JP2010019589A - Inter-vehicle distance detector, drive recorder apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an inter-vehicle distance measurement apparatus and a drive recorder apparatus for accurately detecting an inter-vehicle distance even if axles or road surfaces of a vehicle and the other vehicle are relatively angled in a short distance area. <P>SOLUTION: The inter-vehicle distance measurement apparatus includes: an imaging means 21 for capturing an image of the other vehicle 11; a plate image detecting means 31 for detecting a plate image 13 of a license plate 12 from image data including the other vehicle 11; plate size detecting means 32, 33 for determining a size of the license plate 12; an inter-vehicle distance detecting means 34 for detecting the inter-vehicle distance; a diagonal intersection detecting means 37 for detecting a diagonal intersection of the plate image; inclination angle detecting means 38, 39 for monitoring a movement of the diagonal intersection in a predetermined two-dimensional coordinate, and detecting an angle between the axles of the vehicle and the other vehicle or a relative inclination angle between the road surfaces of the vehicle and the other vehicle; and an inter-vehicle distance correcting means 35 for correcting the inter-vehicle distance based on the angle or the inclination angle. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an inter-vehicle distance detection device that detects an inter-vehicle distance from another vehicle, and more particularly to an inter-vehicle distance detection device and a drive recorder device that detect an inter-vehicle distance from a photographed license plate.

  A drive recorder device that records vehicle data of a driver and uses it for evidence at the time of abnormal approach to an obstacle has been put into practical use. Drive recorder devices may detect vehicle distance and record vehicle data triggered by vehicle distance or relative speed, but distance sensors such as millimeter wave radar and laser radar have constant accuracy regardless of the detected distance. Therefore, it is known that the shorter the distance to be detected, the larger the error. For this reason, in a drive recorder device that requires detection of the inter-vehicle distance when approaching, it is not sufficient as a sensor for detecting the inter-vehicle distance.

  In view of this, a technique for calculating a distance from a preceding vehicle using image data obtained by photographing the front of the vehicle has been proposed (see, for example, Patent Documents 1 and 2). Patent Document 1 describes an inter-vehicle distance detection device that recognizes a license plate classification number from image data, determines a prescribed size of the license plate, and calculates an inter-vehicle distance from the shooting size of the license plate in the image data. Yes.

In Patent Document 2, the distance between the vehicle and the heading to the preceding vehicle is calculated by a radar, the relative angle between the front vehicle and the host vehicle is calculated from the heading distance and the heading, and a license plate photograph calculated from this and image data is taken. A vehicle distance measurement method for calculating the distance to the license plate from the size is described.
JP-A-10-96626 JP 2002-122670 A

  However, the inter-vehicle distance detection device described in Patent Document 1 has a problem that it is only assumed that the vehicle ahead and the host vehicle travel in the same direction (without making a relative angle). For example, when the front vehicle stops at a relative angle to make a right or left turn, the inter-vehicle distance is the left rear end portion or the right rear end portion (hereinafter referred to as the closest portion) of the front vehicle. Appropriate driving assistance is difficult without measuring the inter-vehicle distance.

  The inter-vehicle distance measuring method described in Patent Document 2 detects a relative angle, but since the distance to the license plate is detected as the inter-vehicle distance, as in the inter-vehicle distance detection device described in Patent Document 1, The distance to the closest part cannot be detected.

  In view of the above problems, the present invention provides an inter-vehicle distance measuring device and a drive recorder device that can accurately detect an inter-vehicle distance even when the axle or road surface of another vehicle and the host vehicle are relatively angled in a short-distance region. The purpose is to provide.

  In view of the above-described problems, the present invention detects a plate image of the license plate from image capturing means for photographing another vehicle and image data obtained by photographing the other vehicle in an inter-vehicle distance detection device that detects a distance from the other vehicle. Plate image detecting means, plate size detecting means for determining the size of the license plate, inter-vehicle distance detecting means for detecting the inter-vehicle distance, diagonal intersection detecting means for detecting the diagonal intersection of the plate images, and predetermined two-dimensional Inclination angle detecting means for monitoring the movement of the diagonal intersection in the coordinates and detecting the angle formed by the axle of the other vehicle and the own vehicle or the relative inclination angle of the road surface of the other vehicle and the road surface of the own vehicle; And an inter-vehicle distance correcting means for correcting the inter-vehicle distance based on the inclination angle.

  It is possible to provide an inter-vehicle distance measuring device and a drive recorder device capable of accurately detecting an inter-vehicle distance even if the other vehicle and the axle or road surface of the host vehicle are relatively angled in a short distance region.

The best mode for carrying out the present invention will be described below with reference to the drawings.
FIG. 1 is a diagram for explaining the outline of detection of the inter-vehicle distance. In Japan, the license plate 12 has three sizes (22 × 44, 16.5 × 33, 12.5 × 23 [cm]), and the aspect ratio between the height H and the width W is 1: 2 Therefore, when the host vehicle 50 travels straight behind the forward vehicle 11 that travels straight as shown in the drawing (the two axles are parallel to each other), the plate image 13 of the license plate 12 photographed by the in-vehicle camera 21 of the host vehicle 50 is displayed. The aspect ratio is 1: 2. For this reason, the intersection point O of the diagonal lines of the plate image 13 comes to a position determined by an aspect ratio of 1: 2.

  For example, when the forward vehicle 11 turns to the left and the axles of the forward vehicle 11 and the host vehicle 50 are not parallel, the projection surface that the license plate 12 forms on the image sensor of the in-vehicle camera 21 becomes small. In this case, the height H is not affected by the angle α formed between the axles, but the width W is a function of the angle α formed (W · cos α), so the aspect ratio of the plate image 13 changes. For this reason, the intersection point O of the diagonal lines of the plate image 13 comes to a position determined by the aspect ratio H: W · cos α. Therefore, if the amount of movement of the diagonal intersection point O is detected, the angle α formed becomes clear. For example, when the movement amount is Hp (the aspect ratio is 1: 1), the angle α formed is 60 degrees.

  By the way, since the change in the aspect ratio occurs similarly at the time of a right turn, it is not clear whether the vehicle has made a left turn or a right turn even if the angle α formed is known. However, when making a left turn, the right side of the number plate 12 moves more than the left side as the front vehicle 11 turns left, and when turning right, the left side of the number plate 12 moves more than the right side when the front vehicle 11 turns right. In this case, the intersection of the plate images 13 also moves in the direction in which the vehicle has moved. Therefore, if the coordinate system is fixed with respect to the plate image 13 when the angle formed is zero, the intersection moves to the left in the case of a left turn and moves to the right in the case of a right turn. Therefore, it is possible to detect the direction of right / left turn from the direction of the movement vector.

  As will be described later, since the inter-vehicle distance k to the license plate 12 can be calculated from the size of the plate image 13, it is corrected to the closest portion 14 of the preceding vehicle 11 by correcting with the angle α forming the inter-vehicle distance k. The closest distance L can be calculated. When applied to a drive recorder apparatus, TTC can be calculated based on the closest distance L and recording can be started at an appropriate timing.

  Similarly, even when the preceding vehicle 11 travels on a slope, the projection surface that the license plate 12 forms on the image sensor of the in-vehicle camera 21 becomes small. In this case, the width W is not affected by the relative inclination angle β of the slopes, but the height H is a function of the inclination angle β (H · sin (90−β)). The ratio changes. Therefore, the intersection point O of the diagonal lines of the plate image 13 comes to a position determined by the aspect ratio H · sin (90−β): W.

  Therefore, as in the case of the right / left turn, the inclination angle β can be detected from the movement vector of the diagonal intersection point O by utilizing the fact that the diagonal intersection point O comes to a position determined by the aspect ratio. Further, when the forward vehicle 11 climbs the slope, the plate image 13 also moves upward together with the forward vehicle 11, so the diagonal intersection point O also moves upward. Therefore, if the coordinate system is fixed with respect to the plate image 13 when the angle formed is zero, the intersection point moves upward when climbing a slope, and the intersection point moves downward when going down. Therefore, it is possible to detect the direction of right / left turn from the direction of the movement vector.

  Even when the preceding vehicle 11 climbs or descends the slope, the inter-vehicle distance k to the license plate 12 can be calculated from the size of the plate image 13, and the inter-vehicle distance k is corrected by the inclination angle β. Thus, the closest distance L to the closest portion 14 of the forward vehicle 11 can be calculated. Then, based on the closest distance L, TTC can be calculated and recording of vehicle data can be started at an appropriate timing.

  FIG. 2 is a schematic configuration diagram of the inter-vehicle distance detection device 100. The inter-vehicle distance detection device 100 is applied to the drive recorder device 200, and provides the inter-vehicle distance k and the closest distance L as one of triggers for the drive recorder device 200 to start recording of vehicle data.

  The inter-vehicle distance detection device 100 and the drive recorder device 200 are controlled by the control unit 25. The control unit 25 is a computer having a CPU, ROM, RAM, ASIC (Application Specific Integrated Circuit), RAM, ROM, memory, and the like, and the CPU stores a program stored in the ROM or hardware such as an ASIC. The distance detection unit 27, the TTC detection unit 28, and the vehicle data recording unit 29 are realized by the following. In the figure, the inter-vehicle distance device 100 and the drive recorder device 200 are integrally configured, but the distance detection unit 27 may be realized by another electronic control unit (for example, a navigation ECU ((Electronic Control Unit)).

  The control unit 25 is connected to an in-vehicle camera 21, a vehicle speed sensor 22, a G sensor 23, a stop lamp switch 24, and vehicle data storage means 26 via an in-vehicle LAN such as a CAN (Controller Area Network) and a dedicated line. .

  The in-vehicle camera 21, for example, is mounted on an indoor room mirror and captures a region that extends horizontally in a predetermined angle range toward the front of the vehicle. The in-vehicle camera 21 outputs image data of a predetermined luminance gradation (for example, 256 gradations) by a photoelectric conversion element such as a complementary metal oxide semiconductor (CMOS) or a charge coupled device (CCD). In the present embodiment, the inter-vehicle distance from the front vehicle 11 in front of the host vehicle 50 is detected, but the inter-vehicle distance from the rear other vehicle can be detected in the same manner if a camera for photographing the rear is mounted.

  The vehicle speed sensor 22, the G sensor 23, and the stop lamp switch 24 are sensors that mainly output vehicle data to be recorded by the drive recorder device 200. The vehicle speed sensor 22 is, for example, an MR sensor that measures, as pulses, changes in magnetic flux when convex portions installed at regular intervals pass on the circumference of a rotor provided in each wheel of the host vehicle 50. The wheel speed is measured for each wheel based on the number of pulses per beat. The G sensor 23 is a vibration piece type gyro sensor formed by, for example, micromachining, which detects an electrical signal proportional to the acceleration in the vehicle length direction and the width direction. The stop lamp switch 24 is turned on / off in conjunction with the displacement of the lever of the brake pedal, and detects that the driver has braked the host vehicle 50 by operating the brake pedal (hereinafter simply referred to as a brake operation). The presence or absence of a brake operation may be detected by, for example, the master cylinder pressure. The vehicle data storage means 26 is a storage means for recording vehicle data, and includes, for example, a flash memory, an HDD (Hard Disk Drive), an MRAM (Magnetoresistive Random Access Memory), a portable memory card, and the like. The vehicle data storage means 26 stores vehicle data such as vehicle speed and acceleration and video (image data) taken by the in-vehicle camera 21 in time series. In addition to this, time, sound, weather, and driver status ( Drowsiness, looking aside) may be recorded.

  The drive recorder device 200 is a device that records a driving situation (hereinafter referred to as a data recording situation) that the driver is confident or confused, and continuously records the vehicle data while overwriting the vehicle data. Overwriting the vehicle data for a predetermined time before and after the detection is prohibited. In addition, although vehicle data may be recorded only in a data recording situation, recording omissions can be easily prevented by always recording vehicle data.

  When a data recording situation is detected in a state where TTC (Time To Collation) is equal to or less than a threshold value, the drive recorder device 200 stores vehicle data for a predetermined time (for example, about 20 to 60 seconds) including the time when the data recording situation is detected. Recording is performed, and the recorded portion is overwritten. In the present embodiment, the data recording state is detected by a brake operation, but an impact on the sudden handle or the host vehicle 50 may be used as a trigger. By recording an actual video or the like in the data recording situation, the driver's safety consciousness can be improved, and it can be used as evidence when the vehicle 11 comes into contact with the vehicle 11 ahead.

  The distance detection unit 27 detects the distance between the distance k to the license plate 12 of the preceding vehicle 11 from the size of the plate image 13 included in the image data and the movement vector of the diagonal intersection O, and the closest to the closest part 14. Detect the distance L. The distance detection unit 27 will be described later. The TTC detection unit 28 obtains the relative speed V with the preceding vehicle 11 from the inter-vehicle distance k or the closest distance L, and TTC (= closest distance L / relative speed V from the inter-vehicle distance k or the closest distance L and the relative speed V. ) By obtaining the TTC from the closest distance L, the data recording status can be maintained accurately. The vehicle data recording unit 29 records the vehicle data with overwriting prohibited when a data recording state is detected in a state where the TTC is equal to or less than a predetermined threshold (for example, 2 to 5 seconds).

[Distance detection unit 27]
The distance detection unit 27 will be described based on the functional block diagram of FIG. In the image data, the front vehicle 11 is photographed together with the license plate 12. A known method can be used to detect the license plate 12 from the image data. The license plate 12 is attached at easy-to-see positions on the front and rear surfaces of the vehicle, and since the background color is a single color, the contour of the license plate 12 can be specified from the edge image.

  The plate image detection unit 31 first detects an edge in the vertical direction from the entire image data. The obtained vertical edge is the boundary between the side surface of the forward vehicle 11 and the background. In addition, in order to make specification of the front vehicle 11 easy, you may detect the edge of a horizontal direction.

  In sequentially captured image data, the vertical and horizontal edges disappear intermittently, and the position and length of the disappearing portion vary. The same edge is detected almost continuously.

  Next, the plate image detection unit 31 specifies the area of the license plate 12 from the vertical edge and the horizontal edge of the area of the preceding vehicle 11 in the image data. When the edges are connected, a rectangular area with a fixed aspect ratio of 1: 2 is obtained. Therefore, for example, the license plate 12 can be specified by template matching using a standard template with an aspect ratio of 1: 2. Since the number plate 12 is often attached at the center in the width direction below the bumper, the number plate 12 can be identified early by template matching from this position. Once identified, the license plate 12 can be identified from the same position until the preceding vehicle 11 changes to another vehicle. The identified rectangular area is the plate image 13.

  The plate classification determination unit 32 determines the classification of the license plate 12. As described above, since the Japanese license plate 12 has three sizes, in order to detect the inter-vehicle distance from the size of the plate image 13 to the license plate 12, it is necessary to determine the classification of the license plate 12. is there. There are several methods for determining the classification of the license plate 12. For example, the vehicle communicates with the preceding vehicle 11 by inter-vehicle communication, and receives the size or classification information of the license plate 12 from the preceding vehicle 11. Alternatively, the plate image 13 may be subjected to OCR (Optical Character Recognition) processing, and the classification number described on the license plate 12 may be read. Further, from the shape of the edge in the vertical direction of the forward vehicle 11 obtained when the plate image 13 is extracted, an approximate vehicle type can be specified and the classification of the license plate 12 can be determined. Since the aspect ratios of the edge images of the large vehicle, the passenger car, and the two-wheeled vehicle are different from each other, the classification of the license plate 12 can be determined from the aspect ratio of the circumscribed rectangle of the edge image. A large license plate 12 is attached to a large vehicle such as a truck or a bus, a middle license plate 12 is attached to a passenger car or a light vehicle, and a small license plate 12 is attached to a motorcycle.

  The plate size determination unit 33 determines the plate size of the license plate 12 based on the classification information. For example, it has a database that associates classification numbers and plate sizes (long and horizontal lengths of large, medium, and small plates), and plates can be referenced by referring to classification information acquired by inter-vehicle communication or OCR processing. Get the size. Moreover, you may utilize the database which matched plate size with distinction, such as a large vehicle, a passenger car, and a two-wheeled vehicle.

  The straight line distance detection unit 34 detects the inter-vehicle distance to the license plate 12. FIG. 4 shows an example of a diagram for schematically explaining detection of the inter-vehicle distance x. f is a distance from the lens of the in-vehicle camera 21 to the imaging plane and is known because it is equal to the focal length. K is the inter-vehicle distance from the lens to the license plate 12.

  First, the linear distance detection unit 34 detects the height Hp and the width Wp of the plate image 13. Since the height Hp and the width Wp are the height and width of the license plate 12 in the plate image 13, they are determined from the number of pixels and the pixel pitch that the plate image 13 occupies on the image plane.

As shown in the figure, between the height Hp and width Wp of the license plate 12 and the height H and width W of the license plate 12 on the image plane, H: Hp = k: f, W: Wp = k: f Therefore, the inter-vehicle distance k can be calculated from the following equation.
k = f · (H / Hp) or k = f · (W / Wp)
The inter-vehicle distance k may be detected by a millimeter wave radar device. If the vehicle is relatively far away, the inter-vehicle distance k can be accurately detected by the millimeter wave radar device k.

  The aspect ratio detection unit 36 detects the aspect ratio of the plate image 13. When the aspect ratio is 1: 2, the angle α formed by the preceding vehicle 11 and the host vehicle 50 and the relative road surface inclination angle β may be zero. When the aspect ratio deviates from 1: 2, detection of the movement vector is started.

  The intersection detection unit 37 detects a diagonal intersection O from the plate image 13. FIG. 5A is an example of a diagram schematically illustrating the diagonal intersection point O detected from the plate image 13. The intersection detection unit 37 determines a coordinate system in which the plate image 13 is a two-dimensional plane with the aspect ratio of the plate image 13 being 1: 2. In FIG. 5A, the origin of the coordinate system is the diagonal intersection point O. This origin is determined by, for example, an average value for every 10 frames of image data having an aspect ratio of 1: 2, and when the aspect ratio fluctuates, the last determined origin and the outer edge of the plate image 13 are fixed.

  Therefore, as described with reference to FIG. 1, the diagonal intersection point O moves due to the change in the aspect ratio. For example, when the forward vehicle 11 turns right or left, the diagonal intersection O moves on the X axis, and when traveling on a slope, the diagonal intersection O moves on the Y axis.

  When the aspect ratio fluctuates from the 1: 2 state, the movement vector detection unit 38 starts detecting the movement vector. The aspect ratio changes from 1: 2 to “less than 1: 2” in the case of a right turn or a left turn, and “less than 1 vs. 2” occurs when the vehicle 11 travels on a slope.

  FIG.5 (b) shows an example of the diagonal intersection O in the case of a left turn. In the case of a left turn, the diagonal intersection point O moves to the left (minus direction) on the X axis. Therefore, a right turn or a left turn can be distinguished from the direction of movement of the diagonal intersection point O.

  As shown in FIG. 5B, “Wp−Wp (α) = x” obtained by subtracting the width Wp (α) of the plate image 13 after starting the left turn from the original width Wp is the length of the movement vector. .

  As described with reference to FIG. 1, the width Wp (α) of the plate image 13 after the left turn is started is a function of the angle α, so Wp · cos α = Wp (α). If Wp (α) = Wp−x is used, the length x of the movement vector can be used instead of Wp (α), and the deformation can be made as Wp · cos α = Wp−x.

Therefore, the formed angle α can be obtained by the following equation.
α = cos ⁻¹ ((Wp−x) / Wp)
Here, since Wp = 2Hp, by substituting 2Hp, it is possible to obtain an expression that is not easily influenced by the inter-vehicle distance k that changes every moment.
α = cos ⁻¹ ((2Hp−x) / 2Hp) (1)
Incidentally, as shown in FIG. 5C, the license plate 12 also moves to the left as the forward vehicle 11 turns to the left, so the left side of the plate image 13 moves to the left. For this reason, if the origin of the diagonal intersection point O is fixed, the length x of the movement vector becomes long even if the angle α is the same, as is apparent from a comparison between FIGS. 5B and 5C. End up. Therefore, the movement vector detection unit 38 matches the left side of the plate image 13 with the left side of the plate image 13 when the origin is fixed.

  Furthermore, when making it correspond, the magnitude | size of the plate image 13 by the inter-vehicle distance k is correct | amended. Since the length x of the movement vector changes even if the angle α formed when the inter-vehicle distance k changes, the inter-vehicle distance k is calculated based on the height Hp, and the plate image 13 is enlarged or reduced according to the inter-vehicle distance k. For example, the enlargement / reduction ratio is determined so that the height Hp of the plate image 13 after the start of the left turn becomes the height Hp of the plate image 13 when the origin of the diagonal intersection O is fixed, and Hp and Wp Enlarge or reduce (α). With the above processing, the angle α formed from the length x of the movement vector can be detected.

  In addition, although it may be calculated using the equation (1) in this way, since the relationship between the angle α and the length x of the movement vector is fixed, the angle α that corresponds to the length x of the movement vector corresponds beforehand. In addition, it is easy to detect the angle α made by storing them. For this reason, the inter-vehicle distance detection device 100 has an angle / tilt angle table 40 formed. The length x of the movement vector when registering in the formed angle / tilt angle table 40 is not the size of the plate image 13 when the origin is fixed, but the length of the movement vector in the plate image 13 having a predetermined size. X. For this reason, the inclination angle detection unit 39 normalizes the size of the plate image 13 so as to have a predetermined height Hp, and then makes an angle / inclination angle table 40 based on the length x of the detected movement vector. The inclination angle α can be acquired by referring to. With the above processing, the angle α formed from the length x of the movement vector can be detected. Since the relationship between the angle α and the movement vector varies depending on the plate size, the angle / tilt angle table 40 formed is provided for each plate size.

The inclination angle β may also be calculated using an equation, or may be calculated from the formed angle / inclination angle table 40. When the equation is used, the inclination angle β can be obtained by the following equation. The length y of the movement vector is “Hp−Hp (α) = y”.
β = 90−sin ⁻¹ (Hp−y / Hp)
β = 90−sin ⁻¹ (Wp−2y / Wp) (2)
The inter-vehicle distance correction unit 35 detects the closest distance L to the closest part 14 based on the inter-vehicle distance k and the formed angle α or the inclination angle β.

  6A is an example of a diagram for schematically explaining calculation of the closest distance L using the angle α formed, and FIG. 6B is an example for schematically explaining calculation of the closest distance L using the inclination angle β. The closest distance L is shorter than the inter-vehicle distance k by an approach distance ΔL1 at which the rear end portion of the forward vehicle 11 approaches the host vehicle side due to an angle α formed as illustrated. The approach distance ΔL1 is ΔL1 = (1/2) · m · sin α, where m is the width of the vehicle 11 ahead. After the inter-vehicle distance k is calculated, the vehicle width m of the front vehicle 11 is based on the same principle as in FIG. 4, and the image in the width direction of the front vehicle 11 occupies the inter-vehicle distance k, the focal length f, and the imaging element. It can be calculated from the number of pixels and the pixel pitch. Inter-vehicle distance correction unit 35 In this way, the closest distance L (= inter-vehicle distance k−ΔL1) may be calculated, or a constant stored in advance may be used.

  When the traveling road surface of the preceding vehicle 11 and the host vehicle 50 has a relative inclination angle β, the closest distance L is an approaching distance by which the upper trunk of the preceding vehicle 11 approaches the own vehicle side by the inclination angle β as shown in the figure. ΔL2 is shorter than the inter-vehicle distance k. The approach distance ΔL2 is ΔL2 = n / tan (90−β) where n is the length from the number plate 12 of the preceding vehicle 11 to the trunk upper portion. The length from the license plate 12 to the trunk upper part can be calculated by the same principle as in FIG. As described above, the inter-vehicle distance correction unit 35 calculates the closest distance L (= inter-vehicle distance k−ΔL2).

  The TTC detection unit 28 obtains the relative speed V from the closest distance L for each image data, and detects the TTC from this and the closest distance L. Since the TTC is calculated from the distance and the data recording state is detected, the recording of the vehicle data can be started at an appropriate timing.

[Recording of vehicle data when front vehicle 11 turns left at intersection]
FIG. 7A is an example of a diagram schematically illustrating the forward vehicle 11 that turns left at the intersection and the diagonal intersection O, and FIG. 7B is a flowchart illustrating a procedure for recording the vehicle data with overwriting prohibited. Respectively.

  I) As shown in the upper diagram of FIG. 7A, both the host vehicle 50 and the preceding vehicle 11 are traveling straight ahead. Therefore. The angle α between the axles of each other is zero.

  II) While traveling straight ahead, the host vehicle 50 calculates an inter-vehicle distance k for each photographed image data. That is, the plate size is obtained from the classification of the license plate 12, and the inter-vehicle distance k is calculated from the focal length f, the pixel pitch, and Wp detected from the number of pixels occupied by the plate image 13. During this time, the intersection detection unit 37 calculates the diagonal intersection O while constantly updating it.

  III) As shown in the lower diagram of FIG. 7A, when the forward vehicle 11 made a left turn at the intersection, the pedestrian crossed the pedestrian crossing and stopped in the middle of the left turn. Therefore, an angle α formed between the preceding vehicle 11 and the host vehicle 50 is generated. The aspect ratio detection unit 36 detects that the aspect ratio of the plate image 13 has deviated from 1: 2 at the initial stage of the left turn, and notifies the intersection detection unit 37 of it. Thereby, the intersection detection unit 37 fixes the origin, and plots the diagonal intersection O of the plate image 13 against the origin. Based on the moved diagonal intersection point O, the movement vector detection unit 38 obtains a movement vector and updates it for each image data. The direction in which the diagonal intersection point O moves is the direction in which the forward vehicle 11 is bent.

  As shown in the lower part of FIG. 7A, the diagonal intersection O moves to the left on the X-axis when the vehicle 11 turns left. During this time, the linear distance detector 34 calculates the inter-vehicle distance k based on Hp every time image data is captured. This is because Hp depends only on the inter-vehicle distance k and does not depend on the inclination angle α.

  IV) Next, the tilt angle detection unit 39 calculates the angle α formed using the equation (1). Further, the inter-vehicle distance correction unit 35 calculates the closest distance L using the angle α formed. For example, if the inter-vehicle distance k = 5 [m], the angle α = 30 degrees, and the vehicle width m = 2.0 [m], ΔL1 = 0.5 [m], so the closest distance L is about 4.5. [M].

  V) The TTC detector 28 detects the relative speed V from the time variation of the closest distance L, and calculates TTC from the closest distance L / relative speed V. When a brake operation is detected in a state where TTC is equal to or less than a predetermined threshold, the vehicle data recording unit 29 records vehicle data while prohibiting overwriting.

  As described above, by calculating the closest distance L, the TTC up to the closest portion 14 can be calculated with high accuracy, so that the data recording status can be detected without delay and the vehicle data can be recorded. In the laser radar, the error becomes larger as the relative distance is closer. However, in the present embodiment, the relative distance is corrected by ΔL1, so that the accuracy is not lowered even if the relative distance is short.

[Recording of vehicle data when the vehicle 11 ahead travels on a slope]
Recording of vehicle data when the preceding vehicle 11 travels on a slope will be described. FIG. 8A is an example of a diagram schematically illustrating the forward vehicle 11 traveling on the slope and the diagonal intersection point O, and FIG. 8B is a flowchart illustrating a procedure for recording the vehicle data with overwriting prohibited. Respectively. In FIG. 8, the same steps as those in FIG.

  As shown in I) II), the host vehicle 50 calculates an inter-vehicle distance k for each photographed image data. During this time, the intersection detection unit 37 calculates the diagonal intersection O while constantly updating it.

  III) As shown in the lower diagram of FIG. 8 (a), when the forward vehicle 11 reaches the slope with the inclination angle β, the vehicle is decelerated and the inter-vehicle distance k is shortened. Further, the height Hp (α) of the plate image 13 is shortened by the inclination angle β. The aspect ratio detection unit 36 detects that the aspect ratio of the plate image 13 is deviated from 1: 2 at the entrance of the slope, and notifies the intersection detection unit 37 of it. Thereby, the intersection detection unit 37 fixes the origin, and plots the diagonal intersection O of the plate image 13 against the origin. Based on the moved diagonal intersection point O, the movement vector detection unit 38 obtains a movement vector and updates it for each image data. The direction in which the diagonal intersection point O has moved is the direction in which the forward vehicle 11 has moved.

  As shown in the lower part of FIG. 8A, the diagonal intersection point O moves upward on the Y-axis as the forward vehicle 11 climbs up. During this time, the linear distance detector 34 calculates the inter-vehicle distance k based on Wp every time image data is captured. This is because Wp depends only on the inter-vehicle distance k and does not depend on the inclination angle β.

  IV) Next, the tilt angle detection unit 39 calculates the tilt angle β using Equation (2). Further, the inter-vehicle distance correction unit 35 calculates the closest distance L using the inclination angle β. For example, if the inter-vehicle distance k = 5 [m], the inclination angle β = 30 degrees, and the length from the number plate 12 of the forward vehicle 11 to the trunk upper portion is n = 0.5 [m], ΔL2 = 0.28 [ m], the closest distance L is about 4.72 [m].

  V) The TTC detection unit 28 detects the relative speed V from the time change of the closest distance L, and calculates TTC. When a brake operation is detected in a state where TTC is equal to or less than a predetermined threshold, the vehicle data recording unit 29 records vehicle data while prohibiting overwriting.

[Recording of vehicle data when the preceding vehicle 11 enters a slope going up to the right]
A description will be given of recording of vehicle data when the preceding vehicle 11 travels on a slope with turning. FIG. 9 (a) is an example of a diagram schematically illustrating the forward vehicle 11 and the diagonal intersection O running on a slope that goes up to the right, and FIG. 9 (b) shows a procedure for recording the vehicle data with overwriting prohibited. The flowchart figure to show is shown, respectively. In FIG. 9, the same steps as those in FIG.

  As shown in I) II), the host vehicle 50 calculates an inter-vehicle distance k for each photographed image data. During this time, the intersection detection unit 37 calculates the diagonal intersection O while constantly updating it.

  III) As shown in the middle diagram of FIG. 9A, when the forward vehicle 11 reaches a slope that rises to the right at an inclination angle β, the vehicle is decelerated and the inter-vehicle distance k is shortened. At this time, the front vehicle 11 rarely travels on the slope while maintaining the aspect ratio of Hp (β) and Wp (α) of 1: 2, and the aspect ratio of the plate image 13 at the entrance of the slope is from 1: 2. The deviation is detected and notified to the intersection detection unit 37.

  Thereby, the intersection detection unit 37 fixes the origin, and plots the diagonal intersection O of the plate image 13 against the origin. Based on the moved diagonal intersection point O, the movement vector detection unit 38 obtains a movement vector and updates it for each image data. The direction in which the diagonal intersection point O has moved is the direction in which the forward vehicle 11 has moved.

  As shown in the middle diagram of FIG. 9A, the diagonal intersection O moves diagonally upward to the right due to the right turn of the forward vehicle 11 and the uphill. During this time, the linear distance detector 34 calculates the inter-vehicle distance k based on Wp (α) and Hp (α) every time image data is captured. For example, an average value of the inter-vehicle distance k obtained from both is set as the inter-vehicle distance k.

  As shown in FIG. 9, when the cornering and the slope traveling occur simultaneously, the diagonal intersection point O does not move on the X axis or the Y axis. Also, since both Wp (α) and Hp (β) vary, the other length cannot be calculated from one length. Therefore, when the turning traveling and the climbing or descending traveling occur at the same time, the angle α and the inclination angle β formed from the formed angle / inclination angle table 40 are obtained. Since the movement vector can be decomposed into the X axis and the Y axis, the formed angle α and the inclination angle β can be calculated respectively.

  As shown in the lower diagram of FIG. 9A, when the X component of the movement vector is a and the Y component is b, the angle α formed in advance can be detected from the X component a with the length x of the movement vector. , The inclination angle β can be detected from the Y component b.

  IV) Next, the inter-vehicle distance correction unit 35 calculates the closest distance L using the angle α formed, and calculates the closest distance L using the inclination angle β. As described above, the inter-vehicle distance k = 5 [m], the formed angle α = 30 degrees, the inclination angle β = 30 degrees, the vehicle width m = 2.0 [m], from the number plate 12 of the forward vehicle 11 to the upper trunk Is n = 0.5 [m], the closest distance L is 4.5 [m] and about 4.72 [m], respectively.

  V) The TTC detection unit 28 detects the relative speed V from the time change of the closest distance L, and calculates TTC. The relative speed V is the same regardless of which of the two closest distances L is used. However, the closest distance L differs depending on, for example, which of the left and right rear ends and the upper part of the trunk is the closest part 14. , TTC is calculated using two closest distances L.

  Then, when a brake operation is detected when one of the two TTCs is equal to or less than a predetermined threshold, the TTC detection unit 28 records the vehicle data while the vehicle data recording unit 29 prohibits overwriting.

  As described above, the inter-vehicle distance detection device 100 according to the present embodiment is the closest to the closest part 14 from the angle α formed by the preceding vehicle 11 and the host vehicle 50 and the slope angle β of the slope to the preceding vehicle 11. The distance L can be detected with high accuracy. For this reason, according to TTC with respect to the nearest part 14, recording of vehicle data can be started at an appropriate timing.

It is a figure explaining the outline of detection of distance between vehicles. It is an example of the schematic block diagram of the inter-vehicle distance detection apparatus. It is an example of a functional block diagram about a distance detection part. It is an example of the figure which illustrates detection of inter-vehicle distance x typically. It is an example of the figure which illustrates the diagonal intersection O detected from a plate image typically. It is an example of a diagram for schematically explaining the calculation of the closest distance L using an angle α and the closest distance L using an inclination angle β. It is an example of the figure which illustrates the front vehicle and the diagonal intersection O which turn left at an intersection typically. It is an example of the figure which illustrates the front vehicle and diagonal intersection O which drive | work a slope, typically. It is an example of the figure which illustrates the front vehicle and diagonal intersection O which drive | work a slope going up right.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Front vehicle 12 License plate 13 Plate image 14 Nearest part 21 Car-mounted camera 26 Vehicle data storage means 31 Plate image detection part 35 Inter-vehicle distance correction part 37 Intersection detection part 38 Movement vector detection part 39 Inclination angle detection part 50 Own vehicle 100 Between vehicles Distance detection device 200 Drive recorder device

Claims (8)

In the inter-vehicle distance detection device that detects the distance to other vehicles,
Photographing means for photographing other vehicles;
Plate image detection means for detecting a plate image of the license plate from image data obtained by photographing another vehicle;
Plate size detection means for determining the size of the license plate;
An inter-vehicle distance detecting means for detecting an inter-vehicle distance;
Diagonal intersection detection means for detecting a diagonal intersection of the plate image;
Inclination angle detection means for monitoring the movement of the diagonal intersection in predetermined two-dimensional coordinates and detecting the angle formed by the axle of the other vehicle and the own vehicle or the relative inclination angle of the road surface of the other vehicle and the own vehicle. When,
An inter-vehicle distance correcting means for correcting the inter-vehicle distance based on the angle or the inclination angle formed;
A vehicle-to-vehicle distance detection device comprising: The inter-vehicle distance correction means corrects the inter-vehicle distance and detects the closest distance to the closest part of another vehicle with respect to the own vehicle.
The inter-vehicle distance detection device according to claim 1. The inclination angle detection means detects the traveling direction of another vehicle from the moving direction of the diagonal intersection, and detects the angle formed or the inclination angle from the movement amount of the diagonal intersection.
The inter-vehicle distance detection device according to claim 1 or 2. A table in which the movement amount is registered in association with the angle formed or the inclination angle;
The inclination angle detection means detects the angle formed or the inclination angle with reference to the table based on the movement amount;
The inter-vehicle distance detection device according to any one of claims 1 to 3. When the diagonal intersection moves in a substantially horizontal direction with respect to the plate image,
The tilt angle detection means detects the angle formed from the ratio of the width of the plate image after movement to the width of the plate image before the diagonal intersection moves.
The inter-vehicle distance detection device according to claim 1. When the diagonal intersection moves in a direction substantially perpendicular to the plate image,
The tilt angle detection means detects the tilt angle from the ratio of the height of the plate image after movement to the height of the plate image before the diagonal intersection moves.
The inter-vehicle distance detection device according to claim 1. When the diagonal intersection moves in a manner that includes a horizontal component and a vertical component of the plate image,
The inclination angle detection means decomposes the movement amount into a horizontal component and a vertical component,
Detecting the angle formed by referring to the table based on the horizontal component and the tilt angle referring to the table based on the vertical component;
The inter-vehicle distance detection device according to claim 4. The inter-vehicle distance detection device according to any one of claims 1 to 7,
Vehicle data detection means for detecting vehicle data;
Data recording status detecting means for detecting the vehicle status for recording the vehicle data;
Vehicle data recording means for recording the vehicle data in a storage means;
The drive recorder apparatus characterized by having.

Images