High-precision large-range pose measurement method and system based on absolute space coding calibration plate

Abstract

The invention discloses a high-precision large-range pose measurement method and system based on an absolute space coding calibration board, and relates to the technical field of vision measurement, and the method comprises the steps: determining a target mark circle center pixel coordinate, a target grating center point pixel coordinate and a target coding center point pixel coordinate of a target minimum recognition unit in the calibration board; according to the absolute phase distribution of the two-dimensional sinusoidal grating pattern of the target minimum identification unit, the pixel coordinate of the target grating center point and the pixel coordinate of the target coding center point, determining the actual physical distance of the same sight line point of the grating center point and the coding center point; and determining an initial pose through the mark circle center world coordinate of the target minimum recognition unit, the target mark circle center pixel coordinate and the camera internal reference matrix, and performing depth correction on the basis of the target coding center point pixel coordinate, the actual physical distance and the target grating center point world coordinate to output the target pose. And carrying out pose calculation based on the designed absolute type space coding calibration plate so as to realize large-range high-precision space calibration.

Description

Technical Field

[0001] The present invention relates to the field of visual measurement technology, and in particular to a high-precision, large-range posture measurement method and system based on an absolute spatial coding calibration plate. Background Art

[0002] With its advantages of low cost, non-contact nature, and flexibility, visual measurement technology has gradually become a research hotspot for improving the accuracy of pose error compensation for multi-axis equipment and robotic end-points. It is widely used in industrial fields such as object motion tracking, robotic end-point guidance, and kinematic calibration of multi-axis platforms. Visual measurement technology obtains image information from feature points on the surface of a calibration plate to measure pose, and then identifies the deviation between the actual motion model of the mechanical system and the theoretical model to complete the system's motion error compensation. Clearly, visual calibration is a crucial foundation for subsequent high-precision operations.

[0003] Traditional visual calibration methods mostly rely on the measurement of checkerboard or dot matrix calibration plates. They are highly accurate in small-scale calibration, but have obvious limitations when facing large-scale high-precision calibration. For example, the checkerboard calibration plate is sensitive to shooting angle and lighting conditions. The collected image must completely cover the entire checkerboard to provide sufficient geometric constraints to solve the pose, which limits its application in precise posture calibration under large-scale motion.

[0004] When visual measurement technology is used to calibrate motion platforms, high precision and wide-range measurement requirements are often incompatible. Large-sized calibration plates cover a wide area, but the density of feature points per unit area is reduced, resulting in low local accuracy. High-resolution, small-field-of-view measurements require reducing the calibration plate size, which reduces the measurement range. In addition, planar calibration plates lack depth information in the depth direction, and limited geometric constraints can easily lead to low accuracy in extrinsic parameter solutions. This is particularly evident when the calibration plate has a limited viewing angle or is tilted significantly. When the calibration plate's minimum coding unit is too large, tilting it will cause the coding unit to exceed the depth of field. Increasing the field of view can increase the number of image pixels, but this will increase the image processing burden. Reducing the image pixel accuracy will lead to an increase in the final kinematic calibration error.

[0005] Therefore, current visual measurement is difficult to take into account both large-scale and high-precision calibration requirements. It is urgent to provide a method that can improve the geometry and range limitations of the planar calibration plate to achieve high-precision large-scale pose measurement. Summary of the Invention

[0006] The present invention provides a high-precision, large-range pose measurement method and system based on an absolute spatial encoding calibration plate, which solves the technical problem that existing visual measurement is difficult to take into account both large-range and high-precision calibration requirements.

[0007] The first aspect of the present invention provides a high-precision, large-range pose measurement method based on an absolute spatial coding calibration plate, which involves a double-layer transparent absolute spatial coding calibration plate. The upper surface of the double-layer transparent absolute spatial coding calibration plate is arrayed with Aruco two-dimensional codes and the lower surface is arrayed with a two-dimensional sinusoidal grating pattern. Circular marking point groups are equidistant between adjacent Aruco two-dimensional codes. The Aruco two-dimensional codes that form the minimum identification unit constitute a unique coding identification matrix. The coding center point of the minimum identification unit corresponds one-to-one with the grating center point of the two-dimensional sinusoidal grating pattern. The method includes:

[0008] The target calibration plate is used to capture images to determine the pixel coordinates of the target mark center of the target minimum identification unit, the pixel coordinates of the target grating center point, and the pixel coordinates of the target code center point;

[0009] Determine the actual physical distance between the corresponding grating center point and the co-line-of-sight point of the code center point according to the absolute phase distribution of the two-dimensional sinusoidal grating pattern to which the target minimum identification unit belongs, the pixel coordinates of the target grating center point, and the pixel coordinates of the target code center point;

[0010] The world coordinates of the center of the target mark are matched by the coding recognition matrix of the target minimum recognition unit, and the pose is solved by combining the pixel coordinates of the center of the target mark and the camera intrinsic parameter matrix to determine the initial pose;

[0011] The initial posture is locally corrected for depth information based on the pixel coordinates of the target encoding center point, the actual physical distance, and the world coordinates of the grating center point of the target minimum identification unit, and the target posture is output.

[0012] Optionally, the method of using a target calibration plate to capture an image and determine the target mark circle center pixel coordinates, the target grating center point pixel coordinates, and the target code center point pixel coordinates of the target minimum identification unit includes:

[0013] After extracting the minimum target recognition unit from the image captured by the target calibration plate, perform corner recognition to determine the actual coordinates of the QR code corner points;

[0014] Based on the actual coordinates of the corner points of the two-dimensional code, a minimum mark recognition area containing a circular mark point group is circled in the target minimum recognition unit;

[0015] Searching the preset calibration plate coding table according to the coding recognition matrix of the target minimum recognition unit to determine the coordinates of the designed two-dimensional code corner points;

[0016] Based on the transformation relationship between the actual two-dimensional code corner point coordinates and the associated design two-dimensional code corner point coordinates, performing perspective transformation on the image captured by the target calibration plate to determine a transformed image;

[0017] Extracting the initial marker center pixel coordinates and the initial grating center pixel coordinates of the circular marker point group in each of the minimum marker recognition areas from the transformed image;

[0018] Using the pixel coordinates of the center of each initial mark circle to perform geometric center fitting, determine the initial encoding center point pixel coordinates of the encoding center point of the target minimum recognition unit;

[0019] The initial mark circle center pixel coordinates, the initial grating center point pixel coordinates and the initial code center point pixel coordinates are inversely transformed into the target calibration plate captured image to determine the target mark circle center pixel coordinates, the target grating center point pixel coordinates and the target code center point pixel coordinates.

[0020] Optionally, determining the actual physical distance between the corresponding grating center point and the co-line-of-sight point of the encoding center point according to the absolute phase distribution of the two-dimensional sinusoidal grating pattern to which the target minimum identification unit belongs, the pixel coordinates of the target grating center point, and the pixel coordinates of the target encoding center point includes:

[0021] Cut out a grating image of the area where the two-dimensional sinusoidal grating pattern of the target minimum recognition unit is located from the image captured by the target calibration plate;

[0022] performing a two-dimensional Fourier transform, a bandpass filter, a two-dimensional inverse Fourier transform, and a phase unwrapping on the grating image in sequence to determine an absolute phase distribution;

[0023] Determining a first absolute phase of a co-linear point associated with a corresponding encoding center point from the absolute phase distribution according to the pixel coordinates of the encoding center point;

[0024] Determining a second absolute phase of a corresponding grating center point from the absolute phase distribution according to the pixel coordinates of the target grating center point;

[0025] Determine a corresponding pixel distance using a phase difference between the first absolute phase and the second absolute phase;

[0026] An actual physical distance between the co-linear point and the grating center point is determined according to the pixel distance and the physical distance corresponding to the unit pixel distance.

[0027] Optionally, the matching of the world coordinates of the center of the target marker by the coding recognition matrix of the target minimum recognition unit, combining the pixel coordinates of the center of the target marker and the camera intrinsic parameter matrix to perform pose calculation to determine the initial pose, includes:

[0028] Searching the preset calibration plate code table based on the code recognition matrix of the target minimum recognition unit to match the world coordinates of the center of the mark circle;

[0029] Using the world coordinates of the center of the marker circle, the pixel coordinates of the center of the target marker circle, and the camera intrinsic parameter matrix, a linear transformation is performed based on pinhole imaging to determine a homography matrix;

[0030] An initial rotation matrix and an initial translation matrix are decomposed from the homography matrix as an initial pose.

[0031] Optionally, the locally correcting the depth information of the initial pose based on the pixel coordinates of the target encoding center point, the actual physical distance, and the world coordinates of the grating center point of the target minimum identification unit, and outputting the target pose, includes:

[0032] Searching the preset calibration plate code table based on the code recognition matrix of the target minimum recognition unit to match the world coordinates of the grating center point;

[0033] Performing a sum operation on the world coordinates of the grating center point and the actual physical distance to determine the world coordinates of the cosmopolitan point;

[0034] Performing a nonlinear least squares solution on the projection error objective function according to the pixel coordinates of the target encoding center point, the world coordinates of the co-linear point and the initial pose to determine the depth information;

[0035] The initial pose is updated based on the depth information to determine a target pose.

[0036] Optionally, the process of determining the image captured by the target calibration plate includes:

[0037] The calibration plate image of the double-layer transparent absolute spatial coding calibration plate is preprocessed and the target calibration plate image is output.

[0038] The second aspect of the present invention provides a high-precision, large-range pose measurement system based on an absolute spatial coding calibration plate, which relates to a double-layer transparent absolute spatial coding calibration plate. The upper surface of the double-layer transparent absolute spatial coding calibration plate is arrayed with Aruco two-dimensional codes and the lower surface is arrayed with a two-dimensional sinusoidal grating pattern. Circular marking point groups are equidistant between adjacent Aruco two-dimensional codes. The Aruco two-dimensional codes that form the minimum identification unit constitute a unique coding identification matrix. The coding center point of the minimum identification unit corresponds one-to-one with the grating center point of the two-dimensional sinusoidal grating pattern. The system includes:

[0039] A pixel coordinate extraction module is used to determine the pixel coordinates of the target mark center of the target minimum recognition unit, the pixel coordinates of the target grating center point, and the pixel coordinates of the target code center point by using the target calibration plate to capture the image;

[0040] a physical distance determination module, configured to determine the actual physical distance between the corresponding grating center point and the co-line-of-sight point of the code center point based on the absolute phase distribution of the two-dimensional sinusoidal grating pattern to which the target minimum identification unit belongs, the pixel coordinates of the target grating center point, and the pixel coordinates of the target code center point;

[0041] A pose calculation module is used to match the world coordinates of the marker circle center through the coding recognition matrix of the target minimum recognition unit, and perform pose calculation based on the pixel coordinates of the target marker circle center and the camera intrinsic parameter matrix to determine the initial pose;

[0042] A posture correction module is used to perform local depth information correction on the initial posture based on the pixel coordinates of the target encoding center point, the actual physical distance and the world coordinates of the grating center point of the target minimum identification unit, and output the target posture.

[0043] The third aspect of the present invention provides a computer device, including a memory and a processor, wherein a computer program is stored in the memory. When the computer program is executed by the processor, the processor executes the steps of the high-precision, large-range posture measurement method based on an absolute spatial encoding calibration plate as described in any one of the above items.

[0044] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed, implements a high-precision, large-range pose measurement method based on an absolute spatial encoding calibration plate as described in any one of the above items.

[0045] A fifth aspect of the present invention provides a computer program product, comprising a computer program / instruction, which, when executed by a processor, implements a high-precision, large-range pose measurement method based on an absolute spatial encoding calibration plate as described in any one of the above items.

[0046] It can be seen from the above technical solutions that the present invention has the following advantages:

[0047] The above-mentioned scheme of the present invention provides a high-precision, large-range posture measurement method based on an absolute spatial coding calibration plate, which involves a double-layer transparent absolute spatial coding calibration plate, including: using a target calibration plate to shoot an image to determine the target mark circle center pixel coordinates, the target grating center point pixel coordinates and the target coding center point pixel coordinates of the target minimum identification unit; determining the actual physical distance between the corresponding grating center point and the cosine line point of the coding center point according to the absolute phase distribution of the two-dimensional sinusoidal grating pattern to which the target minimum identification unit belongs, the target grating center point pixel coordinates and the target coding center point pixel coordinates; matching the mark circle center world coordinates through the coding identification matrix of the target minimum identification unit, combining the target mark circle center pixel coordinates and the camera intrinsic parameter matrix to perform posture solution to determine the initial posture; performing local depth information correction on the initial posture based on the target coding center point pixel coordinates, the actual physical distance and the grating center point world coordinates of the target minimum identification unit, and outputting the target posture. The above solution is based on the designed double-layer transparent absolute spatial coding calibration plate for pose solution. On the one hand, it reduces the minimum identifiable unit of the calibration plate, solves the problem that the traditional calibration method requires shooting the complete calibration plate code, and realizes a larger range of XY measurement and measurement and calibration under large angle changes. On the other hand, it increases the spatial coding dimension of the coding unit, and provides more accurate depth measurement data for calibration through the spatial distribution of calibration points, which helps to achieve large-scale, high-precision spatial calibration. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0049] Figure 1 A flowchart of the steps of a high-precision, large-range pose measurement method based on an absolute spatial encoding calibration plate provided in Example 1 of the present invention;

[0050] Figure 2 A partial view of the absolute spatial coding calibration plate provided in Example 1 of the present invention;

[0051] Figure 3 The camera imaging diagram provided by the first embodiment of the present invention;

[0052] Figure 4 Schematic diagram of the minimum recognition unit provided by the first embodiment of the present invention;

[0053] Figure 5 The two-dimensional sinusoidal grating pattern provided in the first embodiment of the present invention;

[0054] Figure 6 Schematic diagram of absolute spatial coding pose measurement provided by embodiment 1 of the present invention;

[0055] Figure 7 A flowchart of a high-precision, large-range pose measurement method based on an absolute spatial encoding calibration plate provided in the first embodiment of the present invention;

[0056] Figure 8 This is a structural block diagram of a high-precision, large-range pose measurement system based on an absolute spatial encoding calibration plate provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0057] The embodiments of the present invention provide a high-precision, large-range pose measurement method and system based on an absolute spatial encoding calibration plate, which is used to solve the technical problem that existing visual measurement is difficult to take into account both large-range and high-precision calibration requirements.

[0058] In order to make the purpose, features, and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0059] See also Figure 1 , Figure 1 A flowchart of the steps of a high-precision, large-range pose measurement method based on an absolute spatial encoding calibration plate provided in Example 1 of the present invention.

[0060] This embodiment provides a high-precision, large-range pose measurement method based on an absolute spatial coding calibration plate, involving a double-layer transparent absolute spatial coding calibration plate. The method includes steps 101 to 104.

[0061] It should be noted that this embodiment studies and designs a double-layer transparent absolute spatial coding calibration plate, such as Figures 2 to 4 As shown, the upper surface of the double-layer transparent absolute spatial coding calibration plate is arrayed with Aruco two-dimensional codes and the lower surface is arrayed with a two-dimensional sinusoidal grating pattern. Circular marking point groups are equidistant between adjacent Aruco two-dimensional codes. The Aruco two-dimensional codes that form the minimum identification unit constitute a unique coding identification matrix. The coding center point of the minimum identification unit corresponds one-to-one with the grating center point of the two-dimensional sinusoidal grating pattern. In terms of the calibration plate design concept:

[0062] 1) Design of the coding pattern and feature points on the calibration plate:

[0063] The core of this embodiment's extrinsic parameter calibration lies in the following: a designed absolute calibration plate, whose minimum identification unit carries decodable unique coded information. By partially photographing each minimum identification area, the absolute coordinates of these points on the calibration plate can be determined;

[0064] This embodiment uses Aruco coding to achieve rapid positioning and identification, which has the advantages of rapid detection, ease of use, anti-interference ability and low cost; Figure 4 As shown in the figure, every four Aruco QR codes constitute a minimum identification unit (ID0 / ID1 / ID2). If the dictionary for generating Aruco QR codes is DICT_4X4_250, there are 250 uniquely identified QR codes participating in the calibration plate pattern encoding, and every four QR code ID values constitute a 2*2 encoding identification matrix, for example , a total of 250 4 There are many combinations, and more dictionaries with different coding numbers can be selected to generate QR codes according to actual needs. In short, in the absolute coding pattern on the entire surface of the calibration plate, the minimum identification unit that meets the absolute coding template is unique. Therefore, as long as any part of the coding pattern on the calibration plate that is larger than the minimum identification unit is photographed, the absolute position of this part of the coding pattern in the original image can be determined. Therefore, the calibration plate can use the arrangement of the QR code on the upper surface to achieve absolute coding, so as to support a wide range of calibration applications.

[0065] After the identification code is designed and the absolute position is quickly acquired, the camera extrinsic parameters need to be calibrated. This embodiment uses a circular pattern to construct circular markers as calibration feature points. The center of the circular pattern is less affected by lens distortion (such as radial distortion and tangential distortion) in the image. Even if the circle is stretched into an ellipse due to distortion, its center point can still be accurately calculated through geometric fitting (such as the least squares method). The circular markers also have higher sub-pixel detection accuracy. In comparison, the traditional checkerboard corner points are easily distorted, resulting in significantly increased positioning errors.

[0066] In a specific implementation, each minimum recognition unit can carry multiple circular marker point groups. A group of circular marker point groups is set between any two adjacent Aruco QR codes. Each circular marker point group includes one or more evenly distributed circular marker points. Exemplarily, the number of circular marker point groups is four, and each circular marker point group includes four circular marker points, that is, each minimum recognition unit carries a total of 16 circular marker points, which serve as feature points for solving the camera extrinsic parameters, that is, target points.

[0067] 2) Design of the 2D sinusoidal grating pattern on the lower surface of the calibration plate:

[0068] This embodiment Figure 5 The two-dimensional sinusoidal grating pattern shown can be generated into a grayscale image using the following formula:

[0069]

[0070] Where, is a grayscale image, is the brightness DC component, is the fringe frequency in the x direction, is the pixel coordinate in the x direction, is the phase, is the fringe frequency in the y direction, is the pixel coordinate in the y direction;

[0071] A two-dimensional sinusoidal grating pattern can simultaneously acquire phase information in the x and y directions. Since the information in both directions is integrated in the same pattern, the phase data in the x and y directions can be separated without rotating or moving the grating, thereby improving measurement efficiency. By analyzing the phase distribution of the fringes in the image, a mapping relationship between phase difference and physical distance is established. The phase difference can be converted into physical distance to obtain the actual world coordinates of the target. Combined with the circular marking points on the upper surface, the three-dimensional distribution of calibration points is realized, which helps to construct more precise spatial geometric constraints, thereby improving calibration accuracy.

[0072] 3) For the overall design of the calibration plate:

[0073] In the specific implementation, in order to achieve a transparent effect, the calibration plate can be prepared by a glass plate. The thickness of the calibration plate is known when the processing is completed. On the upper surface of the calibration plate, the Aruco QR code is used for the absolute positioning of the calibration plate, and the circular marking point is used for the rapid solution of the calibration plate posture. On the lower surface of the calibration plate, a two-dimensional sinusoidal grating pattern generated based on a mathematical model is used to provide high-precision target information. The world coordinates of the center of the circular pattern of the feature points on the upper surface have been determined during processing. Therefore, the points on the upper surface can form three-dimensional observation data with the target points on the lower surface to jointly participate in the solution of the calibration plate posture. At the same time, it can also be used as a plane target to solve the calibration plate posture. The former posture solution result is preliminarily tested to see if it is correct. The center point of each group of two-dimensional sinusoidal grating patterns on the lower surface (i.e., the grating center point) corresponds to the center point of the line connecting the centers of each group of circular codes on the upper surface (i.e., the code center point). Their world coordinates have also been determined during processing. From the camera's perspective, the upper and lower planes are displayed on the same image. The ideal situation is as follows: Figure 3 shown.

[0074] Step 101: Use a target calibration plate to capture an image and determine the pixel coordinates of the target mark center of the target minimum identification unit, the pixel coordinates of the target grating center point, and the pixel coordinates of the target code center point.

[0075] The image captured by the calibration plate refers to the image obtained by capturing the double-layer transparent absolute spatial coding calibration plate.

[0076] The minimum identification unit refers to the smallest identification area that contains unique coding information that conforms to the absolute coding template.

[0077] The pixel coordinates of the marker center refer to the coordinates of the center of the circular marker point in the circular marker point group in the pixel coordinate system.

[0078] The pixel coordinates of the grating center point refer to the coordinates of the grating center point of the two-dimensional sinusoidal grating pattern in the pixel coordinate system.

[0079] The pixel coordinates of the coding center point refer to the coordinates of the center point of the upper surface containing the QR code in the area where the minimum recognition unit is located in the pixel coordinate system.

[0080] The process of determining the image captured by the target calibration plate includes:

[0081] The calibration plate image of the double-layer transparent absolute spatial coding calibration plate is preprocessed and the target calibration plate image is output.

[0082] It should be noted that after obtaining a captured image of the double-layer transparent absolute spatially encoded calibration plate by photographing it with a camera, the captured image can be preprocessed. This preprocessing can include steps such as binarization, denoising, grayscale correction, and geometric correction (to eliminate lens distortion). Finally, a captured image of the target calibration plate is output for calibration. To address the problem of traditional calibration plates requiring a complete capture, the captured image of the calibration plate can correspond to an image obtained by capturing only a portion of the calibration plate. Calibration can be performed as long as the captured image of the calibration plate contains at least one minimum identifiable unit. Therefore, after obtaining the captured image of the target calibration plate, the minimum identifiable unit is identified and extracted from the captured image of the target calibration plate. One of the extracted minimum identifiable units is selected as the minimum identifiable unit for calibration, namely the target minimum identifiable unit. Pixel coordinates of the target minimum identifiable unit's mark circle center, grating center point, and code center point are then identified, thereby determining the corresponding target mark circle center pixel coordinates, target grating center point pixel coordinates, and target code center point pixel coordinates.

[0083] In a specific implementation of this embodiment, step 101 includes the following sub-steps:

[0084] After extracting the minimum target recognition unit from the image captured by the target calibration plate, perform corner recognition to determine the actual coordinates of the QR code corner points;

[0085] Based on the actual coordinates of the QR code corner points, the minimum mark recognition area containing the circular mark point group is circled in the target minimum recognition unit;

[0086] According to the coding recognition matrix of the target minimum recognition unit, the preset calibration plate coding table is searched to determine the coordinates of the designed QR code corner points;

[0087] Based on the transformation relationship between the actual QR code corner coordinates and the associated design QR code corner coordinates, the image captured by the target calibration plate is perspective transformed to determine the transformed image;

[0088] Extracting the initial marker center pixel coordinates of the circular marker point group and the initial grating center pixel coordinates of each minimum marker recognition area from the transformed image;

[0089] The pixel coordinates of the center of each initial mark circle are used to perform geometric center fitting to determine the initial encoding center point pixel coordinates of the encoding center point of the target minimum recognition unit;

[0090] The pixel coordinates of the initial mark circle center, the pixel coordinates of the initial grating center point and the pixel coordinates of the initial encoding center point are inversely transformed into the image captured by the target calibration plate to determine the pixel coordinates of the target mark circle center, the pixel coordinates of the target grating center point and the pixel coordinates of the target encoding center point.

[0091] The actual QR code corner coordinates refer to the coordinates of the corners of the Aruco QR code in the image captured by the actual calibration plate.

[0092] The minimum marker recognition area refers to the minimum area that can recognize the coordinates of a circular marker point group.

[0093] The coding identification matrix refers to a matrix composed of the two-dimensional code ID values of multiple minimum identification units that constitute the minimum identification unit.

[0094] The calibration plate code table refers to a table containing the Aruco QR code ID value distribution of the entire double-layer transparent absolute spatial coding calibration plate.

[0095] The designed QR code corner point coordinates refer to the coordinates of the corner points of the Aruco QR code that have been determined according to design drawings and other methods during the processing of the double-layer transparent absolute spatial coding calibration plate.

[0096] The transformed image refers to the image obtained after perspective transformation.

[0097] It should be noted that after extracting and determining the target minimum recognition unit from the image captured by the target calibration plate, the target minimum recognition unit is subjected to corner point recognition. Since the minimum recognition unit is surrounded by multiple Aruco QR codes, the minimum recognition unit is formed with the Aruco QR code to form the corner point, thereby determining the actual QR code corner point coordinates, and thereby circle the minimum recognition unit region (ROI) used for solving the posture, thereby extracting the pixel coordinates of the center of the mark of the entire minimum recognition unit in the image captured by the target calibration plate with each minimum recognition unit region as the range. For example, Figure 3As shown, the minimum recognition unit area can be enclosed by two Aruco two-dimensional codes as the boundary, as long as it is convenient to extract the pixel coordinates of the center of the mark circle. Preferably, the minimum recognition unit area does not include the area where the two-dimensional sinusoidal grating pattern is located to avoid affecting the circular mark point extraction effect;

[0098] In order to eliminate the distortion caused by the shooting angle, the image captured by the target calibration plate is projected onto the design plane of the calibration plate using perspective transformation to regularize the image and simplify the feature extraction process. In the perspective transformation process, the coding identification matrix composed of the Aruco QR code ID value on the upper surface of the target minimum identification unit is identified and searched in the calibration plate coding table of the absolute calibration plate to find the row and column position of the coding identification matrix in the calibration plate coding table. Combined with the calibration plate processing design, the corresponding design QR code corner point coordinates can be determined. Based on the one-to-one correspondence between the actual QR code corner point coordinates and the design QR code corner point coordinates, a perspective transformation matrix can be established. Therefore, the perspective transformation of the target calibration plate image is performed based on the perspective transformation matrix to obtain a transformed image.

[0099] The pixel coordinates of the center of each circular marker group and the pixel coordinates of the grating center point of the two-dimensional sinusoidal grating pattern are extracted from the transformed image to obtain the initial marker center pixel coordinates and the initial grating center pixel coordinates. The pixel coordinates of the center point surrounded by each circular marker point are then obtained by connecting the initial marker center pixel coordinates of the circular marker points to determine the pixel coordinates of the initial encoding center point. The extracted feature points are restored to the pixel coordinates of the original image through inverse transformation. The obtained target pixel coordinates can provide high-precision target information.

[0100] Step 102: Determine the actual physical distance between the corresponding grating center point and the co-line-of-sight point of the code center point based on the absolute phase distribution of the two-dimensional sinusoidal grating pattern to which the target minimum identification unit belongs, the pixel coordinates of the target grating center point, and the pixel coordinates of the target code center point.

[0101] Absolute phase distribution refers to the phase data that continuously and uniquely reflects the spatial distribution of grating fringes. For details, please refer to the existing technology for understanding.

[0102] The same line of sight means that in the pinhole imaging camera model, if two three-dimensional points are located on the same camera ray, then their projections on the image will fall on the same pixel point, such as Figure 6 As shown in the figure, point C is the “same-line point” of point A (encoding center point) on the camera’s line of sight.

[0103] The actual physical distance refers to the actual distance between two points in the double-layer transparent absolute spatial coding calibration plate.

[0104] It should be noted that this embodiment extracts fringe phase information from the two-dimensional sinusoidal grating pattern of the target minimum identification unit in the image captured by the target calibration plate to determine the absolute phase distribution. The absolute phase distribution is then combined with the calibration plate, and the two-dimensional phase data is converted into the actual physical distance between the grating center point of the target minimum identification unit and the coaxial line point of the encoding center point through analytical geometry methods, so as to facilitate the subsequent determination of the world coordinates.

[0105] In a specific implementation of this embodiment, step 102 includes the following sub-steps:

[0106] Cut out a grating image of the area where the two-dimensional sinusoidal grating pattern of the target minimum recognition unit is located from the image captured by the target calibration plate;

[0107] The grating image is sequentially subjected to two-dimensional Fourier transform, bandpass filtering, two-dimensional inverse Fourier transform and phase unwrapping to determine the absolute phase distribution;

[0108] Determining the first absolute phase of the co-linear point associated with the corresponding encoding center point from the absolute phase distribution according to the pixel coordinates of the encoding center point;

[0109] Determining the second absolute phase of the corresponding grating center point from the absolute phase distribution according to the pixel coordinates of the target grating center point;

[0110] Determine the corresponding pixel distance using the phase difference between the first absolute phase and the second absolute phase;

[0111] According to the physical distance corresponding to the pixel distance and the unit pixel distance, the actual physical distance between the line of sight point and the center point of the grating is determined.

[0112] A grating image refers to an image containing an area with a two-dimensional sinusoidal grating pattern.

[0113] Pixel distance refers to the distance between two pixels in an image.

[0114] It should be noted that, first, the grating image of the area where the two-dimensional sinusoidal grating pattern of the target minimum recognition unit is located is cropped from the image captured by the target calibration plate. Since the actual image is obtained by shooting with an industrial camera, the background and modulation amplitude may be introduced. The following model is established:

[0115]

[0116] Where, is the pixel coordinate, is the background light (low-frequency component), and are the local modulation amplitudes of the stripes in the x and y directions, and is the x and y direction phase component to be extracted (including actual measurement information);

[0117] The fringe term can be written as:

[0118]

[0119]

[0120] Where, is the phase angle, is an imaginary unit;

[0121] Then perform a two-dimensional Fourier transform, defined as:

[0122]

[0123] Where, is a frequency variable, is the spectrum; the inverse transform is:

[0124]

[0125] According to the above model, after Fourier transform, the spectrum It will include:

[0126] 1) DC component: from , mainly concentrated in nearby;

[0127] 2) Carrier component in the x direction: one at a positive frequency correspond ; one at negative frequency corresponding to its conjugate component;

[0128] 3) Y-direction carrier component: one in correspond One in Corresponding to its conjugate component;

[0129] The actual image will be scaled and distorted during printing and shooting, resulting in the ideal setting The actual frequency may not be completely consistent with that in the image; Perform a two-dimensional Fourier transform to obtain After completing the calculation of spectrum extraction, use the spectrum amplitude ,exist Find the local maximum (peak) on the plane; ideally, the peak should be located close to:

[0130]

[0131] Then, for a certain direction (for example, the x direction), define an x-direction bandpass filter , which satisfies in the frequency domain:

[0132]

[0133] Where, For the filter radius, select a suitable value to surround the entire main peak but exclude other interference components; similarly, for the y direction, define a y-direction bandpass filter:

[0134]

[0135] After using the filter, the filtered spectrum is obtained, including the x-direction filtered spectrum and y-direction filtered spectrum :

[0136]

[0137] In this way, other frequency components are filtered out and only the carrier part in the corresponding direction is retained. The filtered positive first-order spectrum is calculated in the frequency domain in the x and y directions from 、 After translating to the origin, perform a two-dimensional inverse Fourier transform and get:

[0138]

[0139] Where, is the complex fringe signal in the x direction, is the two-dimensional inverse Fourier transform, is the complex fringe signal in the y direction;

[0140] From Equation 11, we get the phase:

[0141]

[0142] Where, is the phase of the fringe in the x direction, is the phase of the fringe in the y direction, is the imaginary part, is the real part;

[0143] Since the phase value obtained is There is a jump problem in the phase, and phase unwrapping is required to expand the phase to obtain a continuous phase function, that is, the absolute phase distribution;

[0144] Calibrate the scale factor according to the actual size of the printed or photographed pattern and , the phase difference can be converted into physical distance; for the same-line-of-sight point, the pixel coordinates of point C are the same as those of point A in the pinhole imaging model. Therefore, the first absolute phase of the corresponding same-line-of-sight point (point C) is determined from the absolute phase distribution according to the pixel coordinates of the encoding center point of point A, and the first absolute phase of the corresponding same-line-of-sight point (point C) is determined according to the pixel coordinates of the target grating center point. The second absolute phase corresponding to the grating center point (point B) is determined from the absolute phase distribution. The phase difference is constructed based on the first absolute phase and the second absolute phase. The offset is calculated based on the phase difference to find the actual physical distance from point C to the grating center point (point B). The process can be referred to the following formula:

[0145]

[0146]

[0147] Where, , that is, the phase difference in the x direction relative to the center point of the grating; , that is, the phase difference in the y direction relative to the center point of the grating, is the x-axis pixel distance relative to the center of the grating, is the pixel distance in the y direction relative to the center point of the grating; and is the physical distance corresponding to the unit pixel distance in the x and y directions, and That is, the actual physical distance from a pixel point in the x-direction and y-direction to the center point of the grating.

[0148] Step 103: Match the world coordinates of the marker circle center through the encoding recognition matrix of the target minimum recognition unit, and perform pose calculation based on the pixel coordinates of the target marker circle center and the camera intrinsic parameter matrix to determine the initial pose.

[0149] The world coordinates of the center of the circle refer to the coordinates of the circular marker point in the calibration plate coordinate system (world coordinate system).

[0150] It should be noted that in this embodiment, the world coordinates of the corresponding feature points on the double-layer transparent absolute spatial coding calibration plate have been determined during processing. Therefore, through the coding recognition matrix of the target minimum identification unit, the world coordinates of the center of the circular mark point of the target minimum identification unit in the calibration plate coordinate system can be determined from the calibration plate, that is, the world coordinates of the mark center. According to the pinhole imaging principle, the world coordinates of the mark center, the pixel coordinates of the target mark center and the camera intrinsic parameter matrix are used to perform preliminary pose solution to determine the initial pose.

[0151] In a specific implementation of this embodiment, step 103 includes the following sub-steps:

[0152] The coding recognition matrix based on the target minimum recognition unit is searched in the preset calibration plate coding table to match the world coordinates of the center of the mark circle;

[0153] The world coordinates of the marker circle center, the pixel coordinates of the target marker circle center and the camera intrinsic parameter matrix are used to perform linear transformation based on pinhole imaging to determine the homography matrix.

[0154] Decompose the initial rotation matrix and initial translation matrix from the homography matrix as the initial pose.

[0155] The homography matrix refers to the matrix that represents the mapping relationship between the calibration plate coordinate system and the pixel coordinate system.

[0156] It should be noted that the coding recognition matrix of the target minimum identification unit is searched in the coding table of the absolute calibration plate. After determining the row and column position of the coding recognition matrix in the coding table, the world coordinates of the center of the mark of the target minimum identification unit can be calculated based on the actual physical spacing between each row and column.

[0157] The known pinhole imaging model is:

[0158]

[0159] Where, is the camera intrinsic parameter matrix (known), is the three-dimensional point of the calibration plate coordinate system, is the transpose, is the pixel coordinate, is the scale factor, is the rotation matrix in the camera coordinate system, is the translation vector in the camera coordinate system; rotation matrix is a column vector , is the rotation matrix column vector, is the matrix column index;

[0160] The calibration plate coordinate system is established on the upper surface, and the circular marking points corresponding to the upper surface satisfy:

[0161]

[0162] Where, is the x-direction coordinate of the calibration plate coordinate system on the surface of the calibration plate, is the y-direction coordinate of the calibration plate coordinate system on the surface of the calibration plate; at this time, the projection equation degenerates into:

[0163]

[0164] Where, is the upper surface scale factor;

[0165] Introducing plane points in homogeneous coordinates , it can be written as:

[0166]

[0167] Define the homography matrix , then the points on the upper surface satisfy:

[0168]

[0169] Then, the coordinate correspondence of multiple circular marking points of the target minimum recognition unit on the surface of the calibration plate is used , can be obtained by using the direct linear transformation (DLT) method ,set up (Each for The vector of the homography matrix column vector), then:

[0170]

[0171]

[0172] Where, is the intermediate variable of the first column vector of the unnormalized rotation matrix in the camera coordinate system, is the intermediate variable of the second column vector of the unnormalized rotation matrix in the camera coordinate system, is the intermediate variable of the unnormalized translation vector in the camera coordinate system; there is a global scale factor So that:

[0173]

[0174] make , according to the orthogonality requirement:

[0175]

[0176] Also required:

[0177]

[0178] At this point, the first two columns of the rotation matrix are obtained from the upper surface With translation vector (in , is the translation vector The translation vectors in the x, y and z directions are obtained from this, but at this time, since all points are located at superior, (and with There is uncertainty in the depth information (related scale), so local correction of the depth information is required.

[0179] Step 104 : Perform local depth information correction on the initial pose based on the pixel coordinates of the target encoding center point, the actual physical distance, and the world coordinates of the grating center point of the target minimum identification unit, and output the target pose.

[0180] Step 104 includes the following sub-steps:

[0181] The coding recognition matrix based on the target minimum recognition unit is searched in the preset calibration plate coding table to match the world coordinates of the grating center point;

[0182] The world coordinates of the grating center point and the actual physical distance are used to perform a sum operation to determine the world coordinates of the same-line-of-sight point;

[0183] The nonlinear least squares method is used to solve the projection error objective function based on the pixel coordinates of the target encoding center point, the world coordinates of the same-line point, and the initial pose to determine the depth information.

[0184] The initial pose is updated based on the depth information to determine the target pose.

[0185] The projection error objective function refers to a function that relies on depth information to construct the projection error.

[0186] Depth information refers to information that can be used to correct the posture from the perspective of depth.

[0187] It should be noted that the line of sight point (point C) of the encoding center point of the target minimum recognition unit corresponds to the lower surface of the calibration plate, so the projection of point C on the lower surface on the image satisfies:

[0188]

[0189] in, are the pixel coordinates of the point on the same line of sight, is the coordinate of the same-line point on the lower surface of the calibration plate coordinate system, that is, the world coordinate of the same-line point. is the thickness of the calibration plate; similar to the method of determining the world coordinates of the center of the mark circle, the world coordinates of the grating center point (point B) are obtained by searching the preset calibration plate code table based on the coding recognition matrix of the target minimum recognition unit. The previous step has extracted the actual physical distance of point C relative to point B through phase information. The world coordinates of the grating center point and the actual physical distance are correspondingly summed to determine the world coordinates of the cosmopolitan point; since point C is the cosmopolitan point of point A, according to the image projection relationship, the pixel coordinates of the two are the same, so the pixel coordinates of the target coding center point are the pixel coordinates of the cosmopolitan point;

[0190] Usually, due to the decomposition 、 and When the scale is selected, it can be considered that the upper surface has been determined and However, since points on the plane cannot provide depth information, and There is ambiguity in the “absolute scale” of , so two unknown parameters can be introduced: :Used to correct The scale of , theoretically, if the decomposition is correct, we should have , : The offset of the translation vector in the z direction (from the upper surface to the camera origin); therefore, the projection equation of the lower surface can be written as:

[0191]

[0192] Where, is the lower surface scale factor; let:

[0193]

[0194] Then we have:

[0195]

[0196] in, is the depth correction term, the plane transformation term Only For the unknown, The same is the unknown parameter, and then use the known image pixel coordinates and the internal parameter matrix Given constraints, solve and ;remember The form is:

[0197]

[0198]

[0199] Where, is the equivalent focal length in the x direction, is the equivalent focal length in the y direction, is the x-axis pixel coordinate of the principal point, is the pixel coordinate in the y direction of the principal point, 、 and is the decomposition of the plane transformation term along the x, y and z directions, 、 and is the decomposition of the depth correction term along the x, y and z directions; With known quantity About, and Contains unknown amount ;

[0200]

[0201] The homogeneous scale is obtained from formula 31:

[0202]

[0203] From Formula 28, Formula 31, and Formula 32, we can get:

[0204]

[0205] Formula 33 is about and ( ) nonlinear constraints, defining the projection error (for and Component) can be used to construct the projection error objective function Use nonlinear least squares (Levenberg-Marquardt method) to find Minimum parameters and , The weight coefficient for balancing the projection error and the unit orthogonality constraint:

[0206]

[0207] After the above steps, the depth information is determined and Finally, the target rotation matrix and target translation vector in the target pose of the absolute spatial encoding calibration plate can be obtained as follows:

[0208]

[0209] Among them, the third column vector of the target rotation matrix .

[0210] For better explanation, refer to Figure 7 , shows the overall framework diagram of the first embodiment of the present invention:

[0211] 1) Information coding: using multiple ARUCO codes to generate a unique coding identification matrix;

[0212] 2) Calibration plate design: Design a double-layer transparent absolute spatial coding calibration plate based on multiple coding identification matrices and two-dimensional sinusoidal grating patterns;

[0213] 3) Image acquisition: Use the camera to capture the double-layer calibration plate to obtain the image of the calibration plate;

[0214] 4) Preprocessing: Perform threshold processing such as binarization on the image captured by the calibration plate;

[0215] 5) Feature extraction: Calculate the coordinates of the smallest recognition unit, including the pixel coordinates of the center of the marking circle, the pixel coordinates of the center point of the grating, and the pixel coordinates of the center point of the encoding;

[0216] 6) Information decoding: Extract the phase information of the two-dimensional pattern of the smallest recognition unit and determine the actual physical distance;

[0217] 7) Posture calculation: The pose is calculated by finding the absolute position of the smallest recognition unit on the calibration board based on the encoding recognition matrix of the smallest recognition unit.

[0218] It should be pointed out that here only the general process of the high-precision large-range posture measurement method based on the absolute spatial encoding calibration plate is briefly described. The specific implementation process of each step can be understood by referring to the relevant content in the aforementioned embodiments. It will not be elaborated here. It can be understood that the present invention does not limit this.

[0219] In an embodiment of the present invention, in a scheme for posture solution based on a designed double-layer transparent absolute spatial encoding calibration plate: 1) the minimum identifiable unit of the calibration plate is reduced, and only local unit coding information (minimum identifiable unit) is required to determine the world coordinate system of the calibration plate, which solves the problem that the traditional calibration method requires shooting the complete calibration plate coding, and realizes a larger range of XY measurement and measurement and calibration under large angle changes; 2) the spatial coding dimension of the coding unit is increased, and more accurate Z-axis measurement data is provided for calibration through the spatial distribution of calibration points; 3) a high-precision posture solution method based on an absolute calibration plate is given; overall, the designed calibration plate breaks through the limitations of traditional calibration plates, realizes large-scale, high-precision spatial calibration, and has a simple structure and is easy to implement, providing an effective tool for precise kinematic modeling and error compensation of multi-axis platforms.

[0220] See also Figure 8 , Figure 8 This is a structural block diagram of a high-precision, large-range pose measurement system based on an absolute spatial encoding calibration plate provided in Example 2 of the present invention.

[0221] This embodiment provides a high-precision, large-range pose measurement system based on an absolute spatial coding calibration plate. The system involves a double-layer transparent absolute spatial coding calibration plate. The upper surface of the double-layer transparent absolute spatial coding calibration plate is arrayed with Aruco two-dimensional codes, and the lower surface is arrayed with a two-dimensional sinusoidal grating pattern. Circular marking point groups are equidistant between adjacent Aruco two-dimensional codes. The Aruco two-dimensional codes that form the minimum identification unit constitute a unique coding identification matrix. The coding center point of the minimum identification unit corresponds one-to-one to the grating center point of the two-dimensional sinusoidal grating pattern. The system includes:

[0222] The pixel coordinate extraction module 801 is used to determine the pixel coordinates of the target mark center of the target minimum identification unit, the pixel coordinates of the target grating center point, and the pixel coordinates of the target code center point by using the target calibration plate to capture the image;

[0223] A physical distance determination module 802 is configured to determine the actual physical distance between the corresponding grating center point and the co-linear line point of the code center point based on the absolute phase distribution of the two-dimensional sinusoidal grating pattern to which the target minimum identification unit belongs, the pixel coordinates of the target grating center point, and the pixel coordinates of the target code center point;

[0224] The pose calculation module 803 is used to match the world coordinates of the marker circle center through the encoding recognition matrix of the target minimum recognition unit, and perform pose calculation based on the pixel coordinates of the target marker circle center and the camera intrinsic parameter matrix to determine the initial pose;

[0225] The posture correction module 804 is used to perform local depth information correction on the initial posture based on the pixel coordinates of the target encoding center point, the actual physical distance and the world coordinates of the grating center point of the target minimum identification unit, and output the target posture.

[0226] Furthermore, the pixel coordinate extraction module 801 is specifically used to:

[0227] After extracting the minimum target recognition unit from the image captured by the target calibration plate, perform corner recognition to determine the actual coordinates of the QR code corner points;

[0228] Based on the actual coordinates of the QR code corner points, the minimum mark recognition area containing the circular mark point group is circled in the target minimum recognition unit;

[0229] According to the coding recognition matrix of the target minimum recognition unit, the preset calibration plate coding table is searched to determine the coordinates of the designed QR code corner points;

[0230] Based on the transformation relationship between the actual QR code corner coordinates and the associated design QR code corner coordinates, the image captured by the target calibration plate is perspective transformed to determine the transformed image;

[0231] Extracting the initial marker center pixel coordinates of the circular marker point group and the initial grating center pixel coordinates of each minimum marker recognition area from the transformed image;

[0232] The pixel coordinates of the center of each initial mark circle are used to perform geometric center fitting to determine the initial encoding center point pixel coordinates of the encoding center point of the target minimum recognition unit;

[0233] The pixel coordinates of the initial mark circle center, the pixel coordinates of the initial grating center point and the pixel coordinates of the initial encoding center point are inversely transformed into the image captured by the target calibration plate to determine the pixel coordinates of the target mark circle center, the pixel coordinates of the target grating center point and the pixel coordinates of the target encoding center point.

[0234] Furthermore, the physical distance determination module 802 is specifically configured to:

[0235] Cut out a grating image of the area where the two-dimensional sinusoidal grating pattern of the target minimum recognition unit is located from the image captured by the target calibration plate;

[0236] The grating image is sequentially subjected to two-dimensional Fourier transform, bandpass filtering, two-dimensional inverse Fourier transform and phase unwrapping to determine the absolute phase distribution;

[0237] Determining the first absolute phase of the co-linear point associated with the corresponding encoding center point from the absolute phase distribution according to the pixel coordinates of the encoding center point;

[0238] Determining the second absolute phase of the corresponding grating center point from the absolute phase distribution according to the pixel coordinates of the target grating center point;

[0239] Determine the corresponding pixel distance using the phase difference between the first absolute phase and the second absolute phase;

[0240] According to the physical distance corresponding to the pixel distance and the unit pixel distance, the actual physical distance between the line of sight point and the center point of the grating is determined.

[0241] Furthermore, the posture solving module 803 is specifically used for:

[0242] The coding recognition matrix based on the target minimum recognition unit is searched in the preset calibration plate coding table to match the world coordinates of the center of the mark circle;

[0243] The world coordinates of the marker circle center, the pixel coordinates of the target marker circle center and the camera intrinsic parameter matrix are used to perform linear transformation based on pinhole imaging to determine the homography matrix.

[0244] Decompose the initial rotation matrix and initial translation matrix from the homography matrix as the initial pose.

[0245] Furthermore, the posture solving module 803 is specifically used for:

[0246] The coding recognition matrix based on the target minimum recognition unit is searched in the preset calibration plate coding table to match the world coordinates of the grating center point;

[0247] The world coordinates of the grating center point and the actual physical distance are used to perform a sum operation to determine the world coordinates of the same-line-of-sight point;

[0248] The nonlinear least squares method is used to solve the projection error objective function based on the pixel coordinates of the target encoding center point, the world coordinates of the same-line point, and the initial pose to determine the depth information.

[0249] The initial pose is updated based on the depth information to determine the target pose.

[0250] Furthermore, a pre-processing module is included for:

[0251] The calibration plate image of the double-layer transparent absolute spatial coding calibration plate is preprocessed and the target calibration plate image is output.

[0252] An embodiment of the present invention also provides a computer device, including a memory and a processor, wherein a computer program is stored in the memory; when the computer program is executed by the processor, the processor executes the steps of a high-precision, large-range posture measurement method based on an absolute spatial encoding calibration plate as in any of the above embodiments.

[0253] An embodiment of the present invention also provides a computer-readable storage medium having a computer program / instruction stored thereon. When the computer program / instruction is executed by a processor, the steps of a high-precision, large-range posture measurement method based on an absolute spatial coding calibration plate as in any of the above embodiments are implemented.

[0254] An embodiment of the present invention also provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of a high-precision, large-range pose measurement method based on an absolute spatial encoding calibration plate as in any of the above embodiments.

[0255] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems and modules can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0256] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0257] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0258] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0259] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0260] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A high-precision, large-range pose measurement method based on an absolute spatial encoding calibration plate, characterized in that: The invention relates to a double-layer transparent absolute spatial coding calibration plate, wherein an Aruco two-dimensional code is arrayed on the upper surface of the double-layer transparent absolute spatial coding calibration plate and a two-dimensional sinusoidal grating pattern is arrayed on the lower surface. Circular marking point groups are equidistantly arranged between adjacent Aruco two-dimensional codes. The Aruco two-dimensional codes forming a minimum identification unit constitute a unique coding identification matrix, and the coding center point of the minimum identification unit corresponds one-to-one to the grating center point of the two-dimensional sinusoidal grating pattern. The method comprises: The target calibration plate is used to capture images to determine the pixel coordinates of the target mark center of the target minimum identification unit, the pixel coordinates of the target grating center point, and the pixel coordinates of the target code center point; Determine the actual physical distance between the corresponding grating center point and the co-line-of-sight point of the code center point according to the absolute phase distribution of the two-dimensional sinusoidal grating pattern to which the target minimum identification unit belongs, the pixel coordinates of the target grating center point, and the pixel coordinates of the target code center point; The world coordinates of the center of the target mark are matched by the coding recognition matrix of the target minimum recognition unit, and the pose is solved by combining the pixel coordinates of the center of the target mark and the camera intrinsic parameter matrix to determine the initial pose; The initial posture is locally corrected for depth information based on the pixel coordinates of the target encoding center point, the actual physical distance, and the world coordinates of the grating center point of the target minimum identification unit, and the target posture is output.

2. The high-precision, large-range pose measurement method based on an absolute spatial encoding calibration plate according to claim 1, characterized in that: The method of using a target calibration plate to capture an image and determine the target mark circle center pixel coordinates, the target grating center point pixel coordinates, and the target code center point pixel coordinates of the target minimum identification unit includes: After extracting the minimum target recognition unit from the image captured by the target calibration plate, perform corner recognition to determine the actual coordinates of the QR code corner points; Based on the actual coordinates of the corner points of the two-dimensional code, a minimum mark recognition area containing a circular mark point group is circled in the target minimum recognition unit; Searching the preset calibration plate coding table according to the coding recognition matrix of the target minimum recognition unit to determine the coordinates of the designed two-dimensional code corner points; Based on the transformation relationship between the actual two-dimensional code corner point coordinates and the associated design two-dimensional code corner point coordinates, performing perspective transformation on the image captured by the target calibration plate to determine a transformed image; Extracting the initial marker center pixel coordinates and the initial grating center pixel coordinates of the circular marker point group in each of the minimum marker recognition areas from the transformed image; Using the pixel coordinates of the center of each initial mark circle to perform geometric center fitting, determine the initial encoding center point pixel coordinates of the encoding center point of the target minimum recognition unit; The initial mark circle center pixel coordinates, the initial grating center point pixel coordinates and the initial code center point pixel coordinates are inversely transformed into the target calibration plate captured image to determine the target mark circle center pixel coordinates, the target grating center point pixel coordinates and the target code center point pixel coordinates.

3. The high-precision, large-range pose measurement method based on an absolute spatial encoding calibration plate according to claim 1, characterized in that: The determining, based on the absolute phase distribution of the two-dimensional sinusoidal grating pattern to which the target minimum identification unit belongs, the pixel coordinates of the target grating center point, and the pixel coordinates of the target code center point, the actual physical distance between the corresponding grating center point and the co-line-of-sight point of the code center point includes: Cut out a grating image of the area where the two-dimensional sinusoidal grating pattern of the target minimum recognition unit is located from the image captured by the target calibration plate; performing a two-dimensional Fourier transform, a bandpass filter, a two-dimensional inverse Fourier transform, and a phase unwrapping on the grating image in sequence to determine an absolute phase distribution; Determining a first absolute phase of a co-linear point associated with a corresponding encoding center point from the absolute phase distribution according to the pixel coordinates of the encoding center point; Determining a second absolute phase of a corresponding grating center point from the absolute phase distribution according to the pixel coordinates of the target grating center point; Determine a corresponding pixel distance using a phase difference between the first absolute phase and the second absolute phase; An actual physical distance between the co-linear point and the grating center point is determined according to the pixel distance and the physical distance corresponding to the unit pixel distance.

4. The high-precision, large-range pose measurement method based on an absolute spatial encoding calibration plate according to claim 1, characterized in that: The method of matching the world coordinates of the center of the target mark by the encoding recognition matrix of the target minimum recognition unit and performing pose calculation in combination with the pixel coordinates of the center of the target mark and the camera intrinsic parameter matrix to determine the initial pose includes: Searching the preset calibration plate code table based on the code recognition matrix of the target minimum recognition unit to match the world coordinates of the center of the mark circle; Using the world coordinates of the center of the marker circle, the pixel coordinates of the center of the target marker circle, and the camera intrinsic parameter matrix, a linear transformation is performed based on pinhole imaging to determine a homography matrix; An initial rotation matrix and an initial translation matrix are decomposed from the homography matrix as an initial pose.

5. The high-precision, large-range pose measurement method based on an absolute spatial encoding calibration plate according to claim 1, characterized in that: The locally correcting the depth information of the initial pose based on the pixel coordinates of the target encoding center point, the actual physical distance, and the world coordinates of the grating center point of the target minimum identification unit, and outputting the target pose, includes: Searching the preset calibration plate code table based on the code recognition matrix of the target minimum recognition unit to match the world coordinates of the grating center point; Performing a sum operation on the world coordinates of the grating center point and the actual physical distance to determine the world coordinates of the cosmopolitan point; Performing a nonlinear least squares solution on the projection error objective function according to the pixel coordinates of the target encoding center point, the world coordinates of the co-linear point and the initial pose to determine the depth information; The initial pose is updated based on the depth information to determine a target pose.

6. The high-precision, large-range pose measurement method based on an absolute spatial encoding calibration plate according to claim 1, characterized in that: The process of determining the image captured by the target calibration plate includes: The calibration plate image of the double-layer transparent absolute spatial coding calibration plate is preprocessed and the target calibration plate image is output.

7. A high-precision, large-range pose measurement system based on an absolute spatial encoding calibration plate, characterized in that: The invention relates to a double-layer transparent absolute spatial coding calibration plate. The upper surface of the double-layer transparent absolute spatial coding calibration plate is arrayed with Aruco two-dimensional codes and the lower surface is arrayed with a two-dimensional sinusoidal grating pattern. Circular marking point groups are equidistantly arranged between adjacent Aruco two-dimensional codes. The Aruco two-dimensional codes that form the minimum identification unit constitute a unique coding identification matrix. The coding center point of the minimum identification unit corresponds one-to-one with the grating center point of the two-dimensional sinusoidal grating pattern. The system includes: A pixel coordinate extraction module is used to determine the pixel coordinates of the target mark center of the target minimum recognition unit, the pixel coordinates of the target grating center point, and the pixel coordinates of the target code center point by using the target calibration plate to capture the image; a physical distance determination module, configured to determine the actual physical distance between the corresponding grating center point and the co-line-of-sight point of the code center point based on the absolute phase distribution of the two-dimensional sinusoidal grating pattern to which the target minimum identification unit belongs, the pixel coordinates of the target grating center point, and the pixel coordinates of the target code center point; A pose calculation module is used to match the world coordinates of the marker circle center through the coding recognition matrix of the target minimum recognition unit, and perform pose calculation based on the pixel coordinates of the target marker circle center and the camera intrinsic parameter matrix to determine the initial pose; A posture correction module is used to perform local depth information correction on the initial posture based on the pixel coordinates of the target encoding center point, the actual physical distance and the world coordinates of the grating center point of the target minimum identification unit, and output the target posture.

8. A computer device, characterized in that: It includes a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the high-precision large-range posture measurement method based on the absolute spatial encoding calibration plate as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by the processor, the steps of the high-precision, large-range pose measurement method based on the absolute spatial encoding calibration plate as described in any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by the processor, the steps of the high-precision, large-range pose measurement method based on the absolute spatial encoding calibration plate as described in any one of claims 1 to 6 are implemented.

Images