JP7469989B2 - Camera Calibration Plate - Google Patents

Description

The present invention relates to a calibration plate used for calibrating a camera.

In cameras used as visual sensors for robots to perform three-dimensional measurements, internal and external parameters are obtained by calibration in order to correct image distortion and calculate the distance to the target point. Internal parameters are specific to the camera, and include focal length, optical center, pixel scaling factor, and lens distortion coefficient. External parameters give the positional relationship of the camera to the target object. For example, the external parameters of a camera fixed to a workplace are information indicating the position and orientation of the camera in the workplace coordinates, which allows conversion from the world coordinates of the workplace to the camera coordinates. In a camera mounted on the moving part of a robot, the camera coordinates also change according to the movement of the robot, but if the external parameters are obtained for one posture of the robot, the change in the camera coordinates accompanying the robot's movement can be easily calculated from the movement of the robot's joints using forward kinematics. In a stereo camera, the relative positions of the two cameras can be determined by obtaining the external parameters of the two cameras, and the distance between the centers of the cameras (baseline length) can be obtained, so the distance to the target point can be calculated using this value and other internal and external parameters.

A grid-like pattern is often used as a calibration pattern for calibrating a camera. Calibration is performed by comparing the positional relationships between grid points on the actual calibration pattern with an image captured by a camera. Therefore, when calibrating, it is necessary to accurately know information indicating the physical characteristics of the calibration pattern, such as the grid spacing of the calibration pattern to be used.

The appropriate calibration pattern is selected from many different types, with different overall pattern sizes, aspect ratios, grid spacing, etc., depending on the resolution of the camera to be calibrated, the focal length of the lens, the shooting distance, the aspect ratio of the image sensor, etc. The characteristic information of the calibration pattern actually used, such as the grid spacing, is manually input from the terminal during the calibration work by referring to a list of calibration patterns, etc.

JP 2010-103730 A JP 2003-307466 A

As mentioned above, during calibration work it was necessary to select one from many types of calibration patterns and specify the grid spacing accurately. Internal parameters needed to be calibrated when adjusting the focus position, etc. Furthermore, external parameters needed to be calibrated every time the baseline length of the stereo camera or the angle between the optical axes of multiple cameras was changed depending on the work content. Therefore, it was necessary to manage the calibration pattern information and use it correctly during calibration work, but managing the calibration pattern information was a tedious job. Above all, a major problem was that even if there was an error in specifying the grid spacing, etc., the calibration work would seemingly proceed without any problems, making it difficult to notice the error.

Patent Document 1 describes the addition of information related to image processing as tag information using a two-dimensional code to the camera calibration plate of an in-vehicle camera. Image processing is, for example, a process of generating an overhead image from an image captured diagonally downward and superimposing vehicle width guide lines, etc. It is said that by providing a different composite pattern for each vehicle model of different size as information related to image processing, the work of setting the image processing contents can be simplified even when cameras are shared between vehicle models. However, due to the characteristics of an in-vehicle camera, it is only necessary to determine the positional relationship of the camera with respect to the vehicle as an external parameter, so that calibration can be performed once before shipping, and there is no need to use multiple calibration patterns. Furthermore, the tag information in Patent Document 1 does not include characteristic information such as the grid spacing of the calibration pattern.

Patent Document 2 describes a calibration chart having first marks provided in at least three locations and second marks that are provided so as to be visually identifiable from the first marks and are used to calculate internal parameters. The first marks are used to obtain the approximate positions of the second marks. However, the first marks in Patent Document 2 also do not include information such as the grid spacing of the second marks, and it is necessary to manage the characteristic information of the calibration pattern separately.

The present invention has been made in consideration of the above, and aims to provide a calibration plate that does not require or at least makes it easy to manage the characteristic information of the calibration pattern, and reduces the room for human error when specifying characteristic information such as the grid spacing of the calibration pattern to be used during camera calibration.

In response to the above problem, the camera calibration plate of the present invention has a grid-shaped calibration pattern that is periodic in both the vertical and horizontal directions, and a two-dimensional code that expresses characteristic information of the calibration pattern. The characteristic information of the calibration pattern is information that indicates the physical characteristics of the calibration pattern, and information that is linked to the information that indicates the physical characteristics. This configuration can reduce the room for human error when calibrating the camera.

Preferably, the feature information includes information regarding the period of the calibration pattern. The period of the calibration pattern is the vertical and horizontal grid spacing of the calibration pattern. The information regarding the period of the calibration pattern may include any one of the period of the calibration pattern, the number of vertical and horizontal periods of the calibration pattern, and a combination of values from which the period or number of periods of the calibration pattern can be calculated.

Preferably, the feature information includes information regarding the length and width of the calibration pattern. The information regarding the length and width of the calibration pattern is a value or a combination of values from which the length and width of the calibration pattern can be calculated, such as the length and width of the calibration pattern, as well as the period and number of periods of the calibration pattern. This makes it easy to determine whether the entire calibration pattern is captured in an image captured by a camera for calibration.

In addition, preferably, the feature information includes information regarding the relative orientation between the two-dimensional code and the calibration pattern. The relative orientation between the two-dimensional code and the calibration pattern refers to the relationship between the vertical and horizontal orientations of the two-dimensional code and the vertical and horizontal orientations of the grid pattern.

Preferably, the calibration pattern is a dot pattern in which circles are arranged in a grid. This is because the positions of the grid points that serve as reference points for calibration can be accurately determined by determining the centers of gravity of the dots on the image captured by the camera for calibration with a small amount of calculation processing.

The two-dimensional code is preferably a matrix type two-dimensional code. The orientation of a matrix type two-dimensional code is easier to determine than other two-dimensional codes such as multi-row types, and it is also possible to calibrate the orientation from the shape of the two-dimensional code.

Preferably, the outer shape of the two-dimensional code is a rectangle. Rectangles include squares. This allows the outer shape of the two-dimensional code to be used to determine the approximate positions of the lattice points.

Preferably, the two-dimensional code is located near a corner of the calibration pattern but outside the calibration pattern. This allows the positions of the corners of the calibration pattern to be used to speed up calculations for extracting the calibration pattern on an image captured by a camera for calibration.

Alternatively, preferably, the two-dimensional code is located at a corner of the calibration pattern. "The two-dimensional code is located at a corner of the calibration pattern" means that the center of the two-dimensional code coincides with the position of a corner of the calibration pattern. This also makes it possible to use the position of the corner of the calibration pattern to speed up calculations for extracting the calibration pattern on an image captured by a camera for calibration.

Preferably, the two-dimensional code further represents information regarding its positional relationship with the corners of the calibration pattern.

When the two-dimensional code is placed at or near the corners of the calibration pattern, preferably, the two-dimensional code is placed at each of a pair of diagonal corners of the calibration pattern, or near each of a pair of diagonal corners of the calibration pattern. This makes it easier to extract the calibration pattern. More preferably, the two-dimensional code is placed at each of the four corners of the calibration pattern, or near each of the four corners of the calibration pattern. By utilizing the positions of all corners of the calibration pattern, the calculation for extracting the calibration pattern on the image captured by a camera for calibration can be further accelerated.

Furthermore, the two-dimensional code only needs to be able to express the characteristic information of the required calibration pattern, and regardless of the above, the number of two-dimensional codes may be one.

The camera calibration plate of the present invention allows the characteristic information of the calibration pattern to be read from the two-dimensional code drawn on the calibration plate itself. This eliminates the need to separately manage the characteristic information of the calibration pattern, or at least makes it easier to manage the characteristic information of the calibration pattern. Furthermore, when calibrating a camera, it is possible to reduce the room for human error in specifying characteristic information such as the grid spacing of the calibration pattern to be used.

FIG. 2 is a diagram showing a camera calibration plate according to the first embodiment; 11A and 11B are diagrams illustrating another camera calibration plate according to the first embodiment. 11A and 11B are diagrams illustrating another camera calibration plate according to the first embodiment. 11A and 11B are diagrams illustrating another camera calibration plate according to the first embodiment. FIG. 1 is a process flow diagram of a camera calibration operation. 5A to 5C are diagrams for explaining the positional relationship between a two-dimensional code and a calibration pattern on the camera calibration plate of the first embodiment. 11A and 11B are diagrams illustrating another camera calibration plate according to the first embodiment. FIG. 13 is a diagram showing a camera calibration plate according to the second embodiment. FIG. 13 is a diagram showing a camera calibration plate according to the second embodiment. FIG. 13 is a diagram showing another camera calibration plate according to the second embodiment.

The first embodiment of the camera calibration plate of the present invention will be described with reference to Figures 1 to 7.

Referring to FIG. 1, the camera calibration plate 10 of this embodiment has a calibration pattern 20 and a two-dimensional code 50 on a flat plate 18 with no irregularities. The calibration work is performed by comparing the positional relationship between the lattice points of the calibration pattern 20 between the actual calibration plate 10 and an image captured by the camera to be calibrated (hereinafter referred to as the "calibration image"). The two-dimensional code 50 represents the characteristic information of the calibration pattern 20.

The calibration plate 10 may be formed by printing the calibration pattern 20 and the two-dimensional code 50 on a film or other material with little expansion and contraction and attaching it to a flat substrate, or by printing it directly on a glass or ceramic plate with little expansion and contraction. In addition, the calibration pattern and the two-dimensional code may be displayed on a flat display as long as sufficient resolution is obtained.

A grid pattern with periodicity in both the vertical and horizontal directions is used as the calibration pattern. If the calibration image is distorted, calculations are easier if the density of grid points is uniform, and it is also easier to search for the pattern. Note that here, vertical and horizontal refer to two perpendicular directions on the calibration plate, and are convenient terms to distinguish between the two directions when the calibration pattern is not a square or when the periodicity differs in the two directions.

The calibration pattern 20 shown in FIG. 1 is a dot pattern 20 in which solid circles (hereinafter referred to as "dots") are arranged in a grid pattern. During calibration, the center of gravity of the dots 21 is taken as the position of the dots (grid points) to calculate the internal and external parameters of the camera (hereinafter, both are collectively referred to as "camera parameters"). The dot pattern 20 occupies a rectangular area with an overall horizontal length of Lh and a vertical length of Lv, with a horizontal grid spacing of Ph and a vertical grid spacing of Pv. The period of the dot pattern 20 is the grid spacing Ph and Pv. Compared to other calibration patterns described later, the dot pattern 20 is superior in that it is easy to calculate the center of gravity on the calibration image even when photographed from an oblique angle.

The number of dots 21 contained in the calibration pattern 20 is preferably 30 or more, and more preferably 100 or more. This is because the greater the number of dots 21, i.e., the number of lattice points, the greater the accuracy of the obtained camera parameters. On the other hand, the number of dots contained in the calibration pattern is preferably 1000 or less. This is because if there are too many dots, the limited resolution of the calibration image makes it impossible to accurately determine the positions of the lattice points.

As a lattice-like calibration pattern having periodicity in the vertical and horizontal directions, in addition to dot patterns, grid patterns and checker patterns can be mentioned. Referring to FIG. 2, the grid pattern 30 is a pattern in which straight lines 31 are drawn in a checkerboard pattern, and the intersections 32 of the vertical and horizontal lines become lattice points. On the outer periphery of the grid pattern, T-shaped intersections 33 become lattice points, but these are not used in the calibration calculation, and calibration is often performed using only the internal lattice points. Since the calculation procedure for extracting the lattice points on the outer periphery from the calibration image is different from that for the internal lattice points, this is mainly for the simplification of the calculation process. Referring to FIG. 3, the checker pattern 40 is a checkerboard-like pattern, and the intersections of two boundary lines 42, that is, the points 43 where the corners of the squares 41 meet, become lattice points. On the outer periphery of the checker pattern, T-shaped intersections 44 of the boundary lines become lattice points, but like the grid pattern, these are often not used in the calibration calculation. Like the dot pattern shown in FIG. 1, the grid pattern 30 and checker pattern 40 have a horizontal grid spacing of Ph and a vertical grid spacing of Pv, and occupy a rectangular area with an overall horizontal length of Lh and a vertical length of Lv.

Returning to FIG. 1, various known two-dimensional codes can be used as the two-dimensional code 50. The two-dimensional code preferably has a rectangular outer shape. This includes squares. As described below, this is because it can be used to determine the approximate position of the dots 21 and the orientation of the proofread paper. Furthermore, the two-dimensional code preferably is a matrix type two-dimensional code. This is because it is easier to determine the orientation compared to other two-dimensional codes such as multi-row types. There are no particular limitations on the type of matrix type two-dimensional code, as long as it is capable of encoding the required amount of information and can be deciphered at the resolution of the proofread image.

Two-dimensional codes generally have various information about themselves according to standards, such as model information, model number information, and error correction codes. Two-dimensional codes also generally include finder patterns, alignment patterns, timing patterns, and other patterns related to reading information, and are configured so that the orientation of the two-dimensional code itself can be determined. For example, in the case of QR Code (registered trademark), a matrix-based two-dimensional code, the vertical and horizontal orientation of the code itself can be determined by the finder patterns located near the three corners of a square.

The two-dimensional code 50 represents characteristic information of the calibration pattern 20. The characteristic information of the calibration pattern is physical characteristic information of the calibration pattern and information linked to information indicating the physical characteristics.

The physical characteristic information of the calibration pattern 20 represented by the two-dimensional code 50 includes information regarding the type of calibration pattern (e.g., whether the calibration pattern is a dot pattern or a grid pattern), the period of the calibration pattern, the number of periods of the calibration pattern, the vertical and horizontal sizes of the calibration pattern, and the relative orientation and positional relationship between the two-dimensional code and the calibration pattern.

The characteristic information of the calibration pattern represented by the two-dimensional code 50 is not limited to information that directly indicates the physical characteristics, but may be information linked to information that indicates the physical characteristics. Specifically, for example, the period and horizontal and vertical sizes of the calibration pattern are stored in advance as multiple tables in the memory unit of the camera or in an external PC to which the camera is connected. In this case, the information represented in the two-dimensional code may be a key number for selecting a specific value from the table.

Preferably, the two-dimensional code 50 represents the physical characteristic information of the calibration pattern 20. This is because the physical characteristic information of the calibration pattern can be read directly from the two-dimensional code on the calibration plate, further reducing the room for human error when calibrating the camera.

Among the physical characteristic information of the calibration pattern 20, the information on the period of the calibration pattern is, for example, the vertical and horizontal periods of the calibration pattern, the number of vertical and horizontal periods of the calibration pattern, or a combination of values from which these values can be calculated. Specifically, for example, the period can be calculated by combining the vertical and horizontal lengths of the calibration pattern and the number of vertical and horizontal periods of the calibration pattern. The period of the calibration pattern 20 is the lattice intervals Ph and Pv. The information on the period of the calibration pattern represented by the two-dimensional code 50 is preferably the vertical and horizontal periods of the calibration pattern, or information from which the vertical and horizontal periods of the calibration pattern can be calculated. Furthermore, when the vertical and horizontal lattice intervals Ph and Pv or the number of vertical and horizontal lattice points of the calibration pattern 20 are different, it is preferable to give the orientation of the calibration pattern 20 and the number of vertical and horizontal periods by the two-dimensional code.

If the period of the calibration pattern is known in advance due to some circumstances, for example if only a series of calibration patterns with the same period are used and the period is stored in the control device for the calibration work, the two-dimensional code 50 may represent only the number of vertical and horizontal periods of the calibration pattern. By reading the number of periods from the two-dimensional code, it is easy to determine whether the entire calibration pattern is shown on the calibration image.

It is not easy to automatically determine whether the entire calibration pattern is visible on the calibration image. For this reason, in the past, a proofreader would visually check the calibration image or leave a wide enough margin around the calibration pattern to determine whether an area without lattice points is visible to a certain extent. If the two-dimensional code 50 represents the number of periods of the calibration pattern, it can be easily determined whether the entire calibration pattern is visible by counting the number of lattice points in the calibration image.

The information regarding the relative orientation of the two-dimensional code 50 and the calibration pattern 20 only needs to be able to determine the length and width of the calibration pattern. Specifically, since the orientation of the two-dimensional code is known as described above, it is only necessary to know whether the length and width of the calibration pattern match the length and width of the two-dimensional code, or whether they are rotated by 90 degrees or some other angle. It is preferable to have the length and width of the two-dimensional code and the calibration pattern match, as this simplifies calculations during the calibration process.

The information on the horizontal and vertical sizes of the calibration pattern 20 is a value or a combination of values from which the horizontal and vertical lengths of the calibration pattern can be calculated, such as the vertical and horizontal lengths of the calibration pattern, or a combination of the period and number of periods of the calibration pattern. For example, the size of the calibration pattern 20 may be given by the horizontal and vertical lengths Lh and Lv of the calibration pattern, or a combination of the spacing Ph and Pv of the dots 21 in the horizontal and vertical directions and the number Nh and Nv. In FIG. 1, the horizontal and vertical lengths of the calibration pattern 20 and the number of dots are mutually convertible, with the relationship Lh=Ph×(Nh-1) and Lv=Pv×(Nv-1). The same applies when the calibration pattern is a grid pattern 30 (FIG. 2) or a checkered pattern 40 (FIG. 3). With this information, it is easy to determine whether the entire calibration pattern is captured on the calibration image.

The information regarding the relative positional relationship between the two-dimensional code 50 and the calibration pattern 20 includes whether the two-dimensional code is on the outside or inside of the calibration pattern, whether the two-dimensional code is located at a corner of the calibration pattern, and whether it is located at the left, right, top, or bottom corner. In addition, the information includes the direction and distance to one corner of the calibration pattern, preferably a corner closest to the two-dimensional code, based on the two-dimensional code. In this way, the approximate position of one corner of the calibration pattern can be known by reading the two-dimensional code.

In addition to the above, various other information related to the calibration pattern 20 and the calibration plate 10 can be encoded and represented in the two-dimensional code 50. For example, information other than the characteristic information of the calibration pattern can be encoded and represented as calibration pattern identification information such as the serial number of the calibration pattern, target camera matching information such as information for matching whether or not the camera is to be calibrated, and information related to the calibration plate, such as the serial number of the calibration plate and calibration plate identification information such as the manufacturing number.

The position of the two-dimensional code 50 is not particularly limited, but it is preferably placed outside the calibration pattern 20 or at a corner of the calibration pattern. This is to make it easier to extract the calibration pattern from the calibration image. Placing a two-dimensional code at a corner of the calibration pattern means placing a two-dimensional code in place of a corner of the calibration pattern. In FIG. 1, the two-dimensional code 50 is placed near the upper left corner of the calibration pattern 20 and outside the calibration pattern 20. In FIG. 4, the two-dimensional code 50 is placed at the upper left corner of the calibration pattern 20.

The number of two-dimensional codes 50 is not limited to one, and may be multiple. The number of two-dimensional codes is preferably four or less. If there are too many two-dimensional codes, the area of the calibration pattern that appears on the captured image will be reduced, lowering the accuracy of the calibration. When the calibration plate has multiple two-dimensional codes, the same characteristic information may be expressed in each two-dimensional code, such as information regarding the period, vertical and horizontal size, and relative orientation of the calibration pattern, or different characteristic information may be assigned to several two-dimensional codes. When used to calibrate a low-resolution camera, it is preferable to divide the information and express it in multiple two-dimensional codes, as this prevents the size of the two-dimensional code from becoming too large.

Next, the method for calibrating a camera using the camera calibration plate 10 of this embodiment will be described according to the process flow in Figure 5.

First, a calibration image is obtained. The calibration plate is photographed (S10), the photographed data is read (S11), and the angle of the calibration plate is changed (S13) until the planned number of photographs is reached (S12), and this process is repeated. The number of photographs taken is one or more, preferably three, more preferably five or more, and preferably 20 or less.

Once the planned number of calibration images have been acquired (S12), the images are selected one by one (S20), the two-dimensional code is read from the images, and the information expressed in the two-dimensional code is acquired (S21). The two-dimensional code can be read by a known method. Next, the entire calibration pattern is extracted from the calibration plate, and the approximate position of each dot in the calibration pattern is calculated (S22), and dots are searched for and extracted near the approximate positions obtained, and their centroids are calculated (S23). The extraction of the calibration pattern, and the calculation of the approximate positions and centroids of the dots can also be performed by known methods, but depending on the shape of the two-dimensional code and the information expressed by the two-dimensional code, it is possible to use this to obtain the approximate positions of the dots with less calculation effort.

When the external shape of the two-dimensional code 50 is rectangular, the external shape itself can be used as one piece of information to extract the calibration pattern 20. Referring to FIG. 6, when the horizontal and vertical directions X, Y of the two-dimensional code 50 can be determined from its external shape, if the two-dimensional code 50 expresses information about its relative orientation with the calibration pattern 20, the horizontal and vertical directions X', Y' of the calibration pattern 20 can be determined. Since the direction of the two-dimensional code 50 obtained from the calibration image contains an error, the horizontal and vertical directions X', Y' of the calibration pattern obtained thereby are also approximate, but since the range for searching the outer edge of the calibration pattern on the calibration image can be narrowed, the amount of calculation required for the search can be reduced. This is particularly advantageous in a calibration image in which the calibration plate 10 is photographed obliquely.

In addition, if the two-dimensional code 50 has information on its own length and width H, W, and information on the length and width of the calibration pattern 20, it is possible to know approximate values of the length and width of the calibration pattern 20 on the calibration image. This makes it possible to further reduce the amount of calculation required to search for the outer edge of the calibration pattern on the calibration image.

Furthermore, if the two-dimensional code 50 expresses information that can specify the position of the corner 25 from the position of the two-dimensional code 50 as information on the positional relationship between itself and the corner 25 of the calibration pattern 20, the approximate position of the outer edge of the calibration pattern 20 on the calibration image can be known based on the two-dimensional code. The information that can specify the position of the corner 25 from the position of the two-dimensional code 50 is, for example, information on the vertical and horizontal lengths of the two-dimensional code 50 and the vertical distance dY and horizontal distance dX from the two-dimensional code 50 to the corner 25 of the calibration pattern 20, more specifically, from the center of the two-dimensional code 50 to the center of gravity of the dot 21a located at the corner 25. The length of the horizontal distance dX between the two-dimensional code 50 and the corner 25 on the calibration image can be obtained based on the ratio to the width W of the two-dimensional code. The same applies to the vertical distance dY. As a result, the position of one corner 25 can be quickly determined based on the information read from the two-dimensional code, and the approximate position of the outer edge of the calibration pattern on the calibration image can be known together with the vertical and horizontal directions and size of the two-dimensional code. As a result, the amount of calculation required to search for the outer edge of the calibration pattern on the calibration image can be further reduced. Note that it is also possible to have the two-dimensional code 50 represent information capable of identifying the position of a distant corner of the calibration pattern 20, but since the direction of the two-dimensional code 50 found from the calibration image contains an error, it is preferable to have the two-dimensional code represent information capable of identifying the position of a close corner of the calibration pattern. From this perspective, it is also preferable to place the two-dimensional code at or near the corner of the calibration pattern.

When extracting the calibration pattern 20, as shown in FIG. 7, if the calibration plate 14 has two-dimensional codes 50a, 50d near a pair of diagonal corners of the calibration pattern, the position of the dot 21a in the upper left corner of the calibration pattern and the approximate values of the directions X', Y' of the upper and left sides of the outer edge of the calibration pattern can be obtained from the shape of the two-dimensional code 50a and the information it represents, and the position of the dot 21d in the lower right corner of the calibration pattern and the approximate values of the directions X", Y" of the lower and right sides of the outer edge of the calibration pattern can be obtained from the shape of the two-dimensional code 50d and the information it represents. If the diagonal corner of the calibration pattern is found using the two-dimensional codes 50a, 50d, it is known that the inside is the calibration pattern, making it easier to extract the calibration pattern and allowing accurate determination of the approximate values even if the image is tilted significantly.

After determining the dot centroids for the calibration image selected in step S20 as described above, if any unprocessed calibration images remain (S24), the next calibration image is selected (S20) and the same process is repeated. Once processing has been completed for all acquired calibration images (S24), calibration calculations are performed to determine the camera parameters (S30).

In the calibration calculation (S30), the positions of the dots 21 on the calibration image are compared with the grid spacing read from the two-dimensional code 50 to calculate the camera parameters. A publicly known method can be used to calculate the camera parameters. For example, the well-known Zhang method can be used to calculate the camera parameters. Note that the order of steps S10 to S24 in FIG. 5 may be changed as appropriate.

The effects of the calibration plates 10 to 14 of this embodiment are summarized below. When a camera is used as a visual sensor for controlling a robot, the arrangement of the camera fixed in the work area is changed according to the work content, and the focus position is adjusted according to the usage situation, etc. In addition, in the case of a stereo camera, since the principle of triangulation is used, if the angle between the two straight lines connecting the object and the two cameras becomes small, the accuracy of distance measurement decreases, so the baseline length of the two cameras is changed according to the size of the work, etc., and the relative angle of the cameras is changed. According to the camera calibration plate of this embodiment, the characteristic information of the calibration pattern required for calibration is expressed in a two-dimensional code drawn on the calibration plate itself, so that the management of the calibration pattern is easy. In addition, if the physical characteristic information of the calibration pattern is expressed in a two-dimensional code, the feature amount of the calibration pattern will not be mixed up. Therefore, the camera calibration plate of this embodiment is particularly suitable for calibrating a stereo camera used as a visual sensor of a robot.

Next, a second embodiment of the camera calibration plate of the present invention will be described with reference to Figures 8 to 10. The camera calibration plate of this embodiment is similar to that of the first embodiment in that a two-dimensional code is placed at or near each of the four corners of the calibration pattern. In the following, a description of the points common to the first embodiment will be omitted.

With reference to FIG. 8, the camera calibration plate 15 has two-dimensional codes 50a to 50d near the four corners 25a to 25d of the calibration pattern 20, respectively. The two-dimensional codes 50a to 50d are all arranged outside the calibration pattern. Features such as information on the period, vertical and horizontal size, and relative orientation of the calibration pattern may be expressed in one two-dimensional code, or may be expressed by assigning them separately to several two-dimensional codes. Preferably, each of the two-dimensional codes 50a to 50d expresses information on the positional relationship with each of the four corners 25a to 25d, which allows the position of each nearby corner to be specified from the position of the two-dimensional code. For example, the two-dimensional code 50a expresses information on the vertical and horizontal lengths of the two-dimensional code 50a and the vertical and horizontal distances from the center of the two-dimensional code 50a to the center of gravity of the dot 21a located at the corner 25a, and the same is true for the two-dimensional codes 50b to 50d. This allows the positions of the four corners 25a-25d of the calibration pattern to be quickly determined based on the information read from the two-dimensional codes 50a-50d in the approximate position calculation of the lattice points (S22) in FIG. 5.

Assuming there is no lens distortion, straight lines on the actual calibration plate are converted into straight lines in the calibration image by projection. When the calibration plate is photographed at an angle, two straight lines that are parallel to each other on the actual calibration plate will not be parallel in the calibration image. However, referring to FIG. 9, the rectangular outer edge 26 of the calibration pattern 20 is converted into a square in the calibration image, so if the positions of the four corners 25a to 25d can be determined, the approximate position of the outer edge 26 of the calibration pattern 20 in the calibration image can be known.

In this embodiment, referring to the calibration plate 16 in FIG. 10, the four two-dimensional codes 50a to 50d may be located at the four corners of the calibration pattern.

The present invention is not limited to the above-described embodiment, and various modifications are possible within the scope of the technical concept.

10-16 Camera calibration plate 18 Flat plate 20 Dot pattern (calibration pattern)
21, 21a to 21d dots (lattice points)
25, 25a to 25d: Corners of the calibration pattern 26: Outer edge of the calibration pattern on the calibration image 30: Grid pattern (calibration pattern)
31 Grid 32 Grid intersection (lattice point)
33 Lattice points on the outer periphery 40 Checker pattern (calibration pattern)
41 Grid 42 Boundary 43 Boundary intersection (grid point)
44 Lattice points on the outer periphery 50, 50a to 50d Two-dimensional code Lh Horizontal length of the calibration pattern Lv Vertical length of the calibration pattern Ph Horizontal lattice spacing Pv Vertical lattice spacing

Claims (11)

A grid-like calibration pattern having periodicity in both the vertical and horizontal directions;
a two-dimensional code representing characteristic information of the calibration pattern ;
the feature information includes information regarding a relative orientation between the two-dimensional code and the calibration pattern;
Camera calibration plate. A grid-like calibration pattern having periodicity in both the vertical and horizontal directions;
a two-dimensional code representing characteristic information of the calibration pattern ;
the two-dimensional code further expresses information regarding a positional relationship with a corner of the calibration pattern;
Camera calibration plate. the characteristic information includes information regarding a period of the calibration pattern;
3. The camera calibration plate according to claim 1 or 2 . the characteristic information includes information regarding the vertical and horizontal sizes of the calibration pattern;
The camera calibration plate according to any one of claims 1 to 3 . The calibration pattern is a dot pattern in which circles are arranged in a grid pattern.
The camera calibration plate according to any one of claims 1 to 4. The two-dimensional code is a matrix type two-dimensional code.
The camera calibration plate according to any one of claims 1 to 5. The two-dimensional code has a rectangular outer shape.
The camera calibration plate according to any one of claims 1 to 6. the two-dimensional code is located near a corner of the calibration pattern and outside the calibration pattern;
The camera calibration plate according to any one of claims 1 to 7. the two-dimensional code is located at a corner of the calibration pattern;
The camera calibration plate according to any one of claims 1 to 7. the two-dimensional codes are arranged on or near a pair of diagonal corners of the calibration pattern,
10. A camera calibration plate according to claim 8 or 9 . the two-dimensional code is disposed at or near each of the four corners of the calibration pattern;
The camera calibration plate according to any one of claims 8 to 10 .

Images