A chessboard design method for binocular vision detection of a laser deposition part

Abstract

The application provides a chessboard design method for binocular vision detection of a laser deposition part, and relates to the technical field of machine vision. According to the laser deposition part to be detected, an initial calibration chessboard is designed, and an initial calibration board is manufactured; a binocular camera device is used to collect a plurality of groups of calibration pictures of the initial calibration board in different poses, and the number of optimal calibration pictures is determined based on evaluation indexes; according to the field of view of the binocular camera and the size of the to-be-detected region of the laser deposition part to be detected, the optimal proportion of the initial calibration board in the field of view is determined; a proportion coefficient is calculated based on the size of the to-be-detected region of the laser deposition part to be detected and the optimal proportion of the initial calibration board in the field of view, and the optimal small-grid side length is determined based on the proportion coefficient; and the optimal chessboard specification corresponding to the laser deposition part to be detected is calculated according to the optimal small-grid side length. The application can effectively improve the calibration accuracy of the binocular camera in the laser deposition manufacturing environment.

Description

Technical Field

[0001] This invention belongs to the field of machine vision technology, and in particular relates to a checkerboard design method for binocular visual inspection of laser-deposited parts. Background Technology

[0002] Laser Deposition Manufacturing (LDM) is an advanced manufacturing technology that uses a laser as a heat source and metal powder or filament as raw material. It achieves layer-by-layer deposition of metal parts by planning the deposition path through 3D model slicing. LDM is characterized by its mold-free operation, ability to directly form large and complex structural parts, high flexibility, and excellent mechanical properties of the manufactured parts. Figure 1 As shown, during the LDM process, the parts undergo repeated cycles of rapid cooling and heating, resulting in a large thermal gradient and residual stress within them, making them highly susceptible to deformation. To ensure the dimensional accuracy of LDM parts, it is necessary to detect their deformation.

[0003] Machine vision utilizes machine technology to simulate the human eye, employing image processing techniques to reconstruct the three-dimensional information of a target object, thereby enabling its measurement. It acquires information by capturing images. In the field of machine vision technology, compared to monocular and multi-view vision technologies, binocular vision technology has been widely applied in recent years due to its advantages such as high information acquisition efficiency, good portability, simple structure, and low cost. It is used in numerous fields including parts quality inspection, drone obstacle avoidance, industrial condition monitoring, 3D reconstruction, and facial recognition. Binocular vision technology uses left and right cameras of the same model and parameters to acquire two-dimensional digital images of a target object from different positions at the same time. By calculating the disparity between corresponding points of the target object's spatial points in the left and right images, and combining the principles of triangulation and coordinate system transformations, the three-dimensional coordinates of the spatial points are obtained, thus acquiring the three-dimensional information of the target object.

[0004] However, in complex LDM environments, using binocular vision technology for deformation detection of parts suffers from low accuracy and poor stability, becoming a major factor limiting its development. Camera calibration, as one of the core technologies of binocular vision measurement, aims to acquire the intrinsic and extrinsic parameters of the camera device. These parameters will be used in subsequent stereo matching. Therefore, the accuracy of camera calibration directly affects the accuracy of stereo matching, and consequently, the measurement accuracy of the binocular vision system.

[0005] Zhang Zhengyou calibration is a commonly used camera calibration method that uses a checkerboard of two-dimensional squares as a calibration plate and is widely applied in various scenarios. Traditional Zhang Zhengyou calibration often uses only a single-sized calibration plate. However, when the size of the part under test (DUT) is larger than the calibration plate, the distance between the DUT and the binocular camera needs to be increased to ensure complete image acquisition. This makes it impossible to guarantee that the object distance during calibration matches the object distance during image acquisition. When the DUT is smaller than the calibration plate, it creates unnecessary calibration areas, increasing calibration errors. If the fit between the calibration parameters and the DUT cannot be properly guaranteed, the calibration accuracy will be unsatisfactory. Summary of the Invention

[0006] To address the shortcomings of existing technologies, in a first aspect, this invention provides a checkerboard design method for binocular vision inspection of laser-deposited parts, comprising the following steps:

[0007] An initial calibration checkerboard pattern was designed based on the laser deposition part to be tested, and an initial calibration plate was fabricated.

[0008] Using a binocular camera device, several sets of calibration images of the initial calibration board in different poses are acquired, and camera calibration parameters are obtained for each set of calibration images. Evaluation indicators are selected from the camera calibration parameters, and the optimal number of calibration images is determined based on the evaluation indicators.

[0009] Based on the field of view of the binocular camera and the size of the test area of ​​the laser deposition part, determine the optimal proportion of the initial calibration plate in the field of view;

[0010] The scaling factor is calculated based on the size of the test area of ​​the laser-deposited part and the optimal proportion of the initial calibration plate in the field of view, and the optimal small grid side length is determined based on the scaling factor.

[0011] The optimal checkerboard grid specification corresponding to the laser deposition part under test is calculated based on the optimal small grid side length, and a high-precision checkerboard calibration plate corresponding to the laser deposition part under test is generated.

[0012] Furthermore, the calibration area of ​​the initial calibration checkerboard is the test area that completely includes the laser deposition part under test; the calibration area of ​​the initial calibration checkerboard is the remaining area after removing the outermost small squares of the checkerboard.

[0013] Furthermore, the evaluation metrics include effective focal length and principal point coordinates;

[0014] The specific method for determining the number of calibration images based on evaluation metrics is as follows:

[0015] Starting with collecting one set of calibration images, the number of calibration images is gradually increased, and evaluation indicators are obtained from the collected calibration images; when the first set of calibration images is collected... Group calibration image evaluation indicators and the first If the difference in the evaluation metrics of the calibration images is less than the preset threshold, it indicates that the evaluation metrics have converged, and the optimal number of calibration images to be collected is determined. Group.

[0016] Furthermore, the optimal occupancy of the initial calibration plate in the field of view is 50%.

[0017] Furthermore, the specific method for calculating the scaling factor based on the size of the test area of ​​the laser-deposited part and the optimal proportion of the initial calibration plate in the field of view is as follows:

[0018] Let L be the actual side length of each small square in the checkerboard pattern, and let L be the length of the longer side of the laser-deposited part to be tested. The short side length of the laser-deposited part to be tested is The length of the longer side of the field of view of the binocular camera is The length of the shorter side of the field of view of the binocular camera is ,when Greater than hour, ;when Less than hour, To obtain the proportionality coefficient ;

[0019] The specific method for determining the actual side length of the small squares on the chessboard based on the scaling factor is as follows:

[0020] If the optimal side length of each small square in the field of view is set to 15mm, then the actual side length of each small square in the chessboard is... .

[0021] Furthermore, the specific method for calculating the optimal checkerboard pattern for the laser-deposited part under test is as follows:

[0022] The number of grids in the width and height directions is calculated to obtain a high-precision checkerboard calibration plate for the laser-deposited part under test, as shown in the following formula:

[0023] ;

[0024] In the formula, and These represent the number of cells in the width and height directions, respectively. The integer symbol.

[0025] Secondly, the present invention proposes an electronic device comprising: one or more processors, and a memory for storing instructions, which, when executed by the one or more processors, cause the one or more processors to perform the checkerboard design method for binocular visual inspection of laser-deposited parts.

[0026] Thirdly, the present invention proposes a computer-readable storage medium storing executable instructions, which, when executed, cause a processor to execute the checkerboard design method for binocular visual inspection of laser-deposited parts.

[0027] Fourthly, the present invention proposes a computer program product, including a computer program or instructions, which are executed by a processor to perform the checkerboard design method for binocular visual inspection of laser-deposited parts.

[0028] The beneficial effects of adopting the above technical solution are as follows: The checkerboard design method for binocular visual inspection of laser-deposited parts provided by this invention, based on the study of the specific relationship between several main factors affecting camera calibration accuracy (including the number of calibration images, the proportion of the calibration board in the camera's field of view, and the number of corner points of the calibration checkerboard) and calibration accuracy, determines the optimal values ​​of these factors through experimental design, and designs an algorithm to calculate the corresponding high-precision calibration checkerboard specifications based on the size of the laser-deposited part under test. This method can effectively improve the fit between camera calibration parameters and the laser-deposited part under test, as well as the calibration instability caused by uncertain calibration conditions. Compared with the traditional Zhang Zhengyou calibration method, it can better guarantee the fit between calibration parameters and the part under test, and can effectively improve the calibration accuracy of binocular camera calibration in the laser-deposited part environment. Attached Figure Description

[0029] Figure 1 A schematic diagram of laser deposition manufacturing as described in the background art of this invention;

[0030] Figure 2 Flowchart of Zhang Zhengyou's camera calibration method provided in Embodiment 1 of the present invention;

[0031] Figure 3 Flowchart of the chessboard design method provided in Embodiment 2 of the present invention;

[0032] Figure 4 The present invention provides two examples of LDM parts to be tested in Embodiment 2, wherein (a) is LDM part 1 to be tested and (b) is LDM part 2 to be tested;

[0033] Figure 5 The present invention provides two high-precision checkerboard calibration plates for LDM parts under test in Embodiment 2, wherein (a) is a high-precision checkerboard calibration plate for LDM part 1 under test, and (b) is a high-precision checkerboard calibration plate for LDM part 2 under test.

[0034] Figure 6The present invention provides two test checkerboard calibration plates for LDM parts under test in Embodiment 2, wherein (a) is the test checkerboard calibration plate for LDM part 1 under test, and (b) is the test checkerboard calibration plate for LDM part 2 under test. Detailed Implementation

[0035] The specific implementation methods of this application will be further described in detail below with reference to the accompanying drawings and embodiments.

[0036] Example 1:

[0037] Zhang Zhengyou's calibration method is a commonly used camera calibration method. It uses a two-dimensional checkerboard as the calibration board, employing a binocular camera to acquire multiple sets of digital images of the calibration board in different poses. The pixel coordinates of the checkerboard corners are extracted, and the initial intrinsic and extrinsic parameters of the camera are calculated using the homography matrix. Then, the lens distortion coefficients are estimated using the nonlinear least squares method. Finally, maximum likelihood estimation is used to further optimize the parameters. The process is as follows: Figure 2 As shown. Zhang Zhengyou's calibration method, due to its numerous advantages such as simple operation, low cost, and high computational efficiency, is currently widely used in camera calibration across various scenarios.

[0038] To improve the compatibility between camera calibration parameters and the laser-deposited part under test (hereinafter referred to as the LDM part), the Zhang Zhengyou calibration method, after obtaining camera parameters through camera calibration, requires placing the part under test at the same object distance as the calibration plate to acquire images for subsequent stereo matching. Traditional Zhang Zhengyou calibration methods often use only a single-size calibration plate for camera calibration, acquiring camera parameters, and then using these parameters for subsequent stereo matching of parts under test of various sizes. However, when the size of the part under test is larger than the size of the calibration plate, the distance between the part under test and the binocular camera needs to be increased to ensure the part can be completely acquired by the binocular camera. This makes it impossible to guarantee that the object distance during calibration is consistent with the object distance during image acquisition. When the size of the part under test is smaller than the size of the calibration plate, it creates unnecessary calibration areas, increasing calibration errors. Therefore, the traditional Zhang Zhengyou calibration method cannot guarantee a good fit between calibration parameters and the part under test, resulting in unsatisfactory calibration accuracy. Furthermore, the traditional Zhang Zhengyou calibration method does not provide precise values ​​for the main factors affecting calibration accuracy, such as the number of calibration images, calibration plate size and specifications, which seriously affects calibration stability.

[0039] To improve the calibration accuracy and stability of the Zhang Zhengyou calibration method, this embodiment proposes, based on the principle of the Zhang Zhengyou calibration method, to create checkerboard calibration plates with corresponding specifications for LDM parts of different sizes to be tested. The main factors affecting calibration accuracy and their relationship with calibration accuracy are clarified, and based on this, a method is designed to calculate the corresponding high-precision calibration checkerboard specifications according to the dimensions of the LDM parts to be tested, thereby improving calibration accuracy and stability.

[0040] This embodiment presents a checkerboard design method for binocular vision inspection of laser-deposited parts, such as... Figure 3 As shown, it includes the following steps:

[0041] Step 1: Design an initial calibration checkerboard based on the LDM part to be tested, and fabricate an initial calibration board;

[0042] The calibration area of ​​the initial calibration checkerboard is determined based on the size of the LDM part to be tested, so that the calibration area of ​​the initial calibration checkerboard can completely cover the test area of ​​the LDM part to be tested. The initial side length of each small square of the initial calibration checkerboard is set to 20mm. The calibration area of ​​the calibration checkerboard is the remaining area after removing the outermost small squares of the checkerboard.

[0043] Step 2: Use a binocular camera to acquire several sets of calibration images of the initial calibration board in different poses, and obtain the camera calibration parameters for the number of calibration images in each set; select evaluation indicators from the camera calibration parameters, and determine the optimal number of calibration images based on the evaluation indicators;

[0044] Evaluation metrics include effective focal length and principal point coordinates, which are used to accurately reflect the changing trend of calibration accuracy in order to obtain the relationship between the number of calibration images and camera calibration parameters.

[0045] The specific method for determining the number of calibration images based on evaluation metrics is as follows:

[0046] Starting with collecting one set of calibration images, the number of calibration images is gradually increased, and evaluation metrics, including effective focal length and principal point coordinates, are obtained from the collected calibration images; when the first set of calibration images is collected... Group calibration image evaluation indicators and the first If the difference in the evaluation metrics of the calibration images is less than the preset threshold, it indicates that the evaluation metrics have converged, and the optimal number of calibration images to be collected is determined. Group;

[0047] This embodiment clarifies the specific relationship between the number of calibration images and calibration accuracy when using the Zhang Zhengyou method for calibration, determining the optimal number of calibration images to be 20. Thirty sets of digital images of the calibration board in different poses were acquired using a binocular camera. By varying only the number of calibration images, camera calibration parameters for the calibration board were obtained under different image counts. The effective focal length and principal point coordinates were selected as evaluation criteria (effective focal length and principal point coordinates accurately reflect the changing trend of calibration accuracy) to obtain the relationship between the number of calibration images and calibration accuracy. In this embodiment, as the number of calibration images increases, the evaluation indicators (i.e., effective focal length and principal point coordinates) continuously change. Initially, the changes fluctuate significantly because the number of calibration images is small, resulting in low and unstable calibration accuracy. As the number of calibration images increases, the effective focal length and principal point coordinates gradually converge. When the number of images reached 20 sets, further increases in the number of images resulted in near-convergence of the effective focal length and principal point coordinates, with no further significant fluctuations. This indicates that after reaching 20 sets of calibration images, further increases in the number of images no longer negatively impacted calibration accuracy. Therefore, the optimal number of calibration images was determined to be 20 sets.

[0048] Step 3: Determine the optimal proportion of the initial calibration plate in the field of view based on the field size of the binocular camera and the size of the test area of ​​the LDM part to be tested;

[0049] When acquiring calibration images with a stereo camera, it is essential to ensure that the calibration board occupies a certain proportion within the camera's field of view for clear image recognition. Traditional Zhang Zhengyou calibration methods require the calibration board to occupy as large a proportion as possible in the field of view, but there is no specific specification for this proportion. Furthermore, an excessively large or small proportion will prevent the calibration algorithm from clearly recognizing the calibration image, resulting in low calibration accuracy and poor stability.

[0050] This embodiment involves fabricating calibration plates of different specifications for LDM parts of varying sizes. The optimal proportion of the calibration plate in the field of view differs for parts of different shapes and sizes. In actual measurement environments, although part specifications may vary, the size of the binocular camera's field of view remains constant. Therefore, this embodiment experimentally studies the impact of the ratio between the side length of the calibration plate and the corresponding side length of the field of view in the calibration image on calibration accuracy, thereby obtaining the optimal proportion of the calibration plate in the field of view during calibration.

[0051] The binocular camera used in this embodiment is 1280×720 pixels, with an actual field of view size of 230mm×130mm. Twenty sets of binocular images of the calibration board in different poses were acquired under different object distances, i.e., different proportions of the calibration board and the field of view, for calibration, obtaining calibration parameters under different proportions. The size of the area to be measured for a certain part is... ,if For parts larger than 230mm × 130mm (i.e., 1280 / 720), the proportion of the corresponding calibration plate in the field of view is determined by the long side of the part (the long side of the calibration area of ​​the calibration plate); if If the size is less than 230mm / 130mm (i.e., 1280 / 720), the proportion of the corresponding calibration plate in the field of view is determined by the short side of the part (the short side of the calibration area of ​​the calibration plate).

[0052] Based on the principles of camera calibration and triangulation, when the pixel coordinates of a point on the test piece in the left and right images are known, the three-dimensional coordinates of that point can be obtained from the camera parameters obtained during calibration. This invention uses the first set of calibration board images acquired under different aspect ratios as calibration accuracy evaluation images for three-dimensional measurement. First, distortion correction is performed on the corresponding accuracy evaluation images using the distortion coefficients obtained from calibration. Then, two adjacent checkerboard corner points are selected from the accuracy evaluation images, and their pixel coordinates in the left and right images are extracted. These coordinates are then combined with the calibration parameters to calculate the three-dimensional coordinates of the two points. Finally, the distance between the two points can be calculated based on their three-dimensional coordinates.

[0053] In this embodiment, the side length of the small square in the accuracy evaluation checkerboard is 20mm. Therefore, the closer the distance between two points calculated through the above process is to 20mm, the higher the camera calibration accuracy. Experimental results show that when the ratio of the corresponding side of the calibration plate's calibration area to the corresponding side of the field of view is approximately 50%, the measured distance between two adjacent corner points of the checkerboard is closest to 20mm. In summary, the optimal ratio of the part's side length (i.e., the side length of the calibration plate's calibration area corresponding to the part's specifications) to the corresponding side length of the field of view is determined to be 50%.

[0054] Step 4: Calculate the scaling factor based on the size of the test area of ​​the LDM part to be tested and the optimal proportion of the initial calibration plate in the field of view, and determine the optimal small grid side length based on the scaling factor;

[0055] This embodiment proposes to create corresponding checkerboard calibration plates for calibration of parts under test (LDMs) of different specifications. The number of checkerboard corner points is determined by the size of the test area of ​​the LDM part and the side length of each small square in the checkerboard. For LDM parts of a fixed size, increasing the number of checkerboard corner points improves calibration accuracy. However, after reaching a certain value, further increasing the number of corner points actually decreases accuracy. This is because when the number of checkerboard corner points is too large, the side length of each small square in the checkerboard becomes too small. The calibration algorithm needs to identify the corner points and the side lengths of the small squares for calibration. If the side length of the small squares is too small, it cannot be clearly identified by the calibration algorithm, leading to reduced calibration accuracy. Therefore, this embodiment designs an experiment to study the impact of the side length of the small squares on the calibration accuracy, thereby obtaining the smallest side length of the small square that can be clearly identified by the calibration algorithm in the checkerboard image acquired by the camera device.

[0056] This embodiment addresses LDM parts of different sizes by fabricating checkerboard calibration plates with varying side lengths for each small grid cell, tailored to the test area of ​​the LDM part. During calibration plate fabrication, it is ensured that the checkerboard calibration area completely encompasses the test area of ​​the part. With the calibration plate occupying 50% of the field of view, 20 sets of digital images are acquired at different poses for calibration. The evaluation image is a checkerboard calibration plate with each small grid cell measuring 20mm × 20mm. The highest calibration accuracy is achieved when the distance between two points calculated from the calibration parameters is closest to 20mm; the corresponding side length of each small grid cell on the checkerboard calibration plate at this point is then determined. This is the optimal small grid side length for the corresponding checkerboard calibration plate of the part.

[0057] Different specifications of parts correspond to different optimal small grid side lengths for the checkerboard calibration board. However, in the checkerboard calibration board image acquired by the same device, the smallest small grid side length that the calibration algorithm can clearly identify is fixed. That is, the optimal small grid side length of the checkerboard in the field of view is fixed.

[0058] The optimal side length of a small grid cell in the field of view needs to be obtained based on the actual side length of the small grid cells and their corresponding proportional relationships. Let the actual side length of the small grid cell in the field of view be L, and the length of the long side of the LDM part to be measured be... The short side length of the LDM part to be tested is The length of the longer side of the field of view is The length of the shorter side of the field of view is Calculate the proportionality coefficient :

[0059] when Greater than The optimal ratio of the longer side of the calibration area to the longer side of the field of view on the corresponding calibration plate is 50%. ;when Less than The optimal ratio of the short side of the calibration area to the short side of the field of view on the corresponding calibration plate is 50%. The proportionality coefficient can be obtained. The optimal side length of each small square in the field of view is Then the actual side length of the small grid is Based on multiple sets of experimental results, the optimal side length of the small squares in the checkerboard pattern within the field of view was determined to be... .

[0060] Step 5: Calculate the optimal checkerboard grid specifications corresponding to the LDM part to be tested based on the optimal small grid side length, and generate a high-precision checkerboard grid calibration plate corresponding to the LDM part to be tested.

[0061] The optimal checkerboard pattern for the LDM part to be tested is shown in the following formula:

[0062] ;

[0063] In the formula, and These represent the number of squares in the width and height directions of the chessboard, respectively. The integer sign indicates that the result is less than 1. The largest integer value within the inner value. Since the small squares in the checkerboard are all squares of the same size, and the calculation of equation (1) is often not divisible, in order to ensure that all the small squares in the checkerboard are squares of the same size and that the calibration area can completely contain the part to be measured, the ratio of the side length of the part to the side length of the small square in the checkerboard is first rounded up and then 1 is added to calculate the specification of the checkerboard calibration area. Then, 2 is added to the width and height directions respectively to complete the outer ring of the non-calibrated area of ​​the checkerboard, and the high-precision calibration checkerboard specification corresponding to the LDM part to be measured is obtained.

[0064] Example 2:

[0065] This embodiment designs two calibration checkerboard patterns for LDM parts under test based on the high-precision checkerboard design method for binocular vision inspection of laser deposition manufactured parts described in Embodiment 1. The binocular vision inspection camera used in this embodiment is a WP-GE series area array industrial camera with a frame rate of at least 30 FPS and a GIGE interface with a data transmission speed greater than 100 MB / s. The camera lens is a zoom industrial lens compatible with the camera, with an adjustable focal length range of 7-36 mm. The resolution of a single image is 1280 dpi × 720 dpi. The binocular camera is fixed using a multi-degree-of-freedom camera bracket compatible with the industrial camera, and the camera baseline distance is adjustable from 60-300 mm.

[0066] To improve calibration accuracy, a calibration method is proposed that uses a checkerboard pattern corresponding to the specifications of laser deposition-manufactured parts of different sizes for calibration. The specific relationships between the number of calibration images, the proportion of the calibration board in the field of view, the number of corner points of the calibration checkerboard, and calibration accuracy are clarified, and their optimal values ​​are determined. Based on the established mathematical model, the corresponding high-precision calibration checkerboard specifications are calculated for the dimensions of the part under test. Finally, an experiment is designed to compare the calibration accuracy of this method with the traditional Zhang Zhengyou calibration method, verifying that the calibration method proposed in this embodiment can effectively improve calibration accuracy. The flowchart of this embodiment is as follows. Figure 3 As shown, the two types of LDM parts to be tested in this embodiment are as follows: Figure 4 As shown, (a) is the LDM part to be tested 1, and (b) is the LDM part to be tested 2.

[0067] against Figure 4 For two types of LDM parts to be tested, a calibration checkerboard pattern of corresponding specifications is designed. The specific implementation plan is as follows:

[0068] (1) First, the number of calibration images is determined to be 20, and the calibration plate corresponding to the part accounts for about 50% of the field of view.

[0069] (2) Determine the relationship between the ratio of the long side to the short side of the part and the ratio of the long side to the short side of the binocular camera's field of view, and use this to determine whether the proportion of the calibration plate corresponding to the part in the field of view is determined by the long side or the short side. For part 1, the size of its area to be measured is 130mm × 50mm. Since 130mm / 50mm is greater than 230mm / 130mm, the proportion of its corresponding calibration plate in the field of view is determined by the long side of the part (the long side of the calibration area of ​​the calibration plate). For part 2, the size of its area to be measured is 90mm × 70mm. Since 90mm / 70mm is less than 230mm / 130mm, the proportion of its corresponding calibration plate in the field of view is determined by the short side of the part (the short side of the calibration area of ​​the calibration plate).

[0070] (3) The corresponding scaling factor m is calculated based on the optimal ratio of the calibration area of ​​the corresponding calibration plate to the corresponding side length of the field of view, which is 50%. For part 1, the optimal ratio of the long side of its corresponding calibration plate calibration area to the long side of the field of view is approximately 50%, i.e., am / 230=50%, so the scaling factor m≈0.88 can be obtained. For part 2, the optimal ratio of the short side of its corresponding calibration plate calibration area to the short side of the field of view is approximately 50%, i.e., bm / 130=50%, so the scaling factor m≈0.93 can be obtained.

[0071] (4) Based on the optimal side length of the small square in the field of view being 15mm, the side length of each small square in the actual calibration checkerboard is calculated as L=15 / m. For part 1, L=17mm; for part 2, L=16mm.

[0072] (5) Using the algorithm designed in this embodiment for calculating the high-precision calibration chessboard grid specifications, calculate the number of grids W and H in the width and height directions of the chessboard grid. Based on these specifications, fabricate high-precision calibration chessboard grid calibration plates corresponding to parts 1 and 2, such as... Figure 5 As shown. Part 1 corresponds to a high-precision checkerboard grid of 10×5, with each small square measuring 17mm×17mm. Part 2 corresponds to a high-precision checkerboard grid of 8×7, with each small square measuring 16mm×16mm.

[0073] (6) To more easily verify the calibration accuracy of this embodiment, the measurement objects in this example are the checkerboard grids with each small square having a size of 20mm × 20mm corresponding to parts 1 and 2. The checkerboard grid size corresponding to the accuracy verification of LDM part 1 is 7×3, and the checkerboard grid size corresponding to the accuracy verification of LDM part 2 is 5×4, as shown below. Figure 6 As shown.

[0074] (7) Under the condition that the ratio of the long side of the calibration area to the long side of the field of view of the high-precision calibration plate corresponding to LDM part 1 is about 50%, and under the condition that the ratio of the short side of the calibration area to the short side of the field of view of the high-precision calibration plate corresponding to LDM part 2 is about 50%, 20 digital images of the calibration plate at different angles are obtained respectively, and calibration is carried out according to the traditional Zhang Zhengyou calibration method.

[0075] (8) Acquire a set of binocular images of the chessboard to be tested corresponding to Part 1 under the calibration conditions of the high-precision calibration plate corresponding to Part 1; acquire a set of binocular images of the chessboard to be tested corresponding to Part 2 under the calibration conditions of the high-precision calibration plate corresponding to Part 2. Calculate the three-dimensional coordinates of all corner points of the chessboard in the images to be tested acquired under the calibration conditions and save them as txt format files. Then, use the point cloud processing algorithm based on the PCL point cloud library and the Visual Studio software platform to open the txt format file and display it as a point cloud. Use the two-point distance measurement algorithm in the PCL library to measure the distance between all adjacent corner points of the point cloud data along the same direction.

[0076] (9) The corner spacing of the corresponding chessboard to be tested, calculated according to the calibration parameters obtained from the high-precision chessboard calibration plates corresponding to parts 1 and 2, is concentrated in the range of 19.6mm-20.4mm, with an error fluctuation range of -0.4mm to +0.4mm. However, the corner spacing of the corresponding chessboard to be tested, calculated according to the calibration parameters obtained from the standard chessboard calibration plate used in the traditional Zhang Zhengyou method, is concentrated in the range of 19.4mm-20.6mm, with an error fluctuation range of -0.6mm to +0.6mm.

[0077] Example 3:

[0078] This embodiment proposes an electronic device, including: one or more processors, and a memory for storing instructions. When the instructions are executed by the one or more processors, the one or more processors cause the one or more processors to execute the high-precision checkerboard design method for binocular vision inspection of laser deposition manufactured parts.

[0079] The electronic device may be a mobile phone, computer, or tablet computer, etc., and includes a memory and a processor. The memory stores a computer program, which, when executed by the processor, implements the high-precision checkerboard design method for binocular vision inspection of laser deposition manufactured parts as described in the embodiments. It is understood that the electronic device may also include input / output (I / O) interfaces and communication components.

[0080] The processor is used to execute all or part of the steps in the high-precision checkerboard design method for binocular vision inspection of laser-deposited manufactured parts as described in the above embodiments. The memory is used to store various types of data, which may include, for example, instructions for any application or method in the electronic device, as well as application-related data.

[0081] The processor can be implemented as an Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), controller, microcontroller, microprocessor, or other electronic components, and is used to execute the high-precision checkerboard design method for binocular vision inspection of laser deposition manufactured parts described in the above embodiments.

[0082] Example 4:

[0083] This embodiment proposes a computer-readable storage medium that stores executable instructions. When these instructions are executed, if they are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0084] The computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the high-precision checkerboard design method for binocular visual inspection of laser deposition manufactured parts as described in the various embodiments of this application.

[0085] The aforementioned storage media include: flash memory, hard disks, multimedia cards, card-type memory (e.g., SD (Secure Digital Memory Card) or DX (Memory Data Register, MDR) memory), random access memory (RAM), static random-access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic storage, disks, optical discs, servers, APP (Application) app stores, and other media capable of storing program verification codes. These media store computer programs, which, when executed by a processor, can implement the various steps of the high-precision checkerboard design method for binocular visual inspection of laser-deposited manufactured parts described above.

[0086] Example 5:

[0087] This embodiment proposes a computer program product, including a computer program or instructions, which, when executed by a processor, implements the high-precision checkerboard design method for binocular visual inspection of laser deposition manufactured parts.

[0088] Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a computer program product.

[0089] The various embodiments in this application are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0090] The scope of protection of this application is not limited to the embodiments described above. Obviously, those skilled in the art can make various modifications and variations to this disclosure without departing from the scope and spirit of this disclosure. If such modifications and variations fall within the scope of this disclosure and its equivalents, then the intent of this disclosure also includes these modifications and variations.

Claims

1. A checkerboard design method for binocular vision inspection of laser-deposited parts, characterized in that, Includes the following steps: An initial calibration checkerboard pattern was designed based on the laser deposition part to be tested, and an initial calibration plate was fabricated. Using a binocular camera device, several sets of calibration images of the initial calibration board in different poses are acquired, and camera calibration parameters are obtained for each set of calibration images. Evaluation indicators are selected from the camera calibration parameters, and the optimal number of calibration images is determined based on the evaluation indicators. Based on the field of view of the binocular camera and the size of the test area of ​​the laser deposition part, determine the optimal proportion of the initial calibration plate in the field of view; The scaling factor is calculated based on the size of the test area of ​​the laser-deposited part and the optimal proportion of the initial calibration plate in the field of view, and the optimal small grid side length is determined based on the scaling factor. The optimal checkerboard grid specification corresponding to the laser deposition part under test is calculated based on the optimal small grid side length, and a high-precision checkerboard calibration plate corresponding to the laser deposition part under test is generated.

2. The checkerboard design method for binocular vision inspection of laser-deposited parts according to claim 1, characterized in that, The calibration area of ​​the initial calibration checkerboard is the area to be tested that completely includes the laser-deposited part under test; the calibration area of ​​the initial calibration checkerboard is the area remaining after removing the outermost small squares of the checkerboard.

3. The checkerboard design method for binocular vision inspection of laser-deposited parts according to claim 1, characterized in that, The evaluation metrics include effective focal length and principal point coordinates; The specific method for determining the number of calibration images based on evaluation metrics is as follows: Starting with the collection of 1 set of calibration images, the number of calibration images is gradually increased, and evaluation indicators are obtained from the collected calibration images. When the difference between the evaluation indicators of the nth set of calibration images and the evaluation indicators of the (n-1)th set of calibration images is less than the preset threshold, it indicates that the evaluation indicators have converged, and the optimal number of calibration images to be collected is determined to be n sets.

4. The checkerboard design method for binocular vision inspection of laser-deposited parts according to claim 1, characterized in that, The optimal occupancy of the initial calibration plate in the field of view is 50%.

5. The checkerboard design method for binocular vision inspection of laser-deposited parts according to claim 1, characterized in that, The specific method for calculating the scaling factor based on the size of the test area of ​​the laser-deposited part and the optimal proportion of the initial calibration plate in the field of view is as follows: Let L be the actual side length of each small square in the checkerboard pattern, and let L be the length of the longer side of the laser-deposited part to be tested. The short side length of the laser-deposited part to be tested is The length of the longer side of the field of view of the binocular camera is The length of the shorter side of the field of view of the binocular camera is ,when Greater than hour, ;when Less than hour, To obtain the proportionality coefficient ; The specific method for determining the actual side length of the small squares on the chessboard based on the scaling factor is as follows: If the optimal side length of each small square in the field of view is set to 15mm, then the actual side length of each small square in the chessboard is... .

6. The checkerboard design method for binocular vision inspection of laser-deposited parts according to claim 5, characterized in that, The specific method for calculating the optimal checkerboard pattern for the laser-deposited part under test is as follows: The number of grids in the width and height directions is calculated to obtain a high-precision checkerboard calibration plate for the laser-deposited part under test, as shown in the following formula: ； In the formula, and These represent the number of cells in the width and height directions, respectively. The integer symbol.

7. An electronic device, comprising: One or more processors, and a memory for storing instructions that, when executed by the one or more processors, cause the one or more processors to perform the checkerboard design method for binocular visual inspection of laser-deposited parts as described in any one of claims 1 to 6.

8. A computer-readable storage medium storing executable instructions that, when executed, cause a processor to perform the checkerboard design method for binocular visual inspection of laser-deposited parts as described in any one of claims 1 to 6.

9. A computer program product comprising a computer program or instructions which, when executed by a processor, implement the checkerboard design method for binocular visual inspection of laser-deposited parts as described in any one of claims 1 to 6.

Images