JP2020046239A - Image processing method, image processing apparatus and underground radar device - Google Patents

Abstract

To provide an image processing method and apparatus for facilitating association between a point in the captured image and a symmetric position thereof.SOLUTION: An image processing method includes the steps of: photographing an image of a scale while rotating a camera 71 at a position with a predetermined distance from the scale; recording a correspondence relation between a rotation angle θ of the camera 71 and the image of the scale; and measuring the distance information of a measurement target point MO1 based on a correspondence relation, and a ratio between the height of the camera 71 to the measurement target point MO1 and the height of the camera 71 to a reference point.SELECTED DRAWING: Figure 2

Description

  The present invention relates to an image processing method and an image processing apparatus for managing a point in a captured image as a distance or a position, and an underground radar apparatus using the image processing method and the like.

  In order to extract the arrangement information from the camera image, it is necessary to correct the distortion of the camera image. As a method of correcting such a distortion, a chart or a calibration pattern having a known figure or structure is directly faced with a camera, an image is taken, and a correction for correcting a camera image distortion using an obtained image is performed. It is known that information is acquired in advance (see Patent Documents 1 to 3).

  However, when correcting the distortion of the camera image by the above-described method, it is necessary to arrange a known chart or pattern over the entire area of the captured image and associate the position on the image with the position on the chart or pattern. In the case of a measurement method on the premise that the distance from the camera to the target is known, it may be necessary to arrange the chart far and wide. Furthermore, it is necessary to repeat the calibration every time the distance from the camera to the target changes.

JP 2013-109416 A JP 2003-123064 A JP 2008-283507 A

  The present invention has been made in view of the above background art, and has as its object to provide an image processing method and apparatus that facilitates associating a point in a captured image with a position of an object.

  In order to achieve the above object, the image processing method according to the present invention captures an image of the scale while rotating the camera at a position at a predetermined distance from the scale, and determines the correspondence between the rotation angle of the camera and the image of the scale from the image. The distance information of the measurement target point is measured based on the correspondence and the ratio of the camera height with respect to the measurement target point and the camera height with respect to the reference point.

  In the above image processing method, the image of the scale is taken while rotating the camera at a position at a predetermined distance from the scale, and the correspondence between the rotation angle of the camera and the image of the scale is recorded from the image, so that the front surface of the camera is covered. It is not necessary to arrange a large scale on the surface, and the workability of calibration can be improved. Further, since the measurement is based on the ratio of the camera height with respect to the measurement target point and the camera height with respect to the reference point, measurement can be performed regardless of the camera height, and it is not necessary to dispose a scale far and wide for calibration.

  In a specific aspect of the present invention, the measurement target distance information is a distance in a horizontal direction from a predetermined origin along a reference plane orthogonal to the rotation axis of the camera to a measurement target point. In this case, high-precision measurement focused on the reference plane becomes possible.

  In another aspect of the invention, the origin is a point on the road surface located directly below the camera. In this case, it is possible to perform measurement based directly on the camera.

  In yet another aspect of the invention, the reference point is a point on the scale located directly below the camera. In this case, the scale can be easily arranged, and the accuracy of the calibration can be easily increased.

  In yet another aspect of the present invention, the correspondence includes conversion information that calculates a real height or a virtual height of the rotation axis of the camera with respect to the scale as a fixed standard height and gives a distance from a pixel position. The distance from the origin to the measurement target point is given as a value obtained by multiplying the position obtained by the conversion information by a coefficient that is a ratio between the camera height and the standard height with respect to the measurement target point. In this case, even if the height of the camera with respect to the measurement target point is arbitrarily set regardless of the time of calibration, the distance from the origin to the measurement target point can be accurately calculated from the conversion information.

  In yet another aspect of the invention, the scale has a scale extending in a direction perpendicular to an axis of rotation for rotating the camera. In this case, distance measurement and calibration in a direction perpendicular to the rotation axis are facilitated.

  In still another aspect of the present invention, the camera has a line sensor extending in a direction perpendicular to a rotation axis for rotating the camera. In this case, the distance information of the measurement target point can be measured by one-dimensional information processing.

  In order to achieve the above object, an image processing apparatus according to the present invention operates an information recording unit that holds a correspondence relationship that gives distance information of a measurement target point from a position in an image of the measurement target point captured by the camera, and operates the camera. A signal processing unit that determines the distance information of the measurement target point based on the correspondence relationship and the height of the camera with respect to the measurement target point from the obtained image, and the correspondence relationship rotates the camera at a position at a predetermined distance from the scale. It is created by taking an image of the scale while making it.

  In the image processing method, since the camera is rotated at a position at a predetermined distance from the scale, an image of the scale is taken, and the correspondence between the rotation angle of the camera and the image of the scale is recorded from the image, so that the front surface of the camera is covered. It is not necessary to arrange a large scale on the surface, and the workability of calibration can be improved.

  In order to achieve the above object, an underground radar apparatus according to the present invention includes the above-described image processing apparatus, a camera, a transmitting section that transmits an exploration signal, a receiving section that receives a detection signal, and detection from the receiving section. And a measurement unit having a search processing unit that processes the signal.

  In the underground radar device, the above-described image processing device is incorporated, and the size and arrangement of the measurement target on the ground surface in the horizontal direction can be quickly grasped.

  According to a specific aspect of the present invention, an image processed by the image processing apparatus is linked with measurement data obtained by the measurement unit and stored as data. In this case, the underground exploration information and the arrangement information of the surface object can be matched and stored including the size.

(A) and (B) are a conceptual side view and a plan view illustrating a vehicle incorporating the image processing device and the underground radar device of the embodiment. (A) and (B) are the front view and the top view explaining the distance measurement using a peripheral image acquisition device. (A) is a front view explaining the calibration of the distance measurement, (B) is a front view explaining the calculation method of the distance and the width in the calibration. (A) is a figure explaining the detection state of the scale on a sensor surface, (B) has illustrated the photographed image at the time of measurement. (C) is a diagram illustrating a conversion table obtained by performing calibration at a standard height. It is a block diagram explaining an example of an underground measurement part. FIG. 3 is a block diagram illustrating an underground radar device including a peripheral image acquisition unit. It is a conceptual diagram explaining the two-dimensional map etc. displayed by image processing. 5 is a flowchart illustrating an example of an operation of the control device. FIG. 5 is a conceptual diagram illustrating two-dimensional calibration.

  Hereinafter, an underground radar apparatus including an image processing apparatus according to an embodiment of the present invention will be described with reference to the drawings.

  As shown in FIGS. 1A and 1B, the underground radar device 100 is mounted on a vehicle 200 such as an automobile and performs underground exploration while moving. The underground radar device 100 includes a plurality of underground measurement units 100A that perform underground measurement by radio waves, a peripheral image acquisition unit 100B that acquires a peripheral image of a measurement position or a measurement point, and a control device 100C that performs data processing and the like. And Among these, the peripheral image acquisition unit 100B and the control device 100C are the image processing device 110 that performs image processing on the peripheral image at the measurement position. Further, the plurality of underground measurement units 100A and the control device 100C which also functions as a search processing unit that processes a detection signal from the reception unit 20b (see FIG. 5) constitute a measurement unit 120 that collects search data. ing. The vehicle 200 is provided with a position information acquisition device 200a. The position information acquisition device 200a includes a GPS receiver, a speedometer, an accelerometer, and the like, measures the position or the amount of movement of the vehicle 200, and outputs the result as position-related information.

  The peripheral image acquisition unit 100B includes a camera 71 that captures an image of a periphery of a measurement point when moving with the plurality of underground measurement units (underground radar main body) 100A, and a rotation driving unit that adjusts the rotation posture of the camera 71 during calibration. 75. The camera 71 and the rotation drive unit 75 are fixed to the vehicle 200 via a support 78. The camera 71 captures and outputs a measurement target including a road surface, a shoulder, a sidewalk, a peripheral structure, and the like at a measurement position or a measurement point by the underground measurement unit 100A in units of a linear measurement area MA. That is, the camera 71 has a line-shaped angle of view, and the measurement area MA is elongated in the X direction orthogonal to the course of the vehicle 200. The camera 71 has a linear sensor (not shown) in which one or more rows of CMOS or other light detection elements are arranged as a light detection unit. The continuous image acquired by the camera 71 as the vehicle 200 moves is stored in the recording unit 72 and output to the control device 100C as image data. Note that the rotation drive unit 75 rotates the camera 71 around a rotation axis RX passing through the center of the lens 71a and tilts the camera 71 to a desired angle at the time of the calibration of the camera 71 described later. The rotation axis RX of the camera 71 extends in a direction orthogonal to the reference plane BS of the camera 71 and the optical axis AX of the lens 71a, and as a result, extends in a horizontal direction corresponding to the traveling direction of the vehicle 200 or parallel to the Y direction. .

  In the above, the peripheral image acquisition unit 100B has been described as including the rotation drive unit 75. However, the rotation drive unit 75 only needs to be present at the time of calibration. At the time of measurement, the optical axis AX of the lens 71a is set to the vertical line VL, that is, It is sufficient if the camera 71 is fixed to the support 78 by setting the camera 71 in parallel with the camera, and the rotation drive unit 75 can be omitted. In this case, the camera 71 is fixed in alignment with the rotation drive unit 75 during calibration.

  The distance measurement using the peripheral image acquisition unit 100B will be described with reference to FIG. The Z-axis corresponds to the vertical direction, the X-axis corresponds to a horizontal direction HD extending parallel to the paper surface and extending left and right, and the Y-axis corresponds to a horizontal direction perpendicular to the paper surface. The lens 71a of the camera 71 is arranged at a position at a height H from the measurement plane MS. Here, the height from the measurement surface MS means the distance to the measurement surface MS measured in the normal direction of the measurement surface MS, and is usually measured in the vertical direction. It is not limited. The measurement surface MS is assumed to be a road surface, and is assumed to extend along a predetermined plane RF perpendicular to the optical axis AX or the vertical line VL. Of the scheduled plane RF, the origin O1 immediately below the camera 71 is a point on the road surface located immediately below the camera 71, and is a base point for measurement using the camera 71. Therefore, the optical axis AX of the lens 71a is set to pass through the origin O1. That is, the vertical line VL passing through the origin O1 coincides with the optical axis AX of the lens 71a. The measurement area MA by the camera 71 targets an object on or near the predetermined plane RF including the origin O1 and perpendicular to the optical axis AX or the vertical line VL. The measurement area MA extends around the measurement line BL extending along the horizontal X axis including the origin O1. The lens 71a has a depth of focus that forms an image at a required resolution regardless of a difference in distance to each part of the measurement area MA. The reference plane BS including the measurement line BL and the vertical line VL extends in parallel with the XZ plane including the origin O1 and perpendicular to the traveling direction of the vehicle 200, and is a prerequisite for measurement by the camera 71. The reference plane BS is orthogonal to the rotation axis RX of the camera 71.

The position of the measurement target point MO1 in the direction of the angle θ with respect to the optical axis AX with respect to the camera 71, that is, the distance DM in the X direction from the origin O1 to the center of the measurement target point MO1 is given by H · tan θ. The width W of the measurement target point MO1 in the X direction is approximately given by H · δ / cos ² θ when the angular width viewed from the camera 71 is a relatively small value δ. The angle θ can be determined from the image acquired by the camera 71. When the pixel position at the center of the measurement target point MO1 in the imaging unit of the camera 71 is x (x is an integer), θ = f (x) And DM = H · tan (f (x)) = g (x), and the distance DM can be determined from the pixel position x. Here, θ = f (x) is a unique function related to the imaging characteristics and the like of the lens 71a. g (x) is a function arithmetically derived from f (x). When the pixel positions at both ends of the measurement target point MO1 are x1 and x2, W (x1, x2) = g (x2) -g (x1), and the width W can be determined from the pixel positions x1 and x2. . When the pixel position at the center of the measurement target point MO2 is x, the distance DM ′ in the X direction from the origin O1 to the center of another measurement target point MO2 is a negative value given by DM ′ = g (x). And the width W ′ of the measurement target point MO2 in the X direction is a positive value given by W ′ (x1, x2) = g (x2) −g (x1) or H · δ ′ / cos ² θ ′. Become.

  The function θ = f (x) is obtained by, for example, rotating the camera 71 around the rotation axis RX while checking the rotation angle or the inclination angle thereof, and recording the pixel position x of the target or the scale arranged at the origin O1. Can be determined. DM = g (x) can be arithmetically calculated by using the function θ = f (x). W (x1, x2) can also be arithmetically calculated using DM = g (x). W (x1, x2) can also be determined directly from the scale of the scale.

The calibration of the image processing apparatus will be described with reference to FIG. In FIG. 2, it is assumed that the lens 71 a is arranged at a position of a height H from the measurement surface MS, but every time the height H is changed, the corresponding DM = g (x) or W (x1 , X2), the calibration work becomes enormous. Therefore, calibration is performed at the standard height H0 set at a predetermined distance to determine the angle θ = f (x), which is the rotation angle, and DM = g (x) from the standard θ = f (x). And W (x1, x2). Specifically, the camera 71 is set so that the lens 71a has the standard height H0, and the scale SC is arranged on the calibration plane CS with reference to the reference point RP immediately below the vertical line VL. That is, the reference point RP is a center point on the scale SC disposed immediately below the camera 71. Then, the pixel of the reference point RP on the scale SC is extracted while changing the angle (rotation angle) θ of the camera 71 from, for example, −90 ° to + 90 ° by the rotation driving unit 75, and the pixel position x of the reference point RP is determined. I do. The rotation drive unit 75 incorporates a rotary encoder and other angle detection units, generates a trigger signal in a prescribed angle unit (for example, every 1 °), and captures an image by the camera 71 according to the trigger signal. Will be If the unit or division of the angle (rotation angle) θ is sufficiently small, θ corresponding to each x (x is an integer) can be given, and as a result, θ = f (x) can be determined. If the angle (rotation angle) θ is not finely divided, θ = f (x) can be determined by using the scale DS of the scale SC or by interpolation by interpolation. When the scale DS of the scale SC is used, the projection magnification 1 / cos ² θ is corrected in consideration of the angle (rotation angle) θ, and the overlap is eliminated at the boundary where the angle (rotation angle) θ at which the scale SC is read is switched. To provide a single azimuth or distance DM for each x. That is, boundary processing is performed when a conversion table described later is created, so that the distance to the target point is uniquely determined from the angle or pixel position of the target point.

As shown in FIG. 3B, a method of determining a position on the standard surface SF from an angle θ corresponding to an azimuth, considering a standard surface SF observed with the camera 71 fixed relatively, will be described. . Taking an image of the scale SC while rotating the camera 71 at a predetermined distance from the scale SC installed on the standard surface SF means rotating the calibration surface CS and the scale SC with the camera 71 fixed. . The scale SC has a scale DS extending in a direction perpendicular to the rotation axis RX of the camera 71. By capturing the image of the scale SC while gradually changing the angle θ, the correspondence between the angle θ that is the rotation angle of the camera 71 and the image of the scale SC is recorded. In the figure, the distance D (θ) in the X direction from the reference point RP to the reference projection point PP is H0, where H0 is the height of the lens 71a from the standard surface SF, that is, the height of the lens 71a with respect to the reference point RP. Tan θ = G (x). Here, G (x) is a function similar to g (x) described with reference to FIG. 2, and is data obtained by performing calibration at the standard height H0. The unit width W0 of the scale DS on the scale SC is obliquely enlarged and projected on the standard surface SF, and corresponds to the reference projection width Wp (θ). Reference projection width Wp (theta) is given by Wp (θ) = W0 / cos 2 θ ≒ H0 · φ / cos 2 θ. That is, as shown in FIG. 4A, considering a line sensor 71c extending in a direction perpendicular to the rotation axis RX of the camera 71, the scale DS including the origin image CD of the scale SC image measured for each θ. , That is, the number n (θ) of pixels on the line sensor 71c or the sensor surface 71d in the X direction of the unit width W0 (θ) is recorded in advance to calculate the width W of the measurement target at the reference projection point PP. can do. Specifically, when the actual measurement target at the reference projection point PP has a spread of the number of occupied pixels m in the X direction on the sensor surface 71d, the width W of the measurement target at the reference projection point PP in the X direction is W Is given as Wp (θ) = W0 / cos ² θ × (m / n) using the number of pixels n (θ) obtained in advance. Here, W0 / cos ² θ / n corresponds to an actual interval corresponding to the pixel interval, and is referred to as a pixel conversion interval PW.

  In FIG. 4A, the number of pixels on the y side of the sensor surface 71d of the line sensor 71c is three, but this is merely an example, and the timing of capturing an image by the camera 71 and the movement of the vehicle 200 In consideration of the speed, for example, the number of pixels on the y side can be set so as to obtain a continuous image in the Y direction.

In actual measurement, the height of the lens 71a from the measurement surface MS, that is, the height of the lens 71a with respect to the measurement target point MO1 or the like is a variable value H, but the geometric relationship shown in FIG. Therefore, the distance DM in the horizontal direction or the X direction from the origin O1 along the reference plane BS to the measurement target point MO1 along the reference plane BS is represented by θ for the angle of the measurement target and x for the pixel position.
DM = (H / H0) × D (θ) = (H / H0) × G (x) (1)
Given by Here, G (x) is data obtained by performing calibration at the standard height H0, as described above. The coefficient H / H0 is a ratio between the height of the camera 71 with respect to the measurement target point MO1 or the measurement plane MS and the height of the camera 71 with respect to the reference point RP or the standard plane SF. The distance DM is obtained by multiplying a function G (x) of the position obtained by the conversion information by a coefficient which is a ratio of the height H of the camera 71 to the measurement target point MO1 and the standard height H0. On the other hand, when the pixel positions at both ends of the measurement target are x1 and x2, the width W (x1, x2)
W = (H / H0) × {G (x2) −G (x1)} (2)
Given by The width W (θ, m) derived from the number m of pixels occupied by the measurement object is
W (θ, m) = (H / H0) × W0 / cos ² θ × (m / n)
= (H / H0) × PW × m (3)
Given by

  FIG. 4B illustrates a captured image. The captured image MP includes an image portion MPa corresponding to the measurement target point MO1, and based on Expression (1), a height H of the lens 71a from a pixel position x of the image portion MPa with respect to the image origin PO. In consideration of the above, the distance DM from the origin O1 of the measurement target point MO1 can be calculated. Further, based on the formula (2) or (3), the width W of the measurement target point MO1 is calculated from the pixel range x1, x2 or the pixel width w of the image portion MPa, taking into account the height H of the lens 71a. be able to. Here, the distance DM and the width W are distance information of the measurement target point MO1, and the arithmetic processing unit (signal processing unit) 101 described later measures the distance DM and the width W of the measurement target point MO1.

  FIG. 4C shows a conversion table in which data obtained by performing calibration at the standard height H0 is summarized. This conversion table is created by data processing after calibration, and is stored in the storage unit 102 (see FIG. 6), which is an information recording unit, as will be described in detail later. In the table, “pixel No.” means a pixel position x, “pixel conversion distance” means a function G (x), and “pixel conversion interval” means a pixel conversion interval PW. Here, the pixel conversion interval PW is not a function of θ but a function of x. The function G (x) and the pixel conversion interval PW are obtained by calculating the actual height of the rotation axis RX of the camera 71 with respect to the scale SC as a fixed standard height H0 and providing distance information from the pixel position x. The conversion table in FIG. 4C records the correspondence between the pixel position x obtained by converting the rotation angle θ of the camera 71 and the center image DC or the scale DS of the scale SC as conversion information. Indirectly, it can be said that the correspondence between the rotation angle θ of the camera 71 and the image of the scale SC is recorded.

  In the above description, the standard height H0 is the actual height of the rotation axis RX of the camera 71, but may be the virtual height of the rotation axis RX of the camera 71. That is, the height at which the scale SC is photographed by the camera 71 may be different from the standard height H0 that provides the conversion table of FIG. In this case, an operation for converting the result obtained by the camera 71 into the standard height H0 is required in advance.

  As shown in FIG. 5, the underground measurement unit 100A includes a radar signal generation unit 10, a transmission amplifier 21, a transmission antenna 31, a reception antenna 32, a reception amplifier 22, a sampling processing unit 25, An information acquisition unit 40 and a measurement-side control unit 60 are provided.

  The radar signal generator 10 of the underground measurement unit 100A generates the pulsed transmission wave SP1 at a predetermined timing under the control of the measurement-side control unit 60. The transmission amplifier 21 amplifies the transmission wave SP1 formed by the radar signal generating unit 10, and the transmission antenna 31 is driven by the transmission amplifier 21 to place the search signal S1 as a radio wave corresponding to the transmission wave SP1 on the ground. Radiate towards The radar signal generation unit 10, the transmission amplifier 21, and the transmission antenna 31 function as a transmission unit 20a that periodically transmits the search signal (radio wave) S1. The receiving antenna 32 receives the response wave (radio wave) S2 reflected by the buried object or the other object OB existing in the underground UG and returned as a detection signal. The signal corresponding to the received response wave S2 is amplified and output as a response wave SP2, and the sampling processing unit 25 generates a signal output SR that is a broadly-defined detection signal from the response wave SP2. The receiving antenna 32, the receiving amplifier 22, and the sampling processing unit 25 receive the response wave (detection signal) S2, which is a radio wave, and output a signal output (detection signal) SR in synchronization with the radio wave for underground exploration. It functions as the unit 20b. The sampling processing unit 25 uses the analog reference wave S3 corresponding to the transmission wave SP1 formed by the radar signal generation unit 10 for the response wave SP2 received by the reception antenna 32 and amplified by the reception amplifier 22. Correlation sampling processing is performed while shifting the timing of the wave S3, and a signal component having high correlation with the transmission wave SP is extracted from the response wave SP2. The signal component obtained by the sampling processing unit 25 is output as a digital signal in association with the distance in the depth direction corresponding to the timing difference from the reference wave S3. That is, the signal output SR of the sampling processing unit 25 is an intensity output value indicating the distribution of the buried object or the other object to be searched OB, and calculates the amplitude of the signal component obtained from the response wave SP2 for each distance in the depth direction. It has become something. The sampling processing unit 25 includes a weighting filter and the like in addition to the correlator that performs the correlation sampling processing on the response wave SP2 and the reference wave S3.

  A plurality of underground measurement units 100A are mounted on the vehicle 200 in the X direction, and measure the distribution of the detection state of the response wave in the XY plane by moving the vehicle 200 in the Y direction along the road surface. can do.

  Referring to FIG. 6, control device 100C controls the operation of underground measurement unit 100A and also controls the operation of peripheral image acquisition unit 100B. The control device 100C is a management unit that comprehensively manages and stores measurement data obtained by the underground measurement unit 100A and image data obtained by the peripheral image acquisition unit 100B. That is, control device (management unit) 100C converts the image obtained by imaging unit 71 into a size proportional to the actual state on measurement plane MS or planned plane RF, and gives the Y coordinate in the traveling direction of vehicle 200. At the same time, the data is linked with the data measured by the underground measurement unit 100A and stored.

  The control device (management unit) 100C includes an arithmetic processing unit 101, a storage unit 102, an input unit 103, a display unit 104, and an interface unit 105. The control device 100C specifically includes a computer on which a program for underground exploration is installed. The user can not only observe the measurement result by the underground measurement unit 100A on the display unit 104, but also perform measurement conditions. The setting or selection of signal processing conditions, display conditions, and the like can be instructed to the arithmetic processing unit 101 via the input unit 103.

  The arithmetic processing unit 101 operates based on programs and data stored in the storage unit 102, performs processing based on information obtained from the input unit 103 and the interface unit 105, and stores the progress of the processing and the result in the storage unit 102. It is stored and presented on the display unit 104. In particular, the arithmetic processing unit 101 operates the underground measurement unit 100A via the interface unit 105 based on a program or the like, and processes the detection signal from the underground measurement unit 100A to obtain the point measurement information and the path section measurement. Information, two-dimensional measurement information, and the like can be displayed on the display unit 104. Here, the point measurement information is an intensity pattern representing a detection signal obtained by an exploration at a specific measurement point while the vehicle 200 is moving as a function of a distance in a depth direction, and the path cross-section measurement information is a detection signal of the detection signal. 5 is a chart showing a relationship between a moving distance with respect to a reception intensity and a reflection time indicating an elapsed time from transmission to reception of a radio wave. The two-dimensional measurement information is a reflection distribution map obtained by combining detection signals from a plurality of underground measurement units 100A and projecting them on an XY plane. When a specific moving distance in the chart is designated, the arithmetic processing unit 101 selects an image (local two-dimensional image) corresponding to the designated moving distance and its surroundings from among the images taken by the imaging unit 71. Map), a local two-dimensional map of the extracted road surface or the like is displayed on the display unit 104, and the reflection distribution map can be superimposed and displayed. The arithmetic processing unit 101 generates a two-dimensional map from the image data obtained by the peripheral image obtaining unit 100B as a signal processing unit. Specifically, the arithmetic processing unit 101 converts the size of the image data obtained by the peripheral image obtaining unit 100B based on the conversion table (see FIG. 4C) corresponding to the equation (1) obtained by the calibration. The corrected one-dimensional information is stored in advance, and a two-dimensional image is generated from the one-dimensional information. As a result, the image data obtained by the peripheral image acquisition unit 100B is processed into a local two-dimensional map that spreads in the X and Y directions. The two-dimensional map that is the basis of this local two-dimensional map represents a road surface extending along the movement route of the vehicle 200 and its surroundings. The information is stored in the storage unit 102 in a form linked to the measurement information and the like.

  FIG. 7 shows an example in which a reflection distribution map is superimposed and displayed on a local two-dimensional map of a road surface. In this case, an image of the road surface RO including the side lines MO11, MO13, the center line MO12, and the like is created as the local two-dimensional map TD, and the reflection distribution pattern RD constituting the two-dimensional measurement information is formed on the local two-dimensional map TD. It is superimposed and displayed. This reflection distribution pattern RD corresponds to, for example, a cavity directly below the road surface RO.

  An example of the measurement operation of the arithmetic processing unit 101 will be described with reference to FIG. The arithmetic processing unit 101 instructs the underground measurement unit 100A and the peripheral image acquisition unit 100B to operate via the interface unit 105, and captures image data obtained by the imaging unit 71 while performing underground exploration (step S01). Next, the arithmetic processing unit 101 assigns a Y coordinate to the captured image data (step S02). The Y coordinate of the image data is determined based on the output of the position information acquisition device 200a, the arrangement information of the imaging unit 71, and the like. Next, the arithmetic processing unit (signal processing unit) 101 corrects the X coordinate of the captured image data (step S03). More specifically, the distance information of each point is calculated based on the conversion table (see FIG. 4C) corresponding to the above equation (1) obtained by the calibration with respect to the image data to which the Y coordinate is added in step S02. Is determined, and one-dimensional information whose size is corrected based on the obtained distance DM is calculated and stored in the storage unit 102. At this time, a two-dimensional map in which the one-dimensional information is integrated is created and stored in the storage unit 102. The arithmetic processing unit (signal processing unit) 101 links image data including one-dimensional information and a two-dimensional map obtained by integrating the one-dimensional information with the measurement data obtained by the measurement unit 120 when storing the image data in the storage unit 102. Next, the arithmetic processing unit 101 receives designation of an object by the operator (step S04). When specifying an object, the one-dimensional information obtained in step S03 and the two-dimensional map obtained by integrating the one-dimensional information are presented to the operator via the display unit 104, and the operator uses the input unit 103 to input the one-dimensional information or the two-dimensional map. Specify a specific area. Note that the arithmetic processing unit 101 may extract the object. The arithmetic processing unit 101 detects the center position, the X width, or the Y width of the object specified in step S04 (step S05). The determination of the center position and the X width of the object can be performed on a two-dimensional map, but the conversion table (see FIG. 4C) corresponding to the above equations (1) to (3) obtained by the calibration is used. It can be calculated directly from the image data captured in step S01 based on the above.

  As described above, in the image processing method of the present embodiment, an image of the scale SC is taken while rotating the camera 71 at a position at a predetermined distance from the scale SC, and the rotation angle of the camera 71 and the image of the scale SC are taken from the image. Since the correspondence is recorded, there is no need to arrange a large scale SC so as to cover the front surface of the camera 71, and the workability of calibration can be improved. Further, since the measurement is based on the ratio between the height of the camera 71 with respect to the measurement target point MO1 and the height of the camera 71 with respect to the reference point RP, the measurement can be performed regardless of the height of the camera 71. You don't have to.

[Others]
The structures, shapes, sizes, and arrangements described in the above embodiments are merely schematic representations so that the present invention can be understood and implemented. Therefore, the present invention is not limited to the embodiments described above, and can be modified in various forms without departing from the scope of the technical idea shown in the claims.

  For example, as shown in FIG. 4C, a conversion table is prepared on the premise that the standard height H0 is used. If a function G (x) for x and the horizontal distance is prepared in advance, only the conversion table is read based on the pixel position x, and the coefficient (H / H0) is set according to the installation height of the camera 71. This eliminates the need for multiplying arithmetic processing.

  The image processing device 110 can be incorporated into various inspection devices other than the underground radar device 100. In addition, images of floors, ceilings, walls, and the like, as well as road surfaces, can be taken to obtain images in which distortion related to arrangement and size has been corrected.

  In the above description, the calibration in the case of capturing a one-dimensional image has been described. However, the same applies to the calibration in the case of capturing a two-dimensional image, and FIG. If not only the one parallel to the Y axis shown but also one parallel to the X axis is added and calibration is performed on the scale immediately below while rotating on two axes, the two-dimensional image can be collectively Position conversion and distance conversion become possible. Specifically, as shown in FIG. 9, θx is swung from −A ° to + B ° using a rotation axis parallel to the Y axis, and θy is −C ° using a rotation axis parallel to the X axis. Shake at ~ + D °. At this time, the combination (θx, θy) covers the entire range (θx: -A ° to B °, θy: -C ° to D °). Specifically, for example, the reference point RP or the scale SC is photographed by scanning covering the entire calibration surface CS as indicated by a path AP.

  The camera 71 can tilt the optical axis AX with respect to the vertical line VL at the time of measurement. In this case, the measurement is performed in a state where the angle (rotation angle) θ is offset, and a conversion table corresponding to this is prepared. At the time of calibration, the tilt state (installation angle at the time of measurement) is set to be immediately above the reference point RP, and the state is used as a reference state to perform pixel extraction and the like of the reference point RP, thereby creating a conversion table. In this case, the range of θ is, for example, −90 ° + a to + 90 ° −a (a is an offset amount at the time of installation).

  In the above description, the distance information (that is, the position and the size) is calculated for the target portion on the target by photographing a planar target having a known distance, but the present invention is not limited to this. For example, if a laser measuring device or other measuring device is installed adjacent to the camera 71 and the distance to the target portion or the inclination angle of the target portion can be measured by this, even if the target is non-planar and the distance is unknown, It is also possible to measure distance information (that is, position and size) for the portion of interest.

  DESCRIPTION OF SYMBOLS 10 ... Radar signal generation part, 20a ... Transmission part, 20b ... Reception part, 25 ... Sampling processing part, 40 ... Distance information acquisition part, 60 ... Measurement side control part, 71 ... Camera, 71 ... Image pickup part, 71a ... Lens, 71c: line sensor, 71d: sensor surface, 72: recording unit, 75: rotary drive unit, 100: underground radar device, 100A: underground measurement unit, 100B: peripheral image acquisition unit, 100C: control unit, 101: arithmetic operation Processing unit, 102: storage unit, 110: image processing device, 120: measuring unit, 200: vehicle, 200a: position information acquisition device, AX: optical axis, RX: rotation axis, BS: reference plane, RP: reference point, SF: Standard plane, MS: Measurement plane, O1: Origin, VL: Vertical line, PO: Image origin, PP: Reference projection point, BL: Measurement line, CS: Carriage SC, scale, DS, scale, MA, measurement area, MO1, MO2, measurement target point, MP, photographed image, MPa, image part, OB, exploration target, RD, reflection distribution pattern, RF, planned plane , RO: road surface, SP1: transmission wave, SP2: response wave, TD: local two-dimensional map, UG: underground

Claims (10)

Take an image of the scale while rotating the camera at a predetermined distance from the scale,
Record the correspondence between the rotation angle of the camera and the image of the scale from the image,
An image processing method, wherein distance information of the measurement target point is measured based on the correspondence relationship and a ratio between the height of the camera with respect to the measurement target point and the height of the camera with respect to a reference point.   The image processing method according to claim 1, wherein the distance information of the measurement target is a distance in a horizontal direction from a predetermined origin along a reference plane orthogonal to a rotation axis of the camera to the measurement target point.   The image processing method according to claim 2, wherein the origin is a point on a road surface located immediately below the camera.   The image processing method according to claim 1, wherein the reference point is a point on the scale disposed immediately below the camera. The correspondence includes conversion information that calculates a real height or a virtual height of the rotation axis of the camera with respect to the scale as a fixed standard height and gives a distance from a pixel position,
The distance from the origin to the measurement target point is given as a value obtained by multiplying a position obtained by the conversion information by a coefficient that is a ratio of a height of the camera to the measurement target point and the standard height. The image processing method according to any one of claims 2 to 4.   The image processing method according to claim 1, wherein the scale has a scale extending in a direction perpendicular to a rotation axis that rotates the camera.   The image processing method according to claim 1, wherein the camera has a line sensor extending in a direction perpendicular to a rotation axis that rotates the camera. An information recording unit that holds a correspondence that gives distance information of the measurement target point from a position in the image of the measurement target point captured by the camera,
A signal processing unit that determines distance information of the measurement target point based on the correspondence and the height of the camera with respect to the measurement target point from an image obtained by operating the camera,
The image processing device, wherein the correspondence is created by capturing an image of the scale while rotating the camera at a position at a predetermined distance from the scale. An image processing apparatus according to claim 8,
Said camera;
An underground radar apparatus comprising: a transmission unit that transmits a search signal; a reception unit that receives a detection signal; and a measurement unit that includes a search processing unit that processes a detection signal from the reception unit.   The underground radar device according to claim 9, wherein the image processed by the image processing device is linked to measurement data obtained by the measurement unit and stored as data.

Images