontal flow distribution diagram, is the turning speed of the vehicle 10 (so-called yaw rate, and the vehicle 10 turns to the right). It becomes a positive value when there is) and it becomes larger. Specifically, the demarcation threshold value Dth has a negative value and its absolute value increases as the vehicle 10 turns to the right and the turning speed increases. On the other hand, the demarcation threshold value Dth has a positive value and its absolute value increases as the vehicle 10 turns left and the turning speed increases.

(Estimation of demarcation threshold)
The ECU 30 acquires the demarcation threshold Dth by executing the following steps (S1) to (S4).
(S1) In the horizontal flow distribution chart, the left side evaluation value VL, which is the number of points on the left side of the origin and whose vertical axis value is larger than the specific value (signed real value) Vc, is acquired.
(S2) In the horizontal flow distribution chart, the right evaluation value VR, which is the number of points on the right side of the origin and whose vertical axis value is smaller than the specific value Vc, is acquired.
(S3) The total evaluation value VS, which is the sum of the left evaluation value VL and the right evaluation value VR, is acquired (that is, VS = VL + VR).
(S4) The specific value Vc when the minimum value of the total evaluation values VS is obtained by repeatedly executing the above (S1) to (S3) while changing the specific value Vc until the minimum value of the total evaluation value VS is obtained. Is obtained as the demarcation threshold Dth.

  The left side evaluation value VL is the number of approaching objects on the left side of the infinite point FOE when the demarcation threshold Dth is assumed to be the specific value Vc. Similarly, the right evaluation value VR is the number of approaching objects on the right side of the infinite point FOE when the demarcation threshold Dth is assumed to be the specific value Vc.

  In general, in an image captured by the camera 20, the area in which a stationary object is captured is larger than the area in which an approaching object is captured. Therefore, when the total evaluation value VS is a value other than the minimum value and an attempt is made to extract an approaching object by the specific value Vc corresponding to the total evaluation value VS, it is determined that the stationary object is an approaching object. Is likely to occur. In other words, the demarcation line Ld corresponding to the specific value Vc (that is, the demarcation threshold Dth) that minimizes the total evaluation value VS reduces the possibility of erroneous determination.

  The demarcation threshold Dth can be considered to represent the direction and magnitude of the own vehicle turning vector. In other words, the demarcation threshold value Dth is the vehicle turning vector estimated by executing the steps (S1) to (S4).

(Specific operation)
Next, a specific operation of the ECU 30 will be described. When the approaching object detection process is ON, the CPU of the ECU 30 (hereinafter, also simply referred to as “CPU”) has a function of “approaching object detection” shown by a flowchart in FIG. 7 every time the imaging cycle Δt elapses. Processing routine ".

  Therefore, at an appropriate timing, the CPU starts the process from step 700 of FIG. 7, sequentially executes the processes of steps 705 to 730 described below, and proceeds to step 735.

Step 705: The CPU acquires the image captured by the camera 20. The CPU stores the acquired image in the RAM.
Step 710: The CPU acquires a flow vector based on the image (first image) acquired when this routine was last executed and the image (second image) acquired this time.
Step 715: The CPU extracts the horizontal direction component from the flow vector. That is, the CPU internally generates the horizontal flow distribution map.

  Step 720: The CPU sets the value of the variable i to the demarcation lower limit value Dmin. The demarcation lower limit value Dmin is a negative value and is the minimum value that the demarcation threshold value Dth can take. In other words, the demarcation lower limit value Dmin means that the vehicle 10 turns to the right and the magnitude of the turning speed is a maximum value (when the magnitude of the turning speed becomes larger than this value, the vehicle 10 turns stably. It is the demarcation threshold value Dth when the turning speed at which it becomes impossible.

  On the other hand, the demarcation upper limit value Dmax described later is a positive value and is the maximum value that the demarcation threshold value Dth can take. Specifically, the demarcation lower limit value Dmin is the demarcation threshold value Dth when the vehicle 10 is turning left and the magnitude of the turning speed is the maximum value.

Step 725: The CPU executes the above (S1) to (S3) and acquires the total evaluation value VS when the above-mentioned specific value Vc is equal to the variable i.
Step 730: The CPU adds the demarcation value res to the variable i. The demarcation value res is set so that when the variable i increases by the demarcation value res, the demarcation line Ld in the horizontal flow distribution diagram moves up by one pixel.

  In step 735, the CPU determines whether the variable i is larger than the demarcation upper limit value Dmax. If the variable i is less than or equal to the demarcation upper limit value Dmax, the CPU makes a “No” determination at step 735 to proceed to step 725.

  On the other hand, if the variable i is larger than the demarcation upper limit value Dmax, the CPU makes a “Yes” determination at step 735 to proceed to step 740. In step 740, the CPU selects the variable i that minimizes the total evaluation value VS and sets the variable i as the demarcation threshold Dth.

  Next, the CPU proceeds to step 745 to extract the approaching object. Specifically, the CPU, in the horizontal flow distribution diagram, the subject corresponding to the flow vector of the point above the demarcation line Ld corresponding to the demarcation threshold Dth and to the left of the infinite point FOE, and the min. The object corresponding to the flow vector of the point below the field line Ld and to the right of the infinite point FOE is detected as an approaching object.

  Next, the CPU proceeds to step 750 to determine whether or not an approaching object has been detected. If the subject detected as the approaching object is present, the CPU makes a “Yes” determination at step 750 to proceed to step 755 to warn the driver of the vehicle 10 of the presence of the approaching object. More specifically, the CPU causes the color of the portion where the subject extracted as the approaching object is displayed to be a color different from other portions (in this example, in the image captured by the camera 20 and displayed on the display device 41). In, change to red). In addition, the CPU causes the speaker included in the display device 41 to generate a warning sound. After that, the CPU proceeds to step 795 to end the present routine tentatively.

  On the other hand, if no approaching object is detected, the CPU makes a “No” determination at step 750 to directly proceed to step 795.

  According to the present detection device, a vehicle speed sensor and an angular velocity sensor are not required, and even when the vehicle 10 is turning, a simple calculation based on the demarcation threshold Dth can be used to relatively accurately detect an object approaching the vehicle 10. Can be detected at.

  Although the embodiment of the approaching object detection device for a vehicle according to the present invention has been described above, the present invention is not limited to the above embodiment, and various modifications can be made without departing from the object of the present invention. .. For example, the vehicle 10 according to the present embodiment is equipped with one imaging device (camera 20). The vehicle 10 may be equipped with an imaging device that captures a left-side image and an imaging device that captures a right-side image (that is, a plurality of imaging devices).

  In addition, the ECU 30 according to the present embodiment acquires the optical flow vector by the so-called block matching method in step 710 of FIG. 7. The ECU 30 may acquire the optical flow vector by another method (for example, the gradient method).

  In addition, the ECU 30 according to the present embodiment acquires the flow vector based on the first image and the second image captured after the capturing period Δt of the capturing timing of the first image. However, the ECU 30 obtains the flow vector based on the first image and the second image captured “imaging period Δt × n (n is an integer of 2 or more)” after the capturing time of the first image. You may. Furthermore, the second image does not have to be the latest image.

10 ... Vehicle, 20 ... Camera, 41 ... Display device, 51 ... Other vehicle, 52 ... Other vehicle.

Claims (1)

An imaging device that captures an image including an object that is fixed to the vehicle body of the vehicle and approaches the vehicle,
Based on a first image acquired by the imaging device at a first time point and a second image acquired by the imaging device at a second time point after a predetermined time from the first time point, the second image A plurality of left sides representing the starting point at the first time point, the moving amount and the moving direction from the first time point to the second time point of an arbitrary object photographed on the left side of the infinite point on the image plane of one image. The optical flow vector and the starting point at the first time point, the moving amount and the moving direction from the first time point to the second time point of an arbitrary object photographed on the right side of the point at infinity of the first image are shown. A right side optical flow vector that represents, and a vector acquisition unit that acquires
Zero if the magnitude of the horizontal component of the vector is zero, positive value if the vector has a horizontal component to the right, negative if the vector has a horizontal component to the left. And a horizontal component value acquisition unit that acquires a horizontal component value that becomes larger as the magnitude of the horizontal component of the vector is larger for each of the left optical flow vector and the right optical flow vector. When,
A left evaluation value that is the number of the left optical flow vectors whose horizontal component value is larger than a specific value, and a right evaluation value that is the number of the right optical flow vectors whose horizontal component value is smaller than the specific value. , When obtained for any specific value, a demarcation threshold acquisition unit that acquires as a demarcation threshold the specific value that is the minimum total evaluation value that is the sum of the left evaluation value and the right evaluation value,
A subject corresponding to the left optical flow vector whose horizontal component value is larger than the demarcation threshold and a subject corresponding to the right optical flow vector whose horizontal component value is smaller than the demarcation threshold are detected as approaching objects. An approaching object detection unit that
An approaching object detection device for a vehicle, comprising:

Images