JP2006327495A - Infinite point determination device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an infinite point determination device capable of determining FOE without installing a target. <P>SOLUTION: In the infinite point determination device for determining a position of the infinite point in an image photographed by an on-vehicle camera fixed to a vehicle such that a bonnet is included in a photographing range, X coordinate FOE(X) of FOE and a first candidate value FOE(y1) of Y coordinate of FOE are determined from a position of an apex of the bonnet in the image (step S100) and a second candidate value FOE(y2) of Y coordinate of FOE and a roll angle of the on-vehicle camera 110 are determined from a position of a cut line CL of a headlight in the image (step S200). Further, the first candidate value FOF(y1) and FOE(y2) are corrected (step S300) and the first candidate value FOF(y1') and the second candidate value FOE(y2') after correction are averaged to determine the Y coordinate FOE(Y) of FOE (step S350). <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an infinite point determination device that determines the position of an infinite point in an image captured by a vehicle-mounted camera.

  When a camera is mounted on a vehicle, it may be necessary to determine an infinite point (Focus Of Expansion, hereinafter also referred to as FOE) in an image plane captured by the camera. For example, as described in Patent Document 1, when recognizing a dividing line (lane mark) that divides a lane such as a white line from an image captured by an in-vehicle camera, it is necessary to determine FOE. The position of the lane mark is used for control for automatically driving the vehicle along the lane mark, and for control for issuing a warning to the driver by detecting deviation from the lane of the vehicle.

  The FOE is a point where a pair of lane marks on both sides of a lane intersects with a straight line and extends to an infinite point, and the pair of lane marks intersect and is fixed on the horizon. The relative position between the host vehicle and the lane mark is recognized based on the FOE. Therefore, the FOE needs to be determined accurately.

  The FOE is automatically corrected based on the lane mark in the image picked up by the in-vehicle camera when the vehicle travels. As a premise, it is necessary to determine the initial position when the in-vehicle camera is attached. FIG. 1 is a diagram for explaining a conventional method for determining an initial position of an FOE. As shown in FIG. 1, when determining the FOE, the target 12 is installed at a predetermined position with respect to the vehicle 10.

  FIG. 2 is a diagram showing an image captured by the in-vehicle camera 14, and three targets 12 are installed as shown in FIG. The heights of these three targets 12 are all the same. The center target 12 is to be installed on the center line in the width direction of the vehicle 10. Further, the targets 12 on both sides are installed so that the straight line connecting the three targets 12 is perpendicular to the center line in the width direction of the vehicle 10 and the distance from the center target 12 is equal to each other. It has become.

  After the image shown in FIG. 2 is captured, the center position of each target 12 is detected by image processing. Since the target 12 is installed at a predetermined distance from the vehicle 10 and the height of the target 12 is determined, the depression angle θ1 shown in FIG. 1 can be calculated geometrically. The center target 12 is installed on the center line in the width direction of the vehicle 10. Therefore, if it is possible to determine where the center position of the center target 12 is located in the captured image, the position of the FOE in the image plane can be determined.

Furthermore, the inclination of the in-vehicle camera 14 in the vehicle width direction, that is, the roll angle of the in-vehicle camera 14 is also determined from the inclination of the straight line passing through the center position of each target 12 in the captured image. In image processing for lane mark recognition or the like, an image corrected based on this roll angle is used.
JP 2004-185425 A

  As described above, the conventional FOE determination work is performed by installing the three targets 12 at predetermined positions with respect to the vehicle 10, but the three targets 12 are at predetermined positions with respect to the vehicle 10. It is necessary to work by at least two people to install the camera in the installation, and more than one hour is usually required only by the installation work. Therefore, it has been strongly desired to be able to determine FOE without installing a target.

  The present invention has been made in view of the above-described points, and an object of the present invention is to provide an infinite point determination apparatus capable of determining FOE without installing a target.

In order to achieve the object, the invention according to claim 1 is the infinity point for determining the position of the infinity point in the image picked up by the vehicle-mounted camera fixed to the vehicle so that the bonnet is included in the image pickup range. A decision device,
Vertex determining means for determining the position of the bonnet vertex in the image taken by the vehicle-mounted camera, and the X coordinate of the bonnet vertex determined by the vertex determining means in the image as the X coordinate of the infinity point X coordinate determination means for storing, a storage device storing a Y coordinate difference between a vertex of a bonnet in the image and a Y coordinate of the infinity point, a Y coordinate difference stored in the storage device, and the vertex Y coordinate determination means for determining the Y coordinate of the point at infinity from the Y coordinate of the vertex of the bonnet determined by the determination means.

  The infinity point can be determined in this way because the relative position of the bonnet with respect to the in-vehicle camera is determined as in the conventional target. And if an infinity point is determined from the position of the bonnet vertex in the image imaged by the vehicle-mounted camera in this way, FOE can be determined without installing a target.

  Here, the Y-coordinate of the FOE can be determined using a headlight cut line as in claim 2. The method for determining the X coordinate of the FOE in claim 2 is the same as that in claim 1.

That is, the invention described in claim 2 for achieving the above object is an infinite method for determining the position of an infinite point in an image captured by an in-vehicle camera fixed to the vehicle so that the bonnet is included in the imaging range. A far point determination device,
Vertex determining means for determining the position of the bonnet vertex in the image taken by the vehicle-mounted camera, and the X coordinate of the bonnet vertex determined by the vertex determining means in the image as the X coordinate of the infinity point X coordinate determining means for determining the Y coordinate of the cut line of the headlight in the image captured by the in-vehicle camera in a state where the headlight is irradiated on the irradiation surface located at a predetermined relative position with respect to the vehicle Cutting line determination means for storing, a storage device storing a Y coordinate difference between the headlight cut line in the image and the Y coordinate of the infinity point, and a Y coordinate difference stored in the storage device; A Y-axis for determining the Y coordinate of the infinity point from the Y coordinate of the cut line of the headlight determined by the cut line determination means Characterized in that it comprises a decision means.

  If the distance between the vehicle and the irradiation surface of the headlight is determined, the height of the cut line of the headlight is determined, so that the Y coordinate of the point at infinity can be determined even in the second aspect. And if the infinity point is determined from the position of the top of the bonnet and the position of the headlight cut line in the image taken by the in-vehicle camera as in claim 2, the infinity point is determined without installing the target. Can be determined.

  Here, preferably, as in claim 3, the infinity point determination device is based on a vehicle pitch angle sensor that detects a pitch angle of the vehicle and a vehicle pitch angle detected by the vehicle pitch angle sensor. It is preferable to further include Y coordinate correction means for correcting the Y coordinate determined by the Y coordinate determination means. In this way, if the pitch angle of the vehicle is detected and the Y coordinate correction unit corrects the Y coordinate of the infinity point based on the detected vehicle pitch angle, the vehicle is inclined in the front-rear direction. However, it is possible to obtain an accurate Y coordinate of the point at infinity.

  In Claim 1, the Y coordinate of the infinity point is determined from the bonnet position in the image imaged by the vehicle-mounted camera, and in Claim 2, the position of the headlight cut line in the image imaged by the vehicle-mounted camera However, the final Y coordinate may be determined by comparing the Y coordinate determined from the cut line of the bonnet and the headlight.

  The invention according to claim 4 is an infinity point determination device that determines the position of an infinity point in an image captured by an in-vehicle camera fixed to the vehicle so that the bonnet is included in the imaging range, Vertex determining means for determining the position of the bonnet vertex in the image taken by the vehicle-mounted camera, and the X coordinate of the bonnet vertex determined by the vertex determining means in the image as the X coordinate of the infinity point X coordinate determining means for determining the Y coordinate of the cut line of the headlight in the image captured by the in-vehicle camera in a state where the headlight is irradiated on the irradiation surface located at a predetermined relative position with respect to the vehicle A cut line determining means to perform, a first Y-coordinate difference between a vertex of a bonnet in the image and a Y-coordinate of the infinity point, and the The storage device storing the second Y coordinate difference between the headlight cut line in the image and the Y coordinate of the infinity point, the first Y coordinate difference stored in the storage device, and the vertex determining means A first candidate value calculating means for calculating a first candidate value of the Y coordinate of the infinity point from the Y coordinate of the vertex of the bonnet, a second Y coordinate difference stored in the storage device, and the cut line The second candidate value calculating means for calculating the second candidate value of the Y coordinate of the infinity point from the Y coordinate of the headlight cut line determined by the determining means, and the first candidate value calculating means. A comparison value obtained by comparing the first candidate value of the Y coordinate of the infinity point with the second candidate value of the Y coordinate of the infinity point calculated by the second candidate value calculating means is a predetermined predetermined criterion. Whether it exceeds the value And the Y coordinate comparison means that determines the Y coordinate of the infinity point calculated by the first candidate value calculation means when the Y coordinate comparison means determines that the comparison value does not exceed the determination reference value. Y coordinate determining means for determining the Y coordinate of the infinity point based on at least one of the first candidate value and the second candidate value of the Y coordinate of the infinity point calculated by the second candidate value calculating means. It is characterized by that.

  According to the fourth aspect of the present invention, the first candidate value calculation means calculates the first candidate value of the Y coordinate at the infinity point from the position of the vertex of the bonnet in the image, and the second candidate value calculation means The second candidate value of the Y coordinate at the infinity point is calculated from the position of the cut line of the headlight in the image, and the Y coordinate comparison means obtains a comparison value comparing the first candidate value and the second candidate value. It is determined whether or not a predetermined determination reference value is exceeded. If the comparison value exceeds the criterion value, it can be said that at least one of the first candidate value and the second candidate value contains a large error. In this case, the measurement is automatically remeasured. Or a message indicating that remeasurement should be performed after adjusting the mounting position of the in-vehicle camera. Accordingly, since a large error is less included in the Y coordinate of the infinity point, the infinity point can be determined with higher accuracy.

  If the comparison value does not exceed the criterion value, that is, if the first candidate value and the second candidate value are close, it is considered that both represent relatively accurate values. The coordinate determination means determines the Y coordinate of the infinity point using at least one of the first candidate value and the second candidate value.

  Claim 5 with respect to Claim 4 has the same relationship as Claim 3 with respect to Claim 1 or 2, and based on the pitch angle of the vehicle, the first candidate value and the first candidate value that are candidate values of the Y coordinate of the infinity point 2 to correct candidate values. That is, the invention according to claim 5 is the infinity point determination device according to claim 4, wherein the vehicle pitch angle sensor detects the pitch angle of the vehicle, and the vehicle pitch angle detected by the vehicle pitch angle sensor. And a candidate value correcting unit that corrects the first candidate value and the second candidate value of the Y coordinate based on the first candidate value and the first candidate value corrected by the candidate value correcting unit. The Y coordinate of the point at infinity is determined based on at least one of the two candidate values. In this way, even if the vehicle is inclined in the front-rear direction, it is possible to obtain a more accurate Y coordinate of the infinity point.

  Furthermore, it is preferable that the infinity point determination apparatus according to the first to fifth aspects determines the roll angle of the in-vehicle camera as in the sixth aspect. That is, the invention according to claim 6 is the infinity point determination device according to any one of claims 1 to 5, wherein the irradiation surface located at a predetermined relative position with respect to the vehicle is irradiated with a headlight. The apparatus further comprises roll angle determining means for determining the roll angle of the in-vehicle camera based on the inclination of the cut line of the headlight in the image captured by the in-vehicle camera.

  In this way, the roll angle of the in-vehicle camera can also be determined without installing a target.

  The roll angle of the in-vehicle camera determined in claim 6 is preferably corrected based on the vehicle roll angle as described in claim 7. That is, the invention according to claim 7 is the infinity point determination device according to claim 6, wherein the vehicle roll angle sensor that detects the roll angle of the vehicle and the vehicle roll angle detected by the vehicle roll angle sensor are used. On the basis of this, it further comprises roll angle correction means for correcting the roll angle of the in-vehicle camera determined by the roll angle determination means.

  In this way, even if the vehicle is inclined in the width direction, the roll angle of the in-vehicle camera can be accurately determined.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 3 is a block diagram showing the configuration of the infinity point determination apparatus 100 of the present invention. The infinity point determination device 100 functions as an arithmetic processing device 120 that processes an image captured by the in-vehicle camera 110, a vehicle posture angle sensor 130 that detects a posture angle (roll angle, pitch angle) of the vehicle, and a storage device. E2PROM 140 and a display device 150 are provided.

  The in-vehicle camera 110 is constituted by a CCD camera, and as shown in FIG. 4, the mounting position is the interior of the vehicle 160, for example, the ceiling near the driver's seat. The mounting position of the in-vehicle camera 110 in the vehicle width direction is the center in the vehicle width direction. Moreover, as shown in FIG. 4, the imaging range θ2 in the vertical direction of the in-vehicle camera 110 is a range in which the hood of the vehicle 160 can be imaged and the image above the horizontal direction can be imaged.

  A straight line L1 shown in FIG. 4 is a straight line extending from the in-vehicle camera 110 so as to be in contact with the hood of the vehicle 160, and a contact C is a contact between the straight line L1 and the hood. FIG. 5 is a diagram conceptually showing an image captured by the in-vehicle camera 110. The boundary line (hereinafter referred to as bonnet line) BL of the bonnet shown in FIG. 5 includes a contact C shown in FIG. 4 and is a cut surface parallel to the vehicle width direction, that is, the cut shown in FIG. It coincides with the upper edge of the bonnet on the surface F. The bonnet line BL has a vertex P at the center in the vehicle width direction.

Here, the mounting position of the in-vehicle camera 110 is determined, and the contact C is the same position if the mounting position of the in-vehicle camera 110 is the same. Therefore, the horizontal distance and the vertical distance between the in-vehicle camera 110 and the contact C are always constant. If the horizontal distance between the in-vehicle camera 110 and the contact C is constant, the imaging length h1 in the height direction of the in-vehicle camera 110 at the cut surface F is determined, and as described above, the in-vehicle camera 110 and the contact The vertical distance from C is constant. If the vertical distance between the in-vehicle camera 110 and the contact C is h2, and the difference in Y coordinate from the lower end to the upper end of the image is Y0, the Y coordinate between the vertex P of the bonnet and the FOE in the image The difference (hereinafter referred to as the first Y coordinate difference) Yd1 can be calculated from Equation 1. Therefore, the first Y coordinate difference Yd1 is a constant value and can be calculated in advance. The E2PROM 140 stores a first Y coordinate difference Yd1 calculated in advance.
(Formula 1) Yd1 = Y0 × (h2 / h1)
The E2PROM 140 also stores a second Y coordinate difference Yd2 that is a Y coordinate difference between the headlight cut line CL and the FOE in the image captured by the in-vehicle camera 110. The second Y coordinate difference Yd2 will be described later.

  The vehicle attitude angle sensor 130 functions as a vehicle pitch angle sensor and a vehicle roll angle sensor. For example, the vehicle attitude angle sensor 130 is provided in the vicinity of a four-wheel suspension and detects the distance between the road surface and the vehicle bottom surface in each suspension portion. The vehicle roll angle and pitch angle are detected by comparing the distances between the road surface and the vehicle bottom surface detected by the four distance sensors.

  The arithmetic processing unit 120 includes a microcomputer including a CPU, ROM, RAM, and the like (not shown), and is stored in the E2PROM 140 by executing a program stored in advance in the ROM while using a primary storage function of the RAM. Further, based on the first Y coordinate difference Yd1, the second Y coordinate difference Yd2, the image captured by the in-vehicle camera 110, the vehicle pitch angle detected by the vehicle attitude sensor 130, and the vehicle roll angle, the FOE is determined and the in-vehicle camera 110 Determine the roll angle.

  FIG. 6 is a flowchart illustrating a main part of processing for determining the FOE and the roll angle of the in-vehicle camera 110 in the arithmetic processing unit 120. In FIG. 6, first, in step S100, the first candidate value (FOE (y1) of the X coordinate (FOE (X)) of the FOE and the Y coordinate of the FOE is determined from the position of the top of the bonnet in the image captured by the in-vehicle camera 110. )) Is determined. During the execution of step S100, the in-vehicle camera 110 continuously captures images.

  The process in step S100 of FIG. 6 is shown in detail in FIG. In step S105 of FIG. 7, exposure control is performed on images continuously captured by the in-vehicle camera 110. In this exposure control, even if the imaging range slightly changes due to an attachment error of the in-vehicle camera 110, the contrast is compared with a reference range 170 (for example, the range shown in FIG. 5) that is set in advance so as to be a hood portion. The exposure is adjusted so as to maximize the value.

  In subsequent step S110, an image captured with the exposure adjusted in step S105 is captured as an image processing image. In subsequent step S115, a preset range is cut out from the image captured in step S110. The range to be cut out here is a range below the one-dot chain line shown in FIG. 5, and this range is set so that the bonnet occupies most of the range.

  In subsequent step S120, a pixel value histogram is created for the range cut out in step S115. FIG. 8 shows an example of the pixel value histogram. In FIG. 8, the minimum value of the pixel value is 0 and the maximum value is 255.

  In the subsequent step S125, the pixel value range of the bonnet is determined based on the pixel value histogram. As shown in FIG. 8, this pixel value histogram has one large peak. However, since this pixel value histogram is in a range set so that the bonnet occupies the majority, FIG. The peak shown indicates the pixel value of the bonnet. Therefore, in step S125, the pixel value range of the bonnet is determined from the rise to the fall of the peak.

  In the subsequent step S130, the binarization processing is executed for the range cut out in step S115, in which the range determined as the pixel value range of the bonnet in step S125 is white and the other pixel values are black. In step S135, the upper boundary line of the white range in the binarized image is determined as the bonnet line BL. Further, in the subsequent step S140, the vertex of the bonnet line BL determined in step S135, that is, the vertex of the bonnet in the image is determined. Note that steps S105 to S140 in FIG. 7 correspond to vertex determination means.

  The subsequent step S145 corresponds to X coordinate determination means. Since the vertex of the bonnet is located at the center in the vehicle width direction in the image picked up by the in-vehicle camera 110, in this step S145, the X coordinate of the bonnet vertex determined in step S140 is determined as the X coordinate of the FOE. .

  The subsequent step S150 corresponds to a first candidate value calculation means, and the first candidate of the FOE Y coordinate is obtained by adding the first Y coordinate difference Yd1 stored in the E2PROM 140 to the Y coordinate of the bonnet vertex determined in step S140. The value FOE (y1) is calculated.

  Returning to FIG. 6, after step S100 is executed, step S200 is executed. In step S200, as shown in FIG. 9, the vehicle 160 is placed at a predetermined distance from the wall surface 180 so that the vehicle width direction is parallel to the planar wall surface 180 that is the irradiation surface. This is executed with the room darkened and the wall surface 180 irradiated with a headlight. Moreover, during execution of step S200, the vehicle-mounted camera 110 is continuously capturing images. FIG. 10 is a diagram conceptually illustrating an image captured by the in-vehicle camera 110 during the execution of step S200. As illustrated in FIG. 10, the image captured by the in-vehicle camera 110 includes a headlight cut line. CL is reflected.

  In step S200, the first candidate value (FOE (y1)) of the Y coordinate of the FOE and the roll angle of the in-vehicle camera 110 are determined from the position of the headlight cut line CL in the image captured by the in-vehicle camera 110.

  The process in step S200 is shown in detail in FIG. In FIG. 11, first, in step S <b> 205, exposure control is performed on images continuously captured by the in-vehicle camera 110. In this exposure control, even if the imaging range slightly changes due to an attachment error of the in-vehicle camera 110, the contrast is maximized with respect to a reference range that is set in advance so as to be an irradiation portion of the headlight in the image. Thus, the exposure is adjusted. The reference range is set to a narrow range centered on a point moved downward by a predetermined value from the average position of the headlight cut line determined based on experiments, for example.

  In subsequent step S210, an image captured with the exposure adjusted in step S205 is captured as an image processing image. In subsequent step S215, a preset range is cut out from the image captured in step S210. The range cut out here is the range shown in FIG. This cutout range is set so that the irradiated portion of the headlight shows a large proportion, the X-axis direction of the range matches the imaging range, and the Y-axis direction is the position of the headlight cut line CL in the image. For example, the range of the image range from the bottom 1/3 to the bottom 1/10 is ensured so that the cut line CL of the headlight is surely included even if fluctuates slightly.

  In step S220, similarly to step S120 in FIG. 7, a pixel value histogram is created for the range cut out in step S215. Although the pixel value histogram created in step S220 is not shown in the drawing, the clipping range in step S215 is set so that the irradiated portion of the headlight has a large proportion, so the pixel value histogram created in step S220. Has a large peak in the pixel value corresponding to the irradiated portion of the headlight. Therefore, in step S225, the pixel value region of the irradiated portion of the headlight is determined from the rise to the fall of the maximum peak in the pixel value histogram created in step S220.

  Then, in the subsequent step S230, binarization processing is performed in which the range determined as the pixel value range of the irradiated portion of the headlight is set to white and the other pixel values are set to black with respect to the range cut out in step S215. Execute. In the subsequent step S235, the upper boundary line of the white range in the binarized image is determined as the headlight cut line CL. Further, in the subsequent step S240, the headlight cut line CL determined in step S235 is determined as the minimum automatic line. Linear approximation is performed using multiplication or the like. Note that steps S205 to S240 in FIG. 11 correspond to cut line determination means.

  The subsequent step S245 corresponds to second candidate value calculation means, and the second Y coordinate difference Yd2 (vehicle-mounted camera 110) stored in the E2PEOM 140 is added to the Y coordinate at the center in the width direction of the cut line CL of the headlight linearly approximated in step S240. The second candidate value FOE (y2) of the Y coordinate of the FOE is calculated by adding the Y coordinate difference between the headlight cut line CL and the FOE in the image picked up by.

Here, the second Y coordinate difference Yd2 will be described. Since the distance between the wall surface 180 irradiated with the headlight and the vehicle 160 is determined, the distance between the in-vehicle camera 110 and the wall surface 180 where the attachment site to the vehicle 160 is determined is determined. If the distance from the vehicle 160 to the irradiation surface, that is, the wall surface 180 is determined, the height of the headlight cut line CL on the irradiation surface (wall surface 180) is also determined in advance. Accordingly, the vertical distance h4 between the in-vehicle camera 110 and the headlight cut line CL shown in FIG. 9 can be obtained in advance. In addition, since the imaging length h3 in the height direction of the in-vehicle camera 110 on the wall surface 180 can also be obtained, the second Y is obtained using h3 and h4 and the Y coordinate difference Y0 from the lower end to the upper end of the image. The coordinate difference Yd2 can be calculated from Equation 2.
(Formula 2) Yd2 = Y0 × (h4 / h3)
The subsequent step S250 corresponds to a roll angle determining means, and the inclination of the headlight cut line CL approximated by a straight line in step S240 is determined as the roll angle of the in-vehicle camera 110.

  Returning to FIG. 6, after step S200 is executed, in step S300 corresponding to the candidate value correction means, a signal representing the vehicle pitch angle is read from the vehicle attitude angle sensor 130, and the vehicle pitch angle is a value whose sign is opposite to that of the vehicle pitch angle. Are added to the first candidate value FOE (y1) calculated in step S150 of FIG. 7 and the second candidate value FOE (y2) calculated in step S245 of FIG. 11, respectively, to thereby obtain the first candidate value FOE (y1). The second candidate value FOE (y2) is corrected to a value in a state where the pitch angle of the vehicle is zero. Hereinafter, the corrected first candidate value FOE (y1) is referred to as a corrected first candidate value (y1 '), and the corrected second candidate value FOE (y2) is referred to as a corrected second candidate value FOE (y2).

  The subsequent step S310 corresponds to roll angle correction means. In this step S310, a signal representing the vehicle roll angle is read from the vehicle attitude angle sensor 130, and the roll angle of the in-vehicle camera 110 determined in step S250 of FIG. Correct the value to zero.

  Subsequently, steps S320 to S330 corresponding to the Y coordinate comparison unit are executed. In step S320, a comparison value between the corrected first candidate value (y1 ') and the corrected second candidate value (y2') obtained in step S300 is calculated. Here, the difference between the corrected first candidate value (y1 ′) and the corrected second candidate value (y2 ′) is calculated as the comparison value. Any difference between the first candidate value (y1 ′) and the corrected second candidate value (y2 ′) may be used. For example, the ratio between the two can be calculated as a comparison value.

  In subsequent step S330, it is determined whether or not the comparison value calculated in step S320 is greater than a preset determination reference value TH. The determination in step S330 is affirmative when the corrected first candidate value FOE (y1 ′) and the corrected second candidate value FOE (y2 ′) are relatively far apart, and at least one of them is It can be considered that a large error is included. However, since it cannot be determined which contains a large error or both include a large error, in step S340, a message for instructing remeasurement is displayed on the display device 150 to display the main error. End the routine.

  On the other hand, the determination in step S330 is affirmative when the corrected first candidate value FOE (y1 ′) and the corrected second candidate value FOE (y2 ′) are relatively close values. Since both are considered to have a small error, either one may be used as the FOE Y coordinate FOE (Y). Here, in step S350, the corrected first candidate value FOE (y1 ′) and the corrected second A value obtained by simply averaging the candidate values FOE (y2 ′) is determined as the Y coordinate FOE (Y) of the FOE. This step S350 corresponds to the Y coordinate determination means.

  As described above, according to the present embodiment described above, the FOE is determined from the position of the top of the hood and the position of the cut line CL of the headlight in the image captured by the in-vehicle camera 110. It is no longer necessary to install a target.

  Further, according to the present embodiment, the first candidate value FOE (y1) and the second candidate value FOE (y2) of the Y coordinate of the FOE are corrected based on the pitch angle of the vehicle, so that the vehicle 160 is moved in the front-rear direction. Even if it is inclined, the Y coordinate of the FOE can be determined with high accuracy.

  In addition, according to the present embodiment, the roll angle of the in-vehicle camera 110 is determined from the position of the headlight cut line CL in the image captured by the in-vehicle camera 110, and therefore, the roll angle of the in-vehicle camera 110 also sets a target. Can be decided without. Moreover, since the roll angle is corrected based on the vehicle roll angle, the roll angle of the in-vehicle camera 110 can be accurately determined even if the vehicle 160 is inclined in the vehicle width direction.

  As mentioned above, although embodiment of this invention was described, this invention is not limited to the above-mentioned embodiment, The following embodiment is also contained in the technical scope of this invention, and also the summary other than the following is also included. Various modifications can be made without departing from the scope.

  For example, in the above-described embodiment, the first candidate value FOE (y1) and the second candidate value FOE (y2) of the Y coordinate of the FOE are corrected based on the vehicle pitch angle. When the FOE determination operation is performed in a state where the pitch angle is small, the correction of the first candidate value FOE (y1) and the second candidate value (y2) based on the vehicle pitch angle may not be executed. Further, the correction of the roll angle of the in-vehicle camera 110 based on the vehicle roll angle may not be performed when the roll angle of the in-vehicle camera 110 is determined in a state where the vehicle roll angle is small.

  In the above-described embodiment, the first candidate value FOE (y1) of the Y coordinate of the FOE is calculated from the position of the vertex of the bonnet in the image captured by the in-vehicle camera 110, and the headlight is cut in the image. The second candidate value FOE (y2) of the FOE Y coordinate is calculated from the position of the line CL, and after comparing them, the Y coordinate of the FOE is determined. The first candidate value FOE (y1) or the correction is made. The corrected first candidate value FOE (y1 ′), the second candidate value FOE (y2), or the corrected second candidate value FOE (y2 ′) corrected therefor may be used as the Y coordinate of the FOE. In this case, step S150 functioning as the first candidate value calculating means or step S245 functioning as the second candidate value calculating means functions as the Y coordinate determining means, and based on the vehicle pitch angle. In this case, step S300 functions as a Y coordinate correction unit.

  In step S350 of the above-described embodiment, the Y coordinate FOE (Y) of the FOE is determined by simply averaging the corrected first candidate value FOE (y1 ′) and the corrected second candidate value FOE (y2 ′). However, the Y coordinate FOE (Y) of the FOE may be determined using a weighted average instead of the simple average.

It is a figure for demonstrating the conventional method for determining the initial position of FOE. It is a figure which shows an example of the image imaged with the vehicle-mounted camera 14 in the state of FIG. It is a block diagram which shows the structure of the infinity point determination apparatus 100 with which this invention was applied. It is a figure which shows the state of the operation | work which determines FOE from the position of the vertex of a bonnet in an image from the side of the vehicle 160. FIG. It is a figure which shows notionally the image imaged with the vehicle-mounted camera 110 in the state of FIG. 7 is a flowchart showing a main part of processing for determining the FOE and the roll angle of the vehicle-mounted camera 110 in the arithmetic processing unit 120. It is a flowchart which shows step S100 of FIG. 6 in detail. It is a figure which shows an example of the pixel value histogram produced by step S120 of FIG. FIG. 5 is a diagram showing a state of work for determining the FOE and the roll angle of the in-vehicle camera 110 from the position of the headlight cut line CL in the image from the side of the vehicle 160. It is a figure which shows notionally the image imaged with the vehicle-mounted camera 110 in the state of FIG. It is a flowchart which shows step S200 of FIG. 6 in detail.

Explanation of symbols

100: Infinite point determination device 110: On-vehicle camera 130: Vehicle attitude angle sensor (vehicle roll angle sensor, vehicle pitch angle sensor)
140: E2PROM (storage device)
180: Wall surface (irradiated surface)
S105 to S140: vertex determination means S145: X coordinate determination means S150: first candidate value calculation means S205 to S240: cut line determination means S245: second candidate value calculation means S250: roll angle determination means S300: candidate value correction means S310 : Roll angle correcting means S320 to S330: Y coordinate comparing means S350: Y coordinate determining means

Claims (7)

An infinity point determination device that determines the position of an infinity point in an image captured by an in-vehicle camera fixed to a vehicle so that a bonnet is included in an imaging range,
Vertex determining means for determining the position of the vertex of the bonnet in the image captured by the in-vehicle camera;
X-coordinate determining means that uses the X-coordinate in the image of the bonnet vertex determined by the vertex determining means as the X-coordinate of the infinity point;
A storage device storing a Y coordinate difference between a vertex of a bonnet in the image and a Y coordinate of the infinity point;
Y coordinate determination means for determining the Y coordinate of the infinity point from the Y coordinate difference stored in the storage device and the Y coordinate of the bonnet vertex determined by the vertex determination means, Infinite point determination device. An infinity point determination device that determines the position of an infinity point in an image captured by an in-vehicle camera fixed to a vehicle so that a bonnet is included in an imaging range,
Vertex determining means for determining the position of the vertex of the bonnet in the image captured by the in-vehicle camera;
X-coordinate determining means that uses the X-coordinate in the image of the bonnet vertex determined by the vertex determining means as the X-coordinate of the infinity point;
A cut line determination means for determining a Y coordinate of a cut line of the headlight in an image captured by the in-vehicle camera in a state in which the irradiation surface located at a predetermined relative position with respect to the vehicle is irradiated with the headlight;
A storage device storing a Y coordinate difference between a headlight cut line in the image and a Y coordinate of the infinity point;
Y coordinate determining means for determining the Y coordinate of the infinity point from the Y coordinate difference stored in the storage device and the Y coordinate of the headlight cut line determined by the cut line determining means. Infinite point determination device characterized by. A vehicle pitch angle sensor for detecting a pitch angle of the vehicle;
The Y coordinate correcting means for correcting the Y coordinate determined by the Y coordinate determining means based on the vehicle pitch angle detected by the vehicle pitch angle sensor. Infinite point determination device. An infinity point determination device that determines the position of an infinity point in an image captured by an in-vehicle camera fixed to a vehicle so that a bonnet is included in an imaging range,
Vertex determining means for determining the position of the vertex of the bonnet in the image captured by the in-vehicle camera;
X-coordinate determining means that uses the X-coordinate in the image of the bonnet vertex determined by the vertex determining means as the X-coordinate of the infinity point;
A cut line determination means for determining a Y coordinate of a cut line of the headlight in an image captured by the in-vehicle camera in a state in which the irradiation surface located at a predetermined relative position with respect to the vehicle is irradiated with the headlight;
A first Y coordinate difference between the vertex of the bonnet in the image and the Y coordinate of the infinity point, and a second Y coordinate difference between the headlight cut line and the Y coordinate of the infinity point in the image. A stored storage device;
First candidate value calculation for calculating a first candidate value of the Y coordinate of the infinity point from the first Y coordinate difference stored in the storage device and the Y coordinate of the vertex of the bonnet determined by the vertex determining means Means,
A second candidate value for calculating the second candidate value of the Y coordinate at the infinity point is calculated from the second Y coordinate difference stored in the storage device and the Y coordinate of the cut line of the headlight determined by the cut line determining means. Candidate value calculation means;
A comparison value obtained by comparing the first candidate value of the Y coordinate of the infinity point calculated by the first candidate value calculation means with the second candidate value of the Y coordinate of the infinity point calculated by the second candidate value calculation means Y coordinate comparison means for judging whether or not exceeds a predetermined judgment reference value set in advance,
When the Y coordinate comparison means determines that the comparison value does not exceed the determination reference value, the first candidate value of the Y coordinate at the infinity point calculated by the first candidate value calculation means, and the first An infinite point including Y coordinate determining means for determining the Y coordinate of the infinite point based on at least one of the second candidate values of the Y coordinate of the infinite point calculated by the two candidate value calculating means Decision device. A vehicle pitch angle sensor for detecting a pitch angle of the vehicle;
Candidate value correcting means for correcting the first candidate value and the second candidate value of the Y coordinate based on the pitch angle of the vehicle detected by the vehicle pitch angle sensor;
The Y-coordinate determining unit determines a Y-coordinate of an infinite point based on at least one of the first candidate value and the second candidate value after correction by the candidate value correcting unit. 4. The infinity point determination device according to 4.   A roll angle of the in-vehicle camera based on an inclination of a cut line of the headlight in an image captured by the in-vehicle camera in a state in which the irradiation surface located at a predetermined relative position with respect to the vehicle is irradiated with a headlight. The infinity point determination device according to claim 1, further comprising roll angle determination means for determining. A vehicle roll angle sensor for detecting a roll angle of the vehicle;
The apparatus further comprises roll angle correction means for correcting the roll angle of the in-vehicle camera determined by the roll angle determination means based on the roll angle of the vehicle detected by the vehicle roll angle sensor. Item 7. The infinity point determination device according to Item 6.

Images