JP2011064639A - Distance measuring device and distance measuring method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a distance measuring device and method capable of highly accurately measuring the distance to an object such as a guide signboard or sign on a road using an on-vehicle monocular camera. <P>SOLUTION: The distance measuring device includes: an object detection part 101 to which images of different frames photographed by a monocular camera are input, and which detects an object from the images; an object detection result accumulating part 102 for accumulating images of two or more frames of the object; an object scale calculation part 103 for calculating the length or the number of pixels of the detected object on the image; an object direction calculation part 104 for calculating the direction of the object from characteristic points of the object; a speed-to-object calculation part 105 for calculating the speed of a vehicle toward the object based on the calculated direction of the object and vehicle speed information input from a vehicle speed sensor; and an object distance calculation part 106 for calculating the distance to the object based on the calculated speed to the object and the calculated scale of the object on the image and outputs it. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a distance measuring device and a distance measuring method for measuring a distance from an object using a captured image.

  2. Description of the Related Art Conventionally, it has been considered to capture road conditions with a camera mounted on a vehicle and to assist driving or control the vehicle based on an image obtained thereby. In such a technique, it is important to correctly measure the distance to the captured object.

  Distance measurement to an object can be classified into distance measurement using a stereo camera and distance measurement using a monocular camera depending on the configuration of the camera. The distance measurement by a stereo camera is a method of measuring a distance using a plurality of cameras whose relative positional relationships are known, imaging the same object, and using parallax of the captured images.

  On the other hand, distance measurement with a monocular camera is a method of calculating the distance to an object using a captured image obtained by one camera. Compared with a method using a stereo camera, a method using a monocular camera has a simple system structure, does not require position calibration between cameras, and is low in cost. The present invention relates to a distance measuring method using a monocular camera.

  In addition, there is a method for calculating the distance from an object of which the actual size is known, using the image size of the object on the image and the spec parameters of the camera.

  However, there are many guide signs, non-standardized signs, or other objects of unknown actual size on the road. It is possible to analyze the driving situation of the driver, for example, whether there is no violation of traffic rules or whether there is a dangerous action, based on the distance to these objects and the change pattern of the distance. Therefore, it is necessary to measure the distance to an object whose actual size is unknown, for example, a guide signboard.

  As a method for measuring the distance between a camera and an object using a specific geometric constraint condition existing between the camera and the object, for example, there is a technique disclosed in Patent Document 1. The technique disclosed in Patent Document 1 is based on a method of detecting a grounding point of a preceding vehicle from a captured image and calculating a distance from the preceding vehicle from the position of the grounding point. Is calculated with high accuracy, and the actual size of the vehicle width and the distance of the preceding vehicle are calculated with high accuracy.

  As a method for measuring the distance between the camera and the target object using the scale change of the stationary object on the image and the own vehicle speed, there is a technique disclosed in Patent Document 2, for example. The technique disclosed in Patent Document 2 calculates a change in scale of a structure fixed near a road on an image, and first measures the distance between the vehicle and the structure. Next, the distance between the host vehicle and the preceding vehicle is calculated using the relative positional relationship between the host vehicle, the structure, and the preceding vehicle on the traveling road.

  Conventionally, there is a method called motion parallax as a distance measurement method that does not use the speed of the host vehicle. At least eight corresponding point pairs are detected from two images taken from different positions, and the amount of movement of the position of the camera that captures the two images is calculated from the position of the corresponding points on the image. Consider the image as a stereo camera image and measure the distance.

  As a distance measurement method using the motion parallax method, for example, there is a technique disclosed in Patent Document 3. The technology disclosed in Patent Document 3 determines the scale amount of the three-dimensional real space coordinates that cannot be determined by a simple motion parallax method from the camera installation height calculated from the detected road surface and the actual camera installation height, The three-dimensional real space coordinates of the corresponding points are calculated.

US 2007/0154068 JP 2003-247824 A JP 2006-349607 A

“Detection of text on road signs from video,” IEEE Trans. ITS, vol.6, no.4, Dec 2005.

  There are many signboards, message boards, and non-standard signs on the road. It is possible to analyze the driving situation of the driver using the distance to these objects and the change pattern information of the distance.

  However, the technique disclosed in Patent Document 1 is based on the premise that the preceding vehicle as the object is at the grounding point, and cannot be applied to an object that does not ground. Since it is difficult to estimate the contact point in the vertical direction on the road surface of an object such as a guide signboard, it is difficult to calculate the distance to the guide signboard or the like using this technique.

  On the other hand, in the technique disclosed in Patent Document 2, the output of the vehicle speed sensor is used as the vehicle speed in order to measure the distance between the vehicle and the object (road structure), but the distance to the guide signboard is measured. It does not necessarily match the speed required to do.

  That is, the speed required to measure the distance to the guide sign is a speed directed to an object such as the guide sign, while the output of the speed sensor is the speed in the moving direction of the host vehicle. In particular, in the case of curves, hills, and irregular driving of the own vehicle (lane change, etc.), a correct distance measurement cannot be realized because the error between the two is large.

  On the other hand, in the technique disclosed in Patent Document 3, it is necessary to obtain 100 or more corresponding points from a plurality of planes in order to realize practical distance measurement. However, since the road surface occupying a large image area has a poor texture, it is difficult to obtain corresponding points on the road surface. In addition, it is necessary to avoid corresponding points in moving objects such as preceding cars. Therefore, since it is difficult to acquire a large number of effective corresponding points, it is difficult to measure the distance in the coordinate calculation of the three-dimensional real space.

  The present invention has been made in view of the above-described conventional circumstances, and provides a distance measuring device and a distance measuring method capable of accurately measuring the distance to a guide signboard or an unstandardized sign whose actual size is unknown. The purpose is to do.

  The distance measuring device of the present invention is a distance measuring device that measures a distance to an object using a monocular camera and a vehicle speed sensor mounted on a vehicle, and from the two or more captured images taken by the monocular camera, An object detection unit for detecting an object, an object scale calculation unit for calculating a scale on an image of the detected object, and an object direction calculation for calculating the direction of the object from a feature point of the object An object-oriented speed calculating unit that calculates a speed at which the vehicle is directed to the object based on the calculated direction of the object and the vehicle speed information of the vehicle speed sensor, and a speed directed to the calculated object And an object distance calculation unit for calculating a distance to the object based on the calculated scale of the object on the image.

  According to the above configuration, since the distance to the object is calculated based on the object directing speed that is the speed at which the vehicle is directed to the object, the vehicle is traveling on a curve, a hill, etc. The correct distance measurement can be realized even when driving regularly (lane change, etc.). Further, since a monocular camera is used, the structure of the system is simple, position calibration between cameras is not necessary, and low cost can be realized.

  In the distance measurement device of the present invention, the object direction calculation unit calculates a normal direction of the surface of the object based on three feature points on a non-collinear line on the surface of the object.

  According to the above configuration, since the normal direction of the surface of the object is calculated based on the three non-collinear feature points on the surface of the object, the signboard or the non-standardized sign whose actual size is unknown The normal direction can be calculated accurately.

  In the distance measurement device of the present invention, the object orientation speed calculation unit calculates a speed at which the vehicle is directed toward the object based on a normal direction of the surface of the object.

  According to the above configuration, since the speed at which the vehicle is directed to the object is calculated based on the normal direction of the surface of the object, the vehicle is traveling on a curve, a hill, etc. Even when you are changing lanes, correct distance measurement can be realized.

  Further, the distance measuring device of the present invention has a ranging condition determining unit that determines whether a product of a scale on the image of the object and the distance to the object is known or unknown, and the object distance calculating unit Calculates the distance to the object based on the product when the product is known.

  According to the above configuration, it is determined whether the product of the scale on the image of the object and the distance is known or unknown, and when the product is known, the distance to the object is calculated based on the product. Calculation can be omitted, calculation load can be reduced, and distance calculation can be performed at high speed.

  In addition, the distance measuring device of the present invention has a time series changing unit that changes the playback order from an image captured at a distance closer to the object to an image captured at a distance closer to the target object in the offline image processing. The object is detected in the captured image, the object detection and tracking unit that searches for the object in the image captured at a long distance, and the reproduction order is changed from an image captured at a long distance to an image captured at a short distance. And a time series adjustment unit.

  According to the above configuration, imaging is performed at a close distance in offline image processing for analyzing the driver's operation status, for example, whether there is no violation of traffic rules or dangerous behavior, based on the record of the drive recorder. Since an object having a large scale is detected from the obtained image and the object is traced until it becomes a small scale in an image captured at a long distance, the detection rate of the object at a long distance can be improved.

  The distance measuring method of the present invention is a distance measuring method for measuring the distance to an object using a monocular camera and a vehicle speed sensor mounted on a vehicle, and is based on two or more captured images taken by the monocular camera. Detecting the object; calculating a scale on the image of the detected object; calculating a direction of the object from a feature point of the object; and Based on the direction and the vehicle speed information of the vehicle speed sensor, calculating the speed at which the vehicle is directed to the object, and based on the calculated speed directed to the object and the calculated scale on the image of the object And calculating a distance to the object.

  According to the above configuration, since the distance to the object is calculated based on the object directing speed that is the speed at which the vehicle is directed to the object, the vehicle is traveling on a curve, a hill, etc. The correct distance measurement can be realized even when driving regularly (lane change, etc.). Further, since a monocular camera is used, the structure of the system is simple, position calibration between cameras is not necessary, and low cost can be realized.

  As described above, according to the distance measuring device and the distance measuring method according to the present invention, the distance to the object is calculated based on the object directing speed that is the speed at which the vehicle is directed to the object. Correct distance measurement can be realized even when driving on a curve, hill, etc., or when the vehicle is traveling irregularly (lane change, etc.). Further, since a monocular camera is used, the structure of the system is simple, position calibration between cameras is not necessary, and low cost can be realized.

The figure for demonstrating schematic structure of the distance measuring device concerning Embodiment 1 of this invention. The flowchart for demonstrating the processing step of the distance measurement method concerning Embodiment 1 of this invention. The figure for demonstrating the calculation step of the distance from a camera to a target object in Embodiment 1 of this invention. The figure for demonstrating schematic structure of the distance measuring device concerning Embodiment 2 of this invention. The figure for demonstrating schematic structure of the distance measuring device concerning Embodiment 3 of this invention. The figure for demonstrating the image scale of the image | photographed image frame and target object in embodiment of this invention.

  Hereinafter, embodiments of a distance measuring apparatus and method according to the present invention will be described with reference to the drawings. The basic principle of the present invention is to measure the distance from an object based on the time change of the scale of the object viewed from the own vehicle. The object is detected from two or more captured images taken by a monocular camera. Calculate the scale on the image of the detected object, calculate the direction of the object from the feature points of the object, and direct the vehicle to the object based on the calculated direction of the object and the vehicle speed information of the vehicle speed sensor. The distance to the object is calculated based on the calculated object orientation speed and the calculated scale on the image of the object.

  According to the distance measuring device and method of the present embodiment, the distance to the object is calculated based on the object directing speed, which is the speed at which the vehicle is directed to the object. Correct distance measurement can be realized even when the vehicle is traveling irregularly (lane change, etc.). Further, since a monocular camera is used, the structure of the system is simple, position calibration between cameras is not necessary, and low cost can be realized.

(Embodiment 1)
FIG. 1 shows the configuration of the first embodiment. The distance measuring device of the present embodiment is a distance measuring device that measures the distance to an object such as a guide signboard using a monocular camera mounted on a vehicle and a vehicle speed sensor, and images of different frames (imaging images) taken by the monocular camera. Image), the object detection unit 101 that detects the object from the image, the object detection result accumulation unit 102 that accumulates images of two or more frames of the object, and the length of the detected object on the image An object scale calculation unit 103 for calculating the size or the number of pixels, an object direction calculation unit 104 for calculating the direction of the object from the feature points of the object, and the calculated direction of the object and the input vehicle speed sensor. Based on the vehicle speed information, an object orientation speed calculation unit 105 that calculates a speed at which the vehicle is directed toward the object, a calculated object orientation speed, and a scale on the image of the calculated object. Based on, and a subject distance calculator 106 calculates and outputs the distance to the object. The “feature point” of an object does not mean a characteristic feature of the object but a predetermined point in the object. If it is a predetermined point corresponding to a predetermined position in the object and it is possible to keep track of the position, the predetermined point is qualified as a feature point.

  The object detection unit 101 detects an object from the input image. The input image may be a real-time image from the camera or an offline image accumulated in advance (detailed in the third embodiment). Examples of the object include a guide signboard, a sign, and a message board on the road. An object having a plane perpendicular to the road surface can be selected as the object. As a detection method, a known method for detecting a specific shape or a specific color region from an image can be used.

  The object detection result accumulation unit 102 accumulates object detection results over at least two frames. Here, the accumulated frames are expressed as frames at time t0 and t1. For example, as shown in FIG. 6, when a camera 301 mounted on a vehicle travels in a direction of 30 ° with respect to an object 302 such as a guide signboard in front, a frame 307 captured at time 0 (t0), And the frame 309 photographed at time 1 (t1) is accumulated.

  The object scale calculation unit 103 calculates an image scale that is a scale on the image of the object in a plurality of frames. For example, the length of the specific contour line of the object or the pixel distance between specific feature points is used. The image scale is grasped by, for example, the number of pixels. Here, the image scale to be calculated is denoted by L0 and L1. In the case of FIG. 6, the image scale L0 of the object 306 in the frame 307 captured at time 0 (t0) is calculated, and the image scale L1 of the object 308 in the frame 309 captured at time 1 (t1) is calculated.

  The object direction calculation unit 104 calculates the direction of the object in the real space, that is, the normal direction of the plane of the signboard as the object. A specific method in the processing of the object orientation calculation unit 104 will be described below with reference to FIG.

  In FIG. 3, the camera 301 is shooting in the actual movement direction K of the vehicle on which the camera 301 is mounted. The position of the camera 301 at time t0 is the origin of the world coordinate system XYZ in real space, the direction in which the object 302 is located is the Z-axis direction, the vertical direction is the Y-axis, and the direction orthogonal to the YZ plane is the X-axis.

  After the object detection unit 101 detects the object 302 from the image, first, the object direction calculation unit 104 detects at least three or more feature points from the object region. Here, the three feature points on the non-collinear line of the object 302 are denoted by P1, P2, and P3. In addition, the image coordinates at time t0 are expressed by the following formula.

  In addition, the image coordinates at time t1 are expressed by the following equations.

  Further, the three-dimensional real space coordinates of the feature points Pi, i = 1, 2, 3 are expressed by the following equations.

  Note that the coordinates of the feature points P1, P2, and P3 shown in the equations (1) and (2) are not real space coordinates, and only an image on the xy plane is seen, so there is no z component. Further, since the vehicle moves from time t0 to time t1, the image coordinates of the feature points P1, P2, and P3 are also shifted. According to Non-Patent Document 1, the following relationship exists between real space coordinates and image coordinates.

  Note that T in equation (4) means a transposed matrix.

In equation (4), d is an unknown parameter for the distance traveled by the camera 301 in the Z-axis direction from time t0 to time t1 (see FIG. 3). Further, f is a known parameter for the focal length of the camera 301. Therefore, the world coordinates X _i , Y _i , and Z _i of the camera 301 at time t0 are the movement distance d that the camera 301 moved in the Z-axis direction from time t0 to time t1, and the image coordinates of the feature points. It can be calculated as a function.

  Next, the object orientation calculation unit 104 considers that the three feature points P1, P2, and P3 of the object 302 are on the same plane, and obtains a normal direction vector N of the plane. As shown in FIG. 3, the normal direction vector N is calculated from the following equation.

  That is, the normal direction vector N of the object 302 is an outer product of the vector A from the feature point P1 to the feature point P2 and the vector B from the feature point P3 to the feature point P2. Since the normal direction vector N of the object 302 is related to the absolute distance between the feature points on the image, it is regarded as a function of the moving distance d that the camera 301 has moved in the Z-axis direction from time t0 to time t1. I can do it. This is because, as can be seen from Equation 4, P1 to P3 are functions including d.

  Next, the object direction calculation unit 104 calculates the normalized normal direction of the normal direction vector N of the object 302, that is, the same direction unit vector N / ‖N‖ of the vector N as three feature points P1 and P2. , P3 and the focal length f of the camera 301. However, ‖N‖ is the length of the vector N. The quantities (Nx, Ny, Nz) in the X, Y, and Z axis directions in the normalization normal direction are as follows.

  However, α in the above formula is represented by the following formula.

  Cx, Cy, and Cz in equations (6) and (7) are parameters that can be calculated from the image coordinates of the three feature points P1, P2, and P3, and are described in detail in Non-Patent Document 1. That is, the normalization normal direction can be calculated from the image coordinates of the three feature points P1, P2, and P3 and the focal length f of the camera 301. The normalization normal direction is normalized by setting the length of the normal direction vector N of the object 302 as 1, so that it is not an absolute distance between feature points on the image but a relative change in absolute distance (eg, P1P2 vs. P3P2). ), It is no longer a function of the moving distance d that the camera 301 has moved in the Z-axis direction from time t0 to time t1. That is, even if the moving distance d of the camera is not known, the normal direction of the object can be calculated from the image coordinates of the feature points.

  Next, the object orientation speed calculation unit 105 calculates the speed at which the camera 301 is directed toward the object 302. If a measurement value such as a speed meter input from a vehicle speed detection device (not shown) is displayed as V, the object orientation speed Vz is calculated by the following equation.

  That is, the object directing speed Vz is obtained by the product of the measured value V of the speedometer and the quantity Nz in the Z-axis direction in the normalization normal direction. Since the object directing speed Vz is only in the signboard direction (Z-axis direction), it has only the z component.

  Next, the object distance calculation unit 106 calculates the object distance from the camera 301 to the object 302. First, the distance d that the vehicle has moved in the Z-axis direction from time t0 to time t1 is calculated by the following equation.

  That is, it is obtained by the product of the object directing speed Vz and the time difference (t1-t0).

  Next, from the image scales L0 and L1 calculated by the object scale calculation unit 103 and the distance d that the vehicle has moved in the Z-axis direction from the time t0 to the time t1, the object distance calculation unit 106 A distance D1 to the object 302 is calculated by the following formula.

  That is, the distance D1 is obtained by the product of the distance d that the vehicle has moved in the Z-axis direction from time t0 to time t1 and the ratio L0 / (L1−L0) of the image scales L0 and L1 in the image coordinates.

  FIG. 2 shows a calculation flowchart of the first embodiment. Note that the scale of the object can be obtained from the distance between the feature points P1, P2, and P3 of the object, but in the flowchart of FIG. 2, the scale is calculated from the lengths L0, L1, etc. of the image of the object. An example is shown.

  In step S201, the camera 301 acquires an image at time t0. Next, in step S202, the object detection unit 101 detects the object from the image at time t0. Next, in step S203, the object orientation calculation unit 104 detects image coordinates of the three feature points P1, P2, and P3 of the object. In step S204, the object scale calculation unit 103 calculates the scale of the object from the image at time t0.

  On the other hand, in step S205, the camera 301 acquires an image at time t1. Next, in step S206, the object detection unit 101 detects the object from the image at the time t1. Next, in step S207, the object orientation calculation unit 104 detects the image coordinates of the three feature points P1, P2, and P3 of the object. In step S204, the object scale calculation unit 103 calculates the scale of the object from the image at the time t1.

  Next, in step S208, the object orientation calculation unit 104 determines the real space temporary coordinates (Xi, Yi, Zi), i = 1 of the three feature points P1, P2, P3 of the object at time t0 and time t1. , 2 and 3 are calculated. Next, in step S209, the object orientation calculation unit 104 calculates the normal direction (Nx, Ny, Nz) of the object plane.

  Next, in step S210, the object orientation speed calculation unit 105 calculates the speed in the object direction. Next, in step S211, the object distance calculation unit 106 calculates a movement distance in the object direction. In step S212, the object distance calculation unit 106 calculates the distance to the object.

  Thus, according to the distance measuring device and method of the present embodiment, the distance to the object is calculated based on the object directing speed that is the speed at which the vehicle is directed to the object. Even when the vehicle is traveling or when the vehicle is traveling irregularly (lane change, etc.), correct distance measurement can be realized. Further, since a monocular camera is used, the structure of the system is simple, position calibration between cameras is not necessary, and low cost can be realized.

(Embodiment 2)
If the distance measurement method shown in the first embodiment is executed for all frames, the calculation load becomes heavy. FIG. 4 shows the configuration of the second embodiment. Compared to Embodiment 1 shown in FIG. 1, distance measurement condition determination for determining whether the product of the scale on the image of the object and the distance to the object is known or unknown based on the output of the object detection unit 101 Part 410 is added.

  For the same object, the distance calculation after the third time, for example, the time t2, can be easily calculated using the results at the times t0 and t1. Referring to FIG. 6, the third time corresponds to time 2 (t2), and the image scale of the object 310 is L2 in the frame 311 at time 2 (t2). Note that the distance from the camera 301 to the object 302 at time 0 (t0) is D0, the distance from the camera 304 to the object 302 at time 1 (t1) is D1, and the distance from the camera 305 at time 2 (t2) to the object. The distance to the object 302 is D2.

  The relationship between the sizes L0, L1, L2 of the objects 306, 308, 310 on the image frames 307, 309, 311 and the distances D0, D1, D2 from the cameras 301, 304, 305 to the object 302 on the z-axis ( Table 1) shows.

  Specifically, the distance D2 from the camera 305 to the object 302 at time 2 (t2) is the scale Li (i = 0, 1, 2,...) On the image of the object 302 and the distance Di ( Since the product of i = 0, 1, 2,... is constant, the distance D2 can be calculated directly from the scale L2 at time t2 by calculating D1 · L1 from the first embodiment.

That is, the distance D2 is obtained by the product of the distance D1 from the camera 301 to the object 302 at time t1 and the ratio L1 / L2 of the scales L1 and L2 in the image coordinates.

  The distance measurement condition determination unit 410 determines whether the product α of the scale and distance on the image of the object is known or unknown from the relationship between the images at time t0 and time t1 shown in (Table 1), and the product α is known. In this case, the respective processes in the object detection result accumulation unit 102, the object scale calculation unit 103, the object orientation calculation unit 104, and the object orientation speed calculation unit 105 are omitted. Then, in the object distance calculation unit 106,

The distance D2 after the time t2 is calculated using the product α.

  According to the distance measuring device and the distance measuring method of the present embodiment, it is determined whether the product α of the scale on the image of the object and the distance is known or unknown, and when the product α is known, the target is based on the product α. Since the distance to the object is calculated, repeated calculations can be omitted, the calculation load can be reduced, and the distance calculation can be performed at high speed.

(Embodiment 3)
Usually, the scale on the image of an object at a long distance is small, making detection difficult, and the calculated error is relatively large. On the other hand, if a known tracking method is used for the object detected in the previous frame, the detection rate from the next frame is improved.

  Accordingly, in the offline processing for analyzing the accumulated image, the detection order of the input image is reversed, the processing of Embodiment 1 or Embodiment 2 is executed from the frame with the large image scale of the object, and the distance result is obtained. It is also possible to return to the conventional frame order and output it later.

  FIG. 5 shows a configuration diagram of the third embodiment. The distance measuring device according to the present embodiment is similar to the configuration of the first embodiment, with the time-series changing unit 501 that changes the reproduction order from an image captured at a distance closer to the object to an image captured at a far distance. Normal object detection and tracking unit 502 that detects an object in an image captured at a distance and searches for an object in an image captured at a far distance, and an image captured at a close distance from an image captured at a distant distance A time series adjustment unit 510 for returning in order is added.

  The time-series changing unit 501 arranges the images from the rear in time series in order to reversely reproduce the stored images from the rear. The object detection tracking unit 502 detects a large scale object in an image captured at a close distance, and tracks the object until the object becomes a small scale in an image captured at a far distance. The distance time series adjustment unit 510 returns the reversely reproduced image to the normal order.

  According to the distance measuring device and the distance measuring method of the present embodiment, the offline analysis of the driver's operation status, for example, whether there is no violation of traffic rules or no dangerous behavior, based on the record of the drive recorder In the image processing by, a large-scale object is detected in an image captured at a close distance, and the object is tracked until it becomes a small scale in an image captured at a distant distance. Can be improved.

  The effect of the present invention will be briefly described with reference to FIG. In FIG. 6, the following conditions were set.

  Under the above conditions, when the conventional technique shown in Patent Document 2 is used, the travel distance is 13.9 m, and the distance at time t0 is 69.5 m. However, if the technique shown in this embodiment considering the moving direction of the camera is used, the traveling distance in the Z-axis direction is 12 m, and the actual distance at the time t0 is 60 m. That is, an error of about 14% occurs if the moving direction of the camera is not taken into consideration.

  Although various embodiments of the present invention have been described above, the present invention is not limited to the matters shown in the above-described embodiments, and those skilled in the art will be able to understand based on the claims and the description of the specification and well-known techniques. Such changes and applications are also within the scope of the present invention, and are included in the scope for which protection is sought.

  The present invention provides distances to guide signs and non-standardized signs whose actual size is unknown even when the vehicle is traveling on a curve, hill, etc., or when the vehicle is traveling irregularly (such as changing lanes). Can be used as a distance measuring device and a distance measuring method.

DESCRIPTION OF SYMBOLS 101 Object detection part 102 Object detection result storage part 103 Object scale calculation part 104 Object direction calculation part 105 Object direction speed calculation part 106 Object distance calculation part 301,303,304,305 Camera 302,306,308 , 310 Object 307, 309, 311 Frame 410 Distance measurement condition determination unit 501 Time series change unit 502 Object detection tracking unit 510 Time series adjustment unit

Claims (6)

A distance measuring device for measuring a distance to an object by a monocular camera and a vehicle speed sensor mounted on a vehicle,
An object detection unit for detecting the object from two or more captured images taken by the monocular camera;
An object scale calculator for calculating a scale on the image of the detected object;
An object direction calculation unit for calculating the direction of the object from the feature points of the object;
An object-oriented speed calculating unit that calculates a speed at which the vehicle is directed to the object based on the calculated direction of the object and vehicle speed information of the vehicle speed sensor;
A distance measuring apparatus comprising: an object distance calculating unit that calculates a distance to the object based on the calculated speed directed to the object and the calculated scale of the object on the image. The distance measuring device according to claim 1,
The object orientation calculation unit is a distance measuring device that calculates a normal direction of the surface of the object based on three feature points on a non-collinear line on the surface of the object. The distance measuring device according to claim 2,
The object orientation speed calculation unit is a distance measurement device that calculates a speed at which a vehicle is directed to the object based on a normal direction of a surface of the object. The distance measuring device according to claim 1,
A ranging condition determining unit that determines whether a product of a scale on the image of the object and a distance to the object is known or unknown;
The object distance calculation unit is a distance measurement device that calculates a distance to the object based on the product when the product is known. The distance measuring device according to claim 1,
In the offline image processing, a time-series changing unit that changes the playback order from an image captured at a short distance to an image captured at a distance from the object,
An object detection tracking unit that detects the object in an image captured at a short distance and searches for the object in an image captured at a long distance;
A distance measurement device comprising: a time-series adjustment unit that changes a reproduction order from an image captured at a long distance to an image captured at a short distance. A distance measuring method for measuring a distance to an object with a monocular camera and a vehicle speed sensor mounted on a vehicle,
Detecting the object from two or more captured images taken by the monocular camera;
Calculating a scale on an image of the detected object;
Calculating the orientation of the object from the feature points of the object;
Calculating a speed at which the vehicle is directed to the object based on the calculated direction of the object and vehicle speed information of the vehicle speed sensor;
Calculating a distance to the object based on the calculated speed directed to the object and the calculated scale on the image of the object.

Images