JP2011180020A - Distance recognition device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide such a technology that can improve the recognition accuracy of distance between own vehicle and an object when a photographed image includes the object at a position far from the own vehicle. <P>SOLUTION: A distance between own vehicle and an object is recognized by a recognizer 6e on the basis of a distance Z between own vehicle and an object that is calculated by a distance calculator according to each of edges hEG, lEG and rEG. When an object at a position far from the own vehicle is included in a photographed image F, the distance Z between the own vehicle and the object is recognized based on each of the distances Z calculated by different methods, even when each of the distances Z calculated by a distance calculator 6d according to each of the edges hEG, lEG and rEG may vary from the actual distance between the own vehicle and the object. Accordingly, the recognized distance is prevented from varying greatly from the actual distance Z and the recognition accuracy of the distance Z between the own vehicle and the object can be improved. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a technique for recognizing a distance from an own vehicle to an object based on a photographed image photographed by an in-vehicle camera.

  2. Description of the Related Art Conventionally, a technique for recognizing a distance from an own vehicle to an object based on a horizontal edge of an object included in a captured image captured by an in-vehicle camera is known (for example, see Patent Document 1). That is, the horizontal edge of the object included in the captured image is detected by processing the captured image, and the distance from the vehicle to the object is calculated based on the vertical distance from the lower end of the captured image to the detected horizontal edge. The

JP-A-3-273500 (page 3, upper right column to page 3, lower left column, FIG. 7, etc.)

  By the way, in the upper part of the captured image of the in-vehicle camera (the back side: a position far from the own vehicle), the distance in the depth direction is compressed and compared with the lower part of the captured image (the near side: a position close to the own vehicle). . That is, the range of the distance from the vehicle to the object assigned to each pixel forming the captured image is larger on the back side of the captured image than on the near side of the captured image. Therefore, when a horizontal edge of an object is detected on the back side of the captured image, the object from the recognized vehicle can be detected only by shifting the position of the detected horizontal edge by one pixel in the vertical direction (depth direction) in the captured image. The distance up to changes greatly. Therefore, if a horizontal edge is detected on the far side of the captured image, the distance between the vehicle and the object recognized based on the detected horizontal edge and the actual distance from the vehicle to the object There was a risk that a large deviation would occur.

  The present invention has been made in view of the above problems, and when an object far from the host vehicle is included in the captured image, the recognition accuracy of the distance from the host vehicle to the object can be improved. The purpose is to provide technology.

  In order to achieve the above-described object, a distance recognition device according to the present invention includes a distance recognition device that recognizes a distance from an own vehicle to an object based on a photographed image photographed by an in-vehicle camera. First distance calculating means for calculating a distance from the own vehicle to the object based on a position of a horizontal edge indicating a lower end of the object; and left and right ends of the object included in the upper-view image obtained by projectively converting the captured image. A central axis that bisects the angle formed by the vertical edge is obtained, and at least one of the two vertical edges, the length from the central axis to the lower end of the vertical edge, and the angle of the vertical edge with respect to the central axis Second distance calculating means for calculating a distance from the own vehicle to the object based on the first distance by the first distance calculating means and the second distance calculating means. It is characterized in that the said vehicle on the basis of the second distance and a recognition means for recognizing the distance to the object (claim 1).

  In the distance recognition device according to claim 1, the recognition unit may recognize a distance to the object by averaging the first distance and the second distance. 2).

  The distance recognition apparatus according to claim 1 or 2, wherein at least one of the first distance calculation unit and the second distance calculation unit is the self-image among a plurality of frames of the captured image by the imaging unit. The number of frames in which the reference position of the object when calculating the distance from the vehicle to the object is located in the same pixel of each captured image, and the assigned distance from the own vehicle assigned to the same pixel to the object Based on the above, it is preferable to derive the movement amount of the object by equally dividing the allocated distance by the number of frames, and to calculate the distance from the host vehicle to the object based on the derived movement amount. (Claim 3).

  According to the first aspect of the present invention, when the distance from the own vehicle to the object is recognized based on the captured image captured by the in-vehicle camera, the horizontal edge position indicating the lower end of the object included in the captured image is set. Based on this, the distance from the vehicle to the object is calculated by the first distance calculating means.

  Then, the second distance calculating means obtains a central axis that bisects the angle formed by the vertical edges indicating the left and right ends of the object included in the upper-view image obtained by projective transformation of the captured image, and at least one of the two vertical edges. On the other hand, the distance from the vehicle to the object is calculated based on the length from the central axis to the lower end of the vertical edge and the angle of the vertical edge with respect to the central axis, and the first distance is calculated by the first distance calculating means. And the distance from the own vehicle to the object is recognized by the recognition means based on the second distance by the second distance calculation means.

  Therefore, when an object at a position far from the own vehicle is included in the captured image, the first distance calculated by the first distance calculating unit may be shifted from the actual distance between the own vehicle and the object. Even in such a case, since the distance from the vehicle to the object is recognized based on the first distance and the second distance calculated by different methods, the recognized distance may greatly deviate from the actual distance. Thus, the recognition accuracy of the distance from the vehicle to the object can be improved.

  According to the invention of claim 2, since the distance represented by one pixel is different in each region based on the infinity point in the captured image, the first distance based on the horizontal edge and the second distance based on the vertical edge However, the recognition means recognizes the distance from the own vehicle to the object by averaging the first distance and the second distance, thereby recognizing the object from the recognized own vehicle. Can be further improved.

  According to the invention of claim 3, the object when calculating the distance from the own vehicle to the object in a plurality of frames of the captured image, even though the distance between the own vehicle and the object is relatively moving. Even when the reference position is not moved, at least one of the first distance calculation unit and the second distance calculation unit is configured such that the reference position of each of the captured images is a plurality of frames of the captured images. Based on the number of frames located at the same pixel and the assigned distance from the vehicle to the object assigned to the same pixel, the assigned distance is divided equally by the number of frames to derive the amount of movement of the object for each frame. Since the distance from the own vehicle to the object is calculated based on the derived movement amount, the actual distance from the own vehicle to the object can be recognized with higher accuracy.

It is a block diagram of one embodiment of an object recognition device of the present invention. It is a figure which shows an example of a picked-up image. It is a figure which shows the distance from the calculated own vehicle to an object. It is a figure which shows the moving amount | distance of the object for every flame | frame. It is a flowchart for operation | movement description of FIG.

  An embodiment of the present invention will be described with reference to FIGS.

  FIG. 1 is a block diagram of an embodiment of an object recognition apparatus 2 of the present invention. 2A and 2B are diagrams illustrating an example of a captured image F of the monocular camera 3. FIG. 2A is a captured image F taken from the camera viewpoint, and FIG. 2B is a horizontal view included in the upper-view image tF obtained by projective transformation of the captured image F. An example in which the edge hEG is emphasized is shown, and (c) shows an example in which the vertical edges lEG and rEG included in the upper view image tF obtained by projective transformation of the captured image F are emphasized. FIG. 3 is a diagram showing the calculated distance from the own vehicle 1 to the object at each time. FIG. 3A shows the distance Z from the own vehicle 1 to the object based on the horizontal edge hEG indicating the lower end of the object. ) Indicates the distance Z from the own vehicle 1 to the object based on the vertical edge lEG indicating the left end of the object, and (c) indicates the distance Z from the own vehicle 1 to the object based on the vertical edge rEG indicating the right end of the object. FIG. 4 is a diagram showing the amount of movement of the object for each frame. FIG. 5 is a flowchart for explaining the operation of FIG.

1. Configuration An object recognition device 2 shown in FIG. 1 is an object (another vehicle) from the own vehicle 1 on the basis of a photographed image F photographed by a monocular camera 3 (corresponding to an “on-vehicle camera” of the present invention) mounted on the own vehicle 1. 12), and a microcomputer for processing the captured image F of the monocular camera 3, the storage means 4 for storing the captured image F of the monocular camera 3 and various data, and the like. An ECU 6 having a configuration, a travel monitor 7 such as a CRT attached to an instrument panel or the like, a liquid crystal display, and an alarm speaker 8 are provided.

  In this embodiment, the monocular camera 3 includes a monochrome or color camera capable of shooting in front of the host vehicle 1 and outputs a signal of a captured image F that is continuously and momentarily captured (FIG. 2 ( a)). Further, in order to perform road surface photography in front of the host vehicle 1, the monocular camera 3 is provided at an upper position in front of the host vehicle 1 in the vehicle so as to aim at an obliquely lower road surface from above.

  The storage unit 4 includes various memories, a hard disk drive, and the like. For the captured image F, various data used when the ECU 6 performs various processes on the captured image F output from the monocular camera 3 every moment. The ECU 6 stores various data obtained by performing various processes by the ECU 6.

  The ECU 6 includes the following units by executing a preset object recognition program, for example, when the ignition key is turned on.

(1) Acquisition means 6a
The acquisition unit 6a acquires a captured image F captured by the monocular camera 3 every moment. For example, as shown in FIG. 2 (a), the captured image F captured by the monocular camera 3 captures the other vehicle 12 traveling as a preceding vehicle in front of the vehicle 1 in the traveling lane, and obtaining means. 6a acquires this captured image F. Note that the captured image F of this embodiment is formed by, for example, 480 line images, and each line image is formed by, for example, 640 pixels.

(2) Detection means 6b
The detection means 6b is a vertical edge detection unit that detects a horizontal edge hEG that indicates the lower edge of an object on the road from a captured image F shown in FIG. 2A by a known edge detection process using a horizontal edge histogram or the like. A function as a left edge detection unit and a right edge detection unit for detecting vertical edges lEG and rEG indicating the left and right edges of an object on the road by a known edge detection process using an edge histogram or the like is included in the photographed image F. A horizontal edge hEG at the lower end of the object (other vehicle 12), a vertical edge lEG at the left end, and a vertical edge rEG at the right end are detected.

(3) Conversion means 6c
The conversion means 6c sets a virtual plane on the road surface and captures the captured image F taken at every time by the monocular camera 3 into a top view image tF shown in FIGS. 2 (b) and 2 (c). . Various data relating to the captured image F of each frame that is captured in time series by the monocular camera 3 and is projectively converted by the conversion unit 6 c is stored in the storage unit 4. In this embodiment, as shown in FIG. 2A, a horizontal edge hEG indicating the lower end of the object included in the captured image F, a vertical edge lEG indicating the left end, and a vertical edge rEG indicating the right end are detected. The photographed image F is projectively converted by the converting means 6c. In addition, in the upper-view image tF obtained by projecting the captured image F of the monocular camera 3 onto the virtual plane set on the road surface, the upper end side of the object (the other vehicle 12) is distorted.

(4) Distance calculation means 6d
The distance calculation means 6d is based on the upward view image tF in which the horizontal edge hEG shown in FIG. 12B is emphasized and the upward view image tF in which the vertical edges lEG and rEG shown in FIG. The distance Z from the vehicle 1 to the object is calculated, and in this embodiment, the following functions are provided.

a) First distance calculation means The first distance calculation means is based on the position of the horizontal edge hEG indicating the lower end of the object included in the upper view image tF shown in FIG. First distance) is calculated (see FIG. 3A).

b) Second distance calculation means The second distance calculation means derives angles θl and θr of the vertical edges lEG and rEG indicating the left and right ends of the object included in the upper view image tF with respect to the lower end of the upper view image tF. A central axis CA that bisects the angle formed by both vertical edges lEG and rEG is obtained, the lower end of the vertical edge lEG indicating the left end of the object is estimated as the lower left end of the object, and the length from the central axis CA to the lower end of the vertical edge lEG is estimated. Based on the length L and the angle of the vertical edge lEG with respect to the central axis CA, a distance Z (second distance) from the vehicle 1 to the object is calculated by performing a geometric calculation (FIG. 3B). )reference).

  Further, the second distance calculation means estimates that the lower end of the vertical edge rEG indicating the right end of the object is the lower right end of the object, the length from the central axis CA to the lower end of the vertical edge rEG, and the angle of the vertical edge rEG with respect to the central axis CA. Based on the above, a geometric calculation is performed to calculate a distance Z (third distance) from the vehicle 1 to the object (see FIG. 3C).

  As described above, the distance calculation means 6d is configured to measure the distance from the vehicle 1 to the object based on the horizontal edge hEG and the two vertical edges lEG and rEG as shown in FIGS. 3 (a), (b), and (c). Z (first distance to third distance) is calculated. Note that the intersection point O of the extended lines of the two vertical edges lEG and rEG corresponds to the position of the monocular camera 3. The central axis is a vertical axis passing through the position of the monocular camera 3 in the upper-view image tF.

  3A, 3B, and 3C are included in the captured image F of each frame captured by the monocular camera 3 at times t0, t1, t2, t3, t4, t5, and t6, respectively. A distance Z (first distance to third distance) from the vehicle 1 to the object calculated by the distance calculation means 6d based on each edge hEG, lEG, rEG is shown. The distance Z calculated by the distance calculation means 6d based on the edges hEG, lEG, and rEG included in the captured image F of the same frame may differ from the back side (upper part of the captured image F). ) Is a compressed image in which the distance in the depth direction is compressed, and therefore the range of the distance from the vehicle 1 to the object assigned to each pixel is compared to the near side (lower part) of the captured image F. Because it is big.

  That is, as shown in FIG. 2A, the number of pixels assigned to the same size is smaller on the back side of the captured image F including the scenery far from the host vehicle 1 than on the near side of the captured image F. , The compressed image is a compressed landscape. Therefore, since the number of line images assigned to the same distance range in the depth direction is smaller on the far side of the photographed image F than on the near side of the photographed image F, each edge hEG, lEG detected by the detection unit 6b. , REG is calculated by the distance calculation means 6d only by shifting the position of rEG by one pixel in the front-rear direction (vertical direction of the captured image F). In addition, on the back side of the captured image F that is a compressed image, there is often a slight shift in the position of each edge hEG, lEG, rEG detected by the detection means 6b, and as a result, FIG. , (B), (c), the distance Z calculated by the distance calculation means 6d based on the edges hEG, lEG, rEG included in the captured image F of the same frame may be different.

  Therefore, in this embodiment, the distance calculation means 6d (first distance calculation means, second distance calculation means) performs the following processing to calculate the own vehicle 1 calculated based on the edges hEG, lEG, rEG. The calculation accuracy of the distance Z from the object to the object is improved. That is, the distance calculation means 6d is the reference position of the object (horizontal edge hEG, both vertical edges lEG, rEG) when calculating the distance Z from the own vehicle 1 to the object among a plurality of frames of captured images F by the monocular camera 3. The allocated distance is equally divided by the number of frames based on the number of frames where the lower end of each captured image F is located at the same pixel and the allocated distance from the vehicle 1 to the object allocated to the pixel. Then, the movement amount d of the object for each frame is derived, and the distance Z from the vehicle 1 to the object is calculated based on the derived movement amount d.

  For example, in the example shown in FIG. 4, since there are four frames in which the reference position of the object for calculating the distance is located at the pixel corresponding to the distance Zβ, the assignment from the own vehicle 1 assigned to the pixel to the object is performed. By dividing the distances Zβγ to Zαβ into four equally, the amount of movement d of the object per frame is derived. Then, the distance Z from the vehicle 1 to the object at each time t0 to t5 is calculated based on the derived movement amount d.

In this embodiment, the distance calculation means 6d calculates the distance Z by the following calculation formula.
Z = Zβγ-Dβ · (N-Nβ-1) / Nβ
Z: distance from own vehicle 1 to object Zβγ: smallest value of assigned distance from own vehicle 1 to object possessed by pixel to which distance Zβ is assigned Dβ: distance range of assigned distances Zβγ to Zαβ Nβ: distance Zβ The number of frames in which the reference position of the object (horizontal edge hEG, lower end of both vertical edges lEG and rEG) is located for the distance calculation to the pixel to which N is assigned N: The reference position of the object is assigned the distance Zβ Number from the first frame located at the pixel to the frame for calculating the distance Z (for example, N = 4 if the distance Z is calculated based on the captured image F of the monocular camera 3 at time t4)

  The assigned distance from the own vehicle 1 to the object assigned to each pixel is a value stored in the storage unit 4 in advance, and is a value depending on the attachment position of the monocular camera 3 to the own vehicle 1. Further, as the allocation distance used in the above-described calculation formula, an allocation distance allocated to a pixel corresponding to a position in the captured image F of the reference position of the object for calculating the distance may be used.

  That is, as described above, when the reference position of the object is moved from the pixel assigned the distance Zβ to the pixel assigned the distance Zγ, while the reference position of the object is located at the pixel assigned the distance Zγ, as described above. If the distance Z from the vehicle 1 to the object is calculated based on the assigned distances Zβγ to Zαβ and the reference position of the object moves to a pixel adjacent to the pixel to which the distance Zγ is assigned, the pixel to which the distance Zγ is assigned is calculated. The distance from the vehicle 1 to the object may be calculated based on the assigned distance.

(5) Recognition means 6e
The recognizing unit 6e recognizes the distance Z from the host vehicle 1 to the object based on the first to third distances by the first distance calculating unit and the second distance calculating unit of the distance calculating unit 6d. In this embodiment, the recognition unit 6e recognizes the distance Z from the own vehicle 1 to the object by averaging the first distance to the third distance.

  The recognizing unit 6e monitors a change in the distance Z from the own vehicle 1 to the object (another vehicle 12) recognized as described above from the captured images F of a plurality of frames.

(6) Notification means 6f
Based on the change in the distance Z from the host vehicle 1 to the object (other vehicle 12) monitored by the recognition unit 6e, the notification unit 6f superimposes a warning display on a map for car navigation being displayed on the travel monitor 7, for example. Or visually displaying the presence of an object to the driver. In addition, the speaker 8 generates a voice message and a warning sound as a recognition result to audibly notify the driver of the presence of the object. The driver's driving assistance is performed by the visual or auditory notification to the driver.

2. Operation Next, the operation of the object recognition apparatus 2 configured as described above will be described with reference to FIG. In this embodiment, as shown in FIG. 2A, a description will be given by taking as an example a state in which another vehicle 12 (object) is traveling with the front of the host vehicle 1 in the traveling lane as the preceding vehicle. Further, in the storage means 4, each edge hEG, lEG, as an object candidate detected by the detection means 6b based on the upper-view image tF obtained by projective transformation of the past captured image F captured in time series. Information on the position of each rEG in each image, information on a change in the distance Z from the vehicle 1 recognized by the recognition unit 6e to the object, and the like are stored.

  The process shown in FIG. 5 is a process executed while the host vehicle 1 is traveling. First, when shooting with the monocular camera 3 is started by turning on the ignition key of the host vehicle 1, the process shown in FIG. As shown, the signal of the captured image F in front of the vehicle 1 by the monocular camera 3 is taken into the ECU 6 by the acquisition means 6a every moment (step S1). 2A, the detection means 6b detects the horizontal edge hEG indicating the lower end of the other vehicle 12, and the vertical edges lEG and rEG indicating the left and right ends of the other vehicle 12 (step S2). As shown in (b) and (c), the photographed image F from which the edges hEG, lEG, and rEG are detected is projectively transformed into the upper view image tF by the converting means 6c (step S3).

  Next, the distance Z from the own vehicle 1 to the other vehicle 12 based on the horizontal edge hEG and the vertical edges lEG and rEG included in the upper view image tF is calculated by the distance calculation unit 6d, and the calculated distances Z are calculated. Based on this, the distance Z from the own vehicle 1 to the object is recognized by the recognition means 6e, and the time-series change of the recognized distance Z between the own vehicle 1 and the object is monitored (step S4).

  If the other vehicle 12 is abnormally close to the host vehicle 1, a warning message is displayed on the travel monitor 7, or a voice message or warning sound is output from the speaker 8. Is notified to the driver by the notification means 6f (step S5). That is, in this embodiment, if it is determined that the other vehicle 12 is abnormally approaching the host vehicle 1 by the function provided by the ECU 6 executing the program, the notification by the notification unit 6f is performed.

  As described above, according to this embodiment, when the distance from the vehicle 1 to the object (another vehicle 12) is recognized based on the captured image F captured by the monocular camera 3, it is included in the upward view image tF. The distance Z from the vehicle 1 to the object is calculated by the distance calculation means 6d (first distance calculation means) based on the position of the horizontal edge hEG indicating the lower end of the object to be detected.

  Then, by the distance calculation means 6d (second distance calculation means), the center that bisects the angle formed by the vertical edges lEG and rEG indicating the left and right ends of the object included in the upper-view image tF obtained by projective transformation of the captured image F The axis CA is obtained, and the distance Z from the vehicle 1 to the object is calculated based on the length from the lower end of both vertical edges lEG, rEG to the central axis CA and the angle of the vertical edges lEG, rEG with respect to the central axis CA. Then, the distance from the vehicle 1 to the object is recognized by the recognition unit 6e based on the distance Z from the vehicle 1 to the object calculated by the distance calculation unit 6d based on the edges hEG, lEG, rEG.

  Therefore, when an object far from the host vehicle 1 is included in the captured image F, the distances Z calculated by the distance calculation means 6d based on the edges hEG, lEG, rEG are the same as the host vehicle 1 and the object. Even if there is a possibility of deviating from the actual distance of the vehicle, the distance Z from the vehicle 1 to the object is recognized based on the distances Z calculated by different methods. It is possible to prevent the distance Z from greatly deviating and to improve the recognition accuracy of the distance Z from the vehicle 1 to the object.

  In addition, since the distance represented by one pixel is different in each region based on the infinity point in the captured image F, the first distance based on the horizontal edge hEG and the second and third distances based on the vertical edges lEG and rEG. Any of the distances may include an error, but the recognition means 6e is recognized by recognizing the distance Z from the vehicle 1 to the object by averaging the first distance to the third distance. The recognition accuracy of the distance Z from the vehicle 1 to the object can be further improved.

  In addition, the reference position of the object when calculating the distance Z from the vehicle 1 to the object in the captured image F of a plurality of frames, although the distance Z between the vehicle 1 and the object is relatively moving. Even when the distance is not moved, the distance calculation means 6d includes the number of frames in which the reference position of the object is located in the same pixel of each captured image F among the captured images F of the plurality of frames by the monocular camera 3. Based on the allocated distance from the own vehicle 1 to the object allocated to the same pixel, the allocated distance is divided equally by the number of frames to derive the moving amount d of the object for each frame, and the derived moving amount d Since the distance Z from the own vehicle 1 to the object is calculated based on the above, the actual distance Z from the own vehicle 1 to the object can be recognized more accurately.

  The present invention is not limited to the above-described embodiment, and various modifications other than those described above can be made without departing from the spirit of the present invention. For example, the monocular camera 3 is photographed rearward. You may attach to the rear part of the own vehicle 1 as possible. If comprised in this way, it can be monitored whether the other vehicle which approaches abnormally from the back of the own vehicle 1 exists.

  Further, the distance Z from the own vehicle 1 to the object may be calculated based on the horizontal edge hEG indicating the lower end of the object in the captured image F, and the object based on the upper-view image tF obtained by projectively converting the captured image F. A horizontal edge hEG indicating the lower end of the image may be detected.

  Further, when calculating the distance Z, it is not always necessary to derive the movement amount d for each frame of the object when the reference position of the object remains at the same pixel, and the horizontal edge hEG indicating the lower end of the object. The distance d of the object for each frame may be calculated when calculating at least one of the distance Z based on the above and the distance Z based on the vertical edges lEG and rEG indicating the left and right ends of the object. . The distance calculating means 6d may calculate the distance Z from the own vehicle 1 to the object based on at least one of the vertical edges lEG and rEG indicating the left and right ends of the object.

  Further, the present invention can be applied to various vehicle object recognition.

DESCRIPTION OF SYMBOLS 1 Own vehicle 2 Object recognition apparatus 3 Monocular camera (vehicle camera)
6d Distance calculation means (first distance calculation means, second distance calculation means)
12 Other vehicles (objects)
F Captured image tF Top-view image hEG Horizontal edge lEG, rEG Vertical edge CA Center axis d Movement amount

Claims (3)

In the distance recognition device that recognizes the distance from the vehicle to the object based on the captured image captured by the in-vehicle camera,
First distance calculating means for calculating a distance from the own vehicle to the object based on a position of a horizontal edge indicating a lower end of the object included in the captured image;
A central axis that bisects the angle formed by the vertical edges indicating the left and right ends of the object included in the upper-view image obtained by projective transformation of the captured image is obtained, and the central axis is at least one of the two vertical edges. Second distance calculating means for calculating a distance from the vehicle to the object based on a length from the vertical edge to the lower end of the vertical edge and an angle of the vertical edge with respect to the central axis;
Recognizing means for recognizing a distance from the own vehicle to the object based on a first distance by the first distance calculating means and a second distance by the second distance calculating means. Recognition device. The distance recognition device according to claim 1,
The distance recognition apparatus characterized in that the recognition means recognizes a distance to the object by averaging the first distance and the second distance. The distance recognition device according to claim 1 or 2,
At least one of the first distance calculating means and the second distance calculating means is
Among the captured images of the plurality of frames by the imaging means, the same number of frames where the reference position of the object is calculated at the same pixel of each captured image when calculating the distance from the vehicle to the object Based on the allocated distance from the host vehicle to the object allocated to a pixel, the movement distance of the object is derived by equally dividing the allocated distance by the number of frames, and the derived movement distance is A distance recognition apparatus that calculates a distance from the vehicle to the object based on the vehicle.