WO1998005922A1

WO1998005922A1 - Calibration method

Info

Publication number: WO1998005922A1
Application number: PCT/JP1997/002752
Authority: WO
Inventors: Kosuke Sato; Takayuki Kataoka; Shozo Hirose; Motohide Yasukawa
Original assignee: Komatsu Ltd.; Osaka Gas Information System Research Institute Co., Ltd.
Priority date: 1996-08-07
Filing date: 1997-08-06
Publication date: 1998-02-12
Also published as: JP3453734B2; JPH1047920A

Abstract

A calibration method which can improve accuracy by eliminating an error resulting from a binarization processing. When a measurement plate (1) as a calibration object is subjected to yaw revolution, pitch revolution and movement in a longitudinal direction reltative to a three-dimensional visual sensor (4), the three-dimensional visual sensor (4) acquires a depth map at each movement point, the depth map so acquired is then processed to obtain a three-dimensional coordinates value of each pixel for each depth map, and a calibration value is obtained from this three-dimensional coordinates value.

Description

Specification

Calibration method Technical field

The present invention relates to a calibration method for a three-dimensional measuring device required for measuring a three-dimensional position of an object to be measured by an imaging means such as a camera. Background art

Conventionally, a method shown in FIG. 19 has been known as a calibration method in this type of three-dimensional measuring apparatus (for example, see Japanese Patent Application Laid-Open No. H5-2481988). This method uses a measuring plate 52 engraved with grid lines 51 (or many points) indicating coordinates as a calibration object, and the measuring plate 52 is controlled by a controller 53. The calibration data is obtained by recognizing the grid line 51 with the three-dimensional visual sensor 54 while moving it back and forth, and processing it with the computer 55. In this case, when recognizing the grid line 51, an image processing method called binarization processing is used.

However, according to the conventional calibration method described above, when the grid 51 is extracted by image processing, the grid 51 can be faintly extracted due to a change in the amount of light in the surrounding environment. Or, the center of grid line 51 could not be extracted and a deviated line was extracted, and the resulting coordinate values deviated from the correct position, resulting in accurate calibration. There is a problem that can not be performed.

The present invention has been made in order to solve such a problem, and it is an object of the present invention to provide a calibration method capable of removing errors due to a binarization process and achieving higher accuracy. And for the purpose. Disclosure of the invention

To achieve the above-mentioned object, the calibration method according to the present invention comprises:

The surface of the device under test is irradiated with light by the light projecting device, and the reflected light is imaged by the image capturing device. The information of the imaged reflected light is used to calculate the reflected light based on the principle of triangulation. A calibration method for a three-dimensional measuring device for measuring a position,

A distance image at each moving point when the calibration object is moved one rotation, a pitch rotation, and a back and forth movement with respect to the imaging means is taken by the imaging means, and the taken distance image is taken. The processing is performed to obtain three-dimensional coordinate values for each distance value of each pixel, and a calibration value is obtained from the obtained three-dimensional coordinate values.

In the calibration method of the present invention, for example, a calibration target such as a measuring plate is rotated once, a pitch or a back and forth by a fixed amount with respect to an imaging means such as a camera. Then, the distance image at each moving point is acquired by the imaging means. Then, by processing the acquired distance image, a three-dimensional coordinate value of each pixel for each distance value is obtained, and a calibration value is obtained from the obtained three-dimensional coordinate value.

According to the present invention, a correspondence table between the 256-level distance information (that is, this is a distance value) replaced with the brightness information of each pixel by rotating and moving the calibration object back and forth and the three-dimensional coordinate values (Look-up table), it is possible to eliminate the occurrence of errors due to binarization processing as in conventional extraction of grid lines by image processing, and a high-precision carrier. A break can be realized.

In the present invention, when processing the distance image captured by the image capturing unit, a median around a specific pixel for one image is processed. The value is set as the distance value of the pixel, and the obtained distance value is set multiple times at each position, and the median values of the multiple times are set as the distance value of the pixel. It is preferable to set. By doing so, when the pixel value of the acquired image changes due to irregular reflection on the surface of the calibration object, smoothing processing for removing noise from the acquired image can be performed. High-precision calibration can be performed.

As a noise removal method, when processing a range image captured by the imaging unit, an average value of the circumference of a specific pixel in one image is set as a distance value of the pixel. At the same time, the distance value thus obtained can be set a plurality of times at each position, and the average value of the plurality of times can be set as the value of the distance value of the pixel. The imaging means may be configured to replace the acquired distance image information with light / dark information and output the information.

Further, the imaging means acquires a three-dimensional image of the measured object from reflected light of the coded pattern light projected on the surface of the measured object by the light projecting means. Or a method of acquiring a three-dimensional image of the measured object by imaging the reflected light of the light projected on the surface of the measured object by the two light emitting means by the light projecting means. It may be. BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 is a system configuration diagram of a three-dimensional measuring apparatus according to one embodiment of the present invention,

Figure 2 shows a flow chart for creating a look-up table in the X direction.

Figure 3 shows the flow for creating an X-direction look-up table. Char.

Figure 4 is a flowchart ③ for creating a look-up table in the X direction.

Figure 5 is a flow chart for creating a look-up table in the X direction.

Figure 6 shows a flow chart for creating an X-direction lookup table.

Figure 7 shows a flow chart for creating a lookup table in the Y direction.

Figure 8 is a flow chart for creating a look-up table in the Y direction.

Figure 9 is a flow chart for creating a look-up table in the Y direction.

FIG. 10 shows a flowchart for creating a lookup table in the Y direction.

Figure 11 is a flow chart for creating a look-up table in the Y direction.

Figure 12 is a flow chart for creating a Z-direction look-up table,

Figure 13 shows a flow chart for creating a Z-direction look-up table,

Fig. 14 shows flow charts (3) and (3) for creating a look-up table in the Z direction.

Figure 15 is an illustration of the movement of the measuring plate for acquiring a distance image in the X direction.

FIG. 16 is an explanatory diagram of a calculation method of the distance image data in the X direction. Figure 17 shows the movement of the measuring plate for acquiring the distance image in the Y direction. Figure,

Fig. 18 is an illustration of the movement of the measurement plate for acquiring a distance image in the Z direction.

FIG. 19 is an explanatory diagram of a conventional calibration method. BEST MODE FOR CARRYING OUT THE INVENTION

Next, a specific embodiment of the calibration method according to the present invention will be described with reference to the drawings.

FIG. 1 shows a system configuration diagram of a three-dimensional measuring apparatus according to one embodiment of the present invention. The three-dimensional measuring apparatus of the present embodiment includes a measuring plate 1 as a calibration object constituted by a plain plate, a measuring plate moving device 2 for moving the measuring plate 1 to a desired position, A measuring plate control device 3 for controlling the measuring plate moving device 2; a three-dimensional visual sensor 4 provided opposite to the measuring plate 1; and a control device for controlling the measuring plate control device 3. And a computer 5 for storing and processing the distance image obtained by the three-dimensional visual sensor 4 in accordance with the moving position of the measuring device 1. And a gonio rotary device 2c provided on the linear slide 2a via a rotary table 2b, whereby the measuring plate 1 rotates one turn with respect to the three-dimensional visual sensor 4 (up and down). Rotation around rotation axis), pitch rotation (rotation around horizontal rotation axis) And it has been configured to be movable back and forth. The three-dimensional visual sensor 4 is configured to irradiate a laser beam (coded pattern light) onto the surface of the measurement plate 1, and to capture an image of light reflected from the surface of the measurement plate 1. A CCD camera is provided as imaging means, and an image obtained by the CCD camera is sent to the computer 5. In this way, the distance image at each moving point is three-dimensionally visualized while moving the measuring plate 1 by a fixed amount based on the control signal from the computer 5. Acquired by sensor 4.

The three-dimensional visual sensor 4 outputs the distance between the three-dimensional visual sensor 4 and the measuring plate 1 by converting the information into bright and dark information such that a bright area is displayed near and a dark area is displayed far away. Then, what three-dimensional coordinates actually correspond to the value indicating the light and darkness is taught by performing the calibration.

Next, a calibration method according to the present embodiment will be described with reference to FIGS. 15 to 18 and flowcharts shown in FIGS. 2 to 14. FIG.

First, a flow for creating an X-direction lookup table for storing X-direction position data will be described with reference to FIGS.

S1: All three axes of the movement axis of the measuring plate 1 with respect to the three-dimensional visual sensor 4, in other words, the vertical rotation axis (Y axis), the horizontal rotation axis (X axis), and the front-rear movement axis (Z axis). Move each axis to the origin as a measurement reference.

S 2: In order to acquire distance image data in the X direction, as shown in FIG. 15 (a), the measuring plate 1 is moved + 30 ° around the vertical rotation axis with respect to the three-dimensional visual sensor 4. +30 for one corner. At an angle of

S3 to S5: The distance image data of the measurement plate 1 is acquired by the CCD camera of the three-dimensional vision sensor 4, and the acquired image is used as a media to remove noise due to reflection of the surface of the measurement plate 1 or the like. Correct with an unfilter (smoothing filter). Note that this median filter replaces the median value of the pixel to be measured with the median value (median value) of an area near the pixel (for example, 3 × 3). This process is repeated until the image at each position is acquired three times.

S6 to S7: When three distance images are acquired at each position, the median value (median value) of the same pixel is calculated by comparing the values of the same pixel. Set the final distance value as the final distance value. Thereafter, the obtained distance image is written to a distance image file in a storage device in the computer 5.

S8 to S9: Distance image data is acquired in the same manner as described above by shifting the front and rear positions of the measurement plate 1 backward by a fixed pitch (for example, 0.1 mm). This data is acquired for all distance image files. (1500 times) Repeatedly write the distance image file XP 0. kmt to xp 149. 9 kmt when the X axis rotates forward. In this way, for example, all range image data during the forward rotation of the X axis in the measurement range of 75 mm to 22 mm are acquired.

S10 to S11: Next, as shown in Fig. 15 (b), when the measuring plate 1 is set at an angle of-30 ° with respect to the three-dimensional visual sensor 4, At the same time, return the measuring plate 1 to the home position in the front-rear direction.

S12 to S18: Performs the same processing as S3 to S9 described above, for example, to obtain all the range image data during the negative rotation of the X axis in the measurement range of 75 mm to 25 mm. The acquired data is written to the distance image file xn O .kmt ~ xnl 4 9 9.kmt at the time of negative rotation of the X axis,

S19: Initial setting is first performed to create a look-up table in the X direction. In this initial setting, an undefined value is set to 1 ut -X (when the X axis is positively rotated) and 1 ut -w (when the X axis is negatively rotated) indicating the X coordinate of the lookup table.

S20 to S22: Reads the range image file xpO.kmt to xpl49.9kmt at the time of the normal rotation of the X axis, created as described above.

S23 to S26: Distance that can be obtained depending on how the measurement plate 1 is set. Since there is a gap in the image data and that data becomes noise, processing to eliminate this noise is performed. That is, first the distance image ixj (i; 256 pixels, j; 242 pixels) is compared with the previous image data for each pixel. For example, the value of the pixel at n rows and m columns of the image data xpl O. The image data at the previous position xP9.Compare with the value of the pixel in the same n rows and m columns of kmt, and the pixel is within 50 mm from the top, and the difference between the values of the preceding and following pixels is If the value is 1 2 8 or more, the values of lut X [m] [n] [0] to lutx [m] [n] [255] at that pixel position are undefined to invalidate this data To value

S27 to S28: When the value of 1ut-X of the pixel value is indefinite, set the moving distance when capturing the file read in 1ut-X of that pixel. In this case, the <part of xp «. Kmt of the file indicates the order of the file, and multiplying this file number by 0.1 to obtain the movement amount (hi x O.1 mm). As a result, the position of the first appearing code value of each pixel is stored in the lut —.

S29: i; 256 pixels, j; 242 The processing of S23 to S28 is performed for all pixels.

S30: Determine whether or not all range image files xpO.kmt to xp149.kmt have been processed, that is, whether or not the force has exceeded the value of 1550. In the case of 1,500, the people «

S31 to S41: Distance image file xn O. Kmt to xn 149. 9 kmt to xn 149 during negative rotation of X-axis For distance image file xn O. kmt ~ xpl 4 9 9. Perform the same processing as the processing in kmt (S20 ~ S30), and set the pixel position read at 1 ut-w of each pixel.

S42 to S49: For the value (Z; 0 to 255) of each pixel, the relative distance in the Z-axis direction from the initial position when the X-axis is positively rotated. Let i P = 1 ut-x [i] [j] [z] and the relative distance in the Z-axis direction from the initial position during the negative rotation of the X axis in = 1 ut-w [j] [i] [z]. And set it to ip and in. Then, when these ip and in are not indefinite values, the X coordinate for the pixel value at that position is calculated by the following equation.

X-(ip-in) / (2 x t a n 0) ①

：: Tilt angle of measuring plate 1

As shown in FIG. 16, in this formula, (ip-in) represents the difference between the image position at the time of the X-axis positive rotation and the image position at the time of the X-axis negative rotation. Therefore, the X coordinate of point A, which represents the same pixel value when the measurement plate 1 rotates forward and when it rotates negatively, is expressed by the following equation. The X coordinate of this point A is stored in the above-mentioned memory 1ut-X [i] [j] [z] to save memory.

On the other hand, when i P or in n is an undefined value, an undefined value is set in the memory 1 ut -X [j] [i] [z].

S50: Performs the above-described processing of S44 to S49 for all pixels i; 256 pixels, j; 24 pixels, completes the look-up table in the X direction, and completes the flow. finish.

Next, a look-up table in the Y direction for storing the position data in the Y direction is created in accordance with each of steps T1 to T50 shown in FIGS. 7 to 11.

As shown in Fig. 17 (a) and (b), the flow for creating the Υ-direction look-up table is as follows. The measurement plate 1 is installed with a pitch angle of ± 20 ° (see step 17). The distance image data in the Y-direction is acquired by using the distance image data in the Y direction, and the distance image data is acquired as the distance image data yp O. Kmt to ypl 4 9 9. kmt and yn O. Kmt to yn 1 4 9 9.Except for writing to kmt, This is the same as the processing in each of steps S1 to S50. Therefore, the detailed description of this flow is omitted.

Next, a flow for creating a Z-direction look-up table for storing Z-direction position data will be described with reference to FIGS. 12 to 14.o

U 1: All axes of the movement axis of the measuring plate 1 with respect to the three-dimensional visual sensor 4, in other words, the vertical rotation axis (Y axis), the horizontal rotation axis (X axis), and the front-rear movement axis (Z axis) Move each axis to the origin as a measurement reference.

U2 to U4: Obtain the distance image data of the measurement plate 1 by the CCD camera of the three-dimensional visual sensor 4, and correct the obtained image by a median filter (smoothing filter). This process is repeated until the image at each position is acquired three times.

U5 to U6: When three distance images are acquired at each position, the value of the same pixel is compared to find the median value (median value), and this median value is used as the final distance value. Set as. Thereafter, the obtained distance image is written to a distance image file in a storage device in the computer 5.o

U7 to U8: Obtain distance image data in the same manner as above by shifting the front and rear position of the measurement plate 1 backward by a fixed pitch (for example, 0.1 mm). Acquire this data for all distance image files. (100 times) Repeatedly write to the Z-axis range image file zO.kmt to z99.kmt. Thus, for example, 100 mn! Acquire all the Z-axis range image data within the measurement range of ~ 200 mm.

U 9: Initial settings are made to create a Z-direction look-up table. In this initial setting, an undefined value is set to 1 ut-z indicating the Z coordinate of the lookup table. U10 to U11: Read the pixel position z0.kmt of the z-axis range image file created as described above.

U12 to U15: Perform processing to eliminate noise in the range image data obtained in the same manner as steps S23 to S26 in FIGS. That is, first, each pixel of i and j (i: 256 pixels, j; 242 pixels) is compared with the immediately preceding image data, and the pixel is within 50 mm from the head, and If the difference between the pixels is greater than 128, the value of 1 ut-z for that pixel is made indefinite to invalidate this data.

U16 to U17: When the value of 1utz of the pixel value is undefined, set the moving distance when the file read at 1ut-z of that pixel is captured. As a result, the position of the first appearing code value of each pixel is stored at 1 u t -z.

U18: i; 256 pixels and j; 242 pixels U11 to U17 are processed for all pixels.

U 19: It is checked whether or not the processing has been performed for all the range image files z0.kmt to z99.9kmt.

U 20 to U 25: For the value (Z; 0-255) of each pixel, the position where the specified pixel value is first found is i = 1 ut -z [j] [i ] Set to [z]. Then, when this i is not an indefinite value, the value of i is stored in lut—z [j] [i] [z] to calculate the Z coordinate for the pixel value at that position. .

On the other hand, if i is an indefinite value, check whether all pixel values are present and return to step U22.

U26: Performs the processing of U22 to U25 described above for all pixels i; 256 and j; 242 pixels, completes the Z-direction look-up table and ends the flow I do. According to the present embodiment, a correspondence table (lookup table) between the brightness information of each pixel and the three-dimensional coordinate value is created using a plain plate, and the conventional grid lines are formed by image processing. Significantly higher accuracy can be achieved compared to what is extracted.

In the present embodiment, when the distance image is processed, the correction is performed using the median value. However, the correction may be performed using the average value instead of the median value. It is also possible to combine these corrections with the median value and the correction with the average value, such that the first correction is performed with the median value and the second correction is performed with the average value. It is possible.

In the present embodiment, an example in which a three-dimensional image is obtained by irradiating a coded pattern light to a measurement plate has been described.However, the present invention is directed to a method in which a three-dimensional image is obtained by two cameras. It goes without saying that it can be applied to this.

Claims

The scope of the claims

The surface of the device under test is irradiated with light by the light projecting device, and the reflected light is imaged by the imaging device. The information of the imaged reflected light is used to determine the position of the device under test based on the principle of triangulation. A calibration method for a three-dimensional measuring device that measures

A distance image at each moving point when the calibration object is moved one rotation, a pitch rotation, and a back and forth movement with respect to the imaging means is captured by the imaging means, and the captured distance image is processed. A three-dimensional coordinate value of each pixel for each of the distance images, and a calibration value is obtained from the obtained three-dimensional coordinate value.

In processing the distance image captured by the imaging unit, a median value around a specific pixel in one image is set as a distance image of the pixel, and the distance image obtained in this way is set. 2. The method according to claim 1, wherein the median value is set a plurality of times at each position, and the median values of the plurality of times are set as correction values of the distance image of the pixel.

In processing the distance image captured by the imaging unit, an average value around a specific pixel in one image is set as the distance image of the pixel, and the distance image obtained in this manner is set. 2. The calibration method according to claim 1, wherein is set a plurality of times at each position, and an average value of the plurality of times is set as a correction value of the distance image of the pixel.

4. The calibration method according to claim 1, wherein the imaging unit replaces the acquired distance image information with light / dark information and outputs the information.

The imaging unit projects light onto the surface of the device under test by the projection unit. The method according to any one of claims 1 to 3, wherein a three-dimensional image of the object to be measured is acquired from reflected light of the coded pattern light.

The imaging means is for acquiring a three-dimensional image of the measured object by imaging the reflected light of the light projected on the surface of the measured object by the light projecting means with two cameras. 4. The method for calibrating according to any one of claims 1 to 3.