A 3D vision measurement method and system for a foundry cleaning robot

Abstract

The application discloses a kind of casting cleaning robot 3D vision measurement method and measurement system, and the point cloud data of the first angle of the casting to be cleaned is acquired by 3D binocular camera and digital grating projector, and the relative motion path when 3D binocular camera is photographed to casting is recorded, based on this path, the point cloud data of multiple angles is acquired, point cloud registration operation is carried out, and complete point cloud data is obtained;It is divided into six regions, and the contour information and position information of the cleaning site corresponding to each region are obtained;Based on the contour information of six regions, the model of the casting to be cleaned is obtained by surface reconstruction, and the final contour information and position information of the cleaning site are obtained, and the visual measurement of the casting to be cleaned is completed.The detection method is not affected by external conditions such as light, the accuracy of the detection result is high, and the work efficiency is high, and the characteristics of the workpiece with complex structure can be identified.

Description

Technical Field

[0001] This invention relates to the field of machine vision-assisted robot casting cleaning technology, and in particular to a 3D vision measurement method and system for casting cleaning robots. Background Technology

[0002] With the development of robotics and automation technologies, industrial robots are increasingly replacing human labor in demanding tasks performed in harsh environments. The rapid advancement of industrial robot technology has enabled them to replace manual labor in cleaning impurities from castings after casting. The removal of these impurities is crucial to the final quality of the castings; however, current casting impurity cleaning technologies still cannot meet the demands of high efficiency and high quality in actual production. In many cases, manual processing is still necessary. This not only results in harsh working environments that can harm workers' health, but also makes it difficult to guarantee the effectiveness of impurity removal. Ultimately, the cleaning result largely depends on the worker's experience, significantly impacting production efficiency and quality.

[0003] A crucial aspect of utilizing robotics to replace manual labor in cleaning casting impurities is the identification and measurement of these impurity characteristics. In simple cases, cleaning paths can be planned offline through programming; however, this requires a complete understanding of the workpiece's 3D model, making offline programming complex, time-consuming, and inefficient. In some scenarios, 2D images are used to detect defects, but this method is susceptible to external factors such as lighting conditions, resulting in low accuracy and an inability to identify features in complex structures. Compared to 2D images, 3D point clouds contain more comprehensive and richer representational information, capable of accurately displaying the geometry and dimensions of complex workpieces. While 3D vision technology possesses powerful feature representation capabilities, its application is challenging, and measurement accuracy remains low. Summary of the Invention

[0004] This invention proposes a 3D vision measurement method and system for casting cleaning robots to overcome the above-mentioned problems.

[0005] The technical solution adopted in this application is as follows:

[0006] A 3D vision measurement method for a casting cleaning robot includes the following steps:

[0007] Step 1: Install the casting to be cleaned of impurities onto the positioner fixture of the casting cleaning robot, so that the casting to be cleaned of impurities is fixed at a first angle; the first angle is the angle at which the casting to be cleaned of impurities is positioned when it is installed on the positioner.

[0008] Step 2: Use Zhang Zhengyou's calibration method to complete the hand-eye calibration of the casting cleaning robot and the 3D binocular camera;

[0009] Step 3: Use a digital grating projector and the 3D binocular camera to take pictures of the casting to be cleaned from the first angle to obtain point cloud data from the first angle, and record the relative motion path used by the 3D binocular camera relative to the casting to be cleaned when taking pictures.

[0010] Step 4: Adjust the position of the casting to be cleaned by the positioner of the casting cleaning robot, so that the casting to be cleaned is fixed at the 2nd angle, the 3rd angle, ..., the nth angle; the 1st angle, the 2nd angle, ..., the nth angle are all different.

[0011] Step 5: Based on the relative motion path, a digital grating projector and the 3D binocular camera are used to take pictures of the casting to be cleaned of impurities at the 2nd angle, the 3rd angle, ..., the nth angle, respectively, so as to obtain point cloud data at the 2nd angle, the 3rd angle, ..., the nth angle;

[0012] Step 6: Based on the point cloud data of the first angle, the second angle, the third angle, ..., the nth angle, perform point cloud registration to obtain the complete point cloud data of the casting to be cleaned of impurities.

[0013] Step 7: Filter the complete point cloud data and divide the filtered complete point cloud data into six regions to obtain point cloud data for six regions. Identify and measure the parts to be cleaned in the point cloud data of the six regions respectively, and obtain the contour information and position information of the parts to be cleaned in each region.

[0014] Step 8: Based on the contour information of the part to be cleaned corresponding to each region, perform surface reconstruction to obtain the model of the casting to be cleaned. Then, compare the model of the casting to be cleaned with the standard digital model file of the casting to obtain the final contour information and position information of the part to be cleaned. Based on the final contour information and position information of the part to be cleaned, control the casting robot to clean the casting to be cleaned and complete the visual measurement of the casting to be cleaned.

[0015] Furthermore, the method for dividing the filtered complete point cloud data into six regions is as follows:

[0016] 1) Obtain the two points with the largest distance from the complete point cloud data and connect them to form a line segment;

[0017] 2) Divide the line segment into three equal parts by taking two equal division points on the line segment;

[0018] 3) Obtain the vertical plane that passes through the dividing point and is perpendicular to the line segment;

[0019] 4) Obtain the plane passing through the line segment, and the plane and the vertical plane divide the complete point cloud data into six regions, such that the reference feature of the casting to be cleaned of impurities is located in any of the six regions.

[0020] Furthermore, in step 3, the method for obtaining the point cloud data at the first angle is as follows:

[0021] 1) Using OpenCV technology, a digital raster projector is used to project a structural light pattern onto the casting to be cleaned of impurities;

[0022] 2) Use a 3D vision camera to photograph the casting to be cleaned of impurities, which has a structured light pattern projected onto it. Based on the outline of the casting to be cleaned of impurities, obtain the three-dimensional surface information of the casting to be cleaned of impurities, and then obtain the point cloud data of the casting to be cleaned of impurities.

[0023] Furthermore, after step 5, the method further includes: using the Qt graphical user interface for visualization, based on the Qt software's signal and slot mechanism, to realize the casting imaging results at the 1st angle, 2nd angle, 3rd angle, ..., nth angle, point cloud data at the 1st angle, 2nd angle, 3rd angle, ..., nth angle, complete point cloud data, regional point cloud data, contour information and position information of the parts to be cleaned in each region, the model of the casting, and the final visualization of the impurities to be cleaned.

[0024] A measurement system for a 3D vision measurement method for a casting cleaning robot, characterized in that it includes a data acquisition module, a point cloud data processing module, and a human-computer interaction interface module;

[0025] The data acquisition module includes a 3D binocular camera and a digital grating projector, which are used in conjunction with the positioner of the casting cleaning robot to take pictures of the casting at the first angle, the second angle, the third angle, ..., the nth angle using a relative motion path, and obtain point cloud data of the casting to be cleaned at the first angle, the second angle, the third angle, ..., the nth angle.

[0026] The point cloud data processing module is used to perform point cloud registration based on point cloud data from the 1st angle, 2nd angle, 3rd angle, ..., nth angle to obtain complete point cloud data of the casting; to filter the complete point cloud data and divide the filtered complete point cloud data into six regions to obtain point cloud data of six regions; to identify and measure the parts to be cleaned in the point cloud data of the six regions respectively, and to obtain the contour information and position information of the parts to be cleaned in each region; to perform surface reconstruction based on the contour information of the parts to be cleaned in each region to obtain a model of the casting to be cleaned, and to further compare it with the standard digital model file of the casting to obtain the final contour information and position information of the parts to be cleaned; and to complete the visual measurement of the casting to be cleaned.

[0027] The human-computer interaction interface module is used to utilize the visualization interface of the Qt graphical user interface, based on the signal and slot mechanism of Qt software, to realize the casting shooting results from the 1st angle, the 2nd angle, the 3rd angle, ..., the nth angle, the point cloud data from the 1st angle, the 2nd angle, the 3rd angle, ..., the nth angle, the complete point cloud data, the regional point cloud data, the contour information and position information of the parts to be cleaned in each region, the model of the casting, and the visualization of the final parts of the impurities to be cleaned.

[0028] Beneficial Effects: This invention provides a 3D vision measurement method and system for a casting cleaning robot. It acquires point cloud data of the casting to be cleaned from a first angle using a 3D binocular camera and a digital grating projector. The relative motion path of the 3D binocular camera when photographing the casting is recorded. Based on this path, point cloud data from the second, third, ..., nth angles are acquired, and point cloud registration is performed to obtain complete point cloud data of the casting to be cleaned. After filtering, the complete point cloud data is divided into six regions, and the contour and position information of the corresponding cleaned area in each region are acquired. Further, surface reconstruction is performed based on the contour information of the six regions to obtain a model of the casting to be cleaned, thus obtaining the final contour and position information of the cleaned area. The casting robot is then controlled to clean the casting, completing the visual measurement of the casting to be cleaned. This detection method is unaffected by external conditions such as lighting, has high accuracy and high efficiency, and can identify features of complex workpieces. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a flowchart of the 3D vision measurement method of the present invention;

[0031] Figure 2 This is a block diagram of the 3D vision measurement method module for the casting cleaning robot of the present invention;

[0032] Figure 3 This is a schematic diagram of the 3D vision measurement method in an embodiment of the present invention;

[0033] Figure 4 This is a schematic diagram showing the casting to be cleaned of impurities divided into six regions in an embodiment of the present invention. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] This embodiment discloses a 3D vision measurement method for a casting cleaning robot, used to assist the casting cleaning robot in quickly and accurately measuring the position and contour of impurities to be cleaned in castings, including the following steps, such as... Figure 1 , 3 As shown:

[0036] Step 1: Install the casting to be cleaned of impurities onto the positioner fixture of the casting cleaning robot; so that the casting to be cleaned of impurities is fixed at the first angle;

[0037] Step 2: Use the existing Zhang Zhengyou calibration method to complete the hand-eye calibration of the casting cleaning robot and the 3D binocular camera;

[0038] Step 3: Use a digital grating projector and the 3D binocular camera to take pictures of the casting to be cleaned from the first angle to obtain point cloud data from the first angle, and record the relative motion path of the 3D binocular camera with respect to the casting to be cleaned.

[0039] Preferably, in step 3, the method for obtaining the point cloud data of the first angle is as follows:

[0040] 1) Using OpenCV technology, a digital raster projector is used to project a structural light pattern onto the casting to be cleaned of impurities;

[0041] 2) Use a 3D vision camera to photograph the casting to be cleaned of impurities, which has a structured light pattern projected onto it. Based on the outline of the casting to be cleaned of impurities, obtain the three-dimensional surface information of the casting to be cleaned of impurities, and then obtain the point cloud data of the casting to be cleaned of impurities.

[0042] Specifically, the digital grating projector and the 3D vision camera work as a photographic unit to photograph the casting, and their relative positions are fixed.

[0043] Step 4: Adjust the position of the casting to be cleaned by the positioner of the casting cleaning robot, so that the casting to be cleaned is fixed at the second angle, the third angle, ..., the nth angle respectively;

[0044] Specifically, the casting cleaning robot's end effector is equipped with a 3D binocular camera, and the casting to be cleaned is fixed on the positioner fixture of the casting cleaning robot. The positioner adjusts the posture of the casting fixed on it.

[0045] Step 5: Based on the relative motion path, a digital grating projector and the 3D binocular camera are used to take pictures of the casting to be cleaned of impurities at the 2nd angle, the 3rd angle, ..., the nth angle, respectively, so as to obtain point cloud data at the 2nd angle, the 3rd angle, ..., the nth angle;

[0046] Step 6: Based on the point cloud data of the first angle, the second angle, the third angle, ..., the nth angle, perform point cloud registration to obtain the complete point cloud data of the casting to be cleaned of impurities.

[0047] Specifically, OpenCV technology is used to control a digital grating projector to project structured light patterns onto the casting to be measured. A 3D vision camera obtains the 3D surface information of the casting by mapping the depth information of the casting surface to the grating phase information. The casting is then photographed from multiple angles to acquire point cloud data files. To adapt to the working environment of cleaning large castings, a metal casing is used to encapsulate the 3D binocular camera and digital grating projector; the metal casing provides protection and significantly extends the lifespan of the binocular camera equipment.

[0048] Specifically, the first angle, the second angle, the third angle, ..., the nth angle are all shooting angles set by the staff in the field. The relative motion path is recorded after shooting at the first angle. In other application examples, the relative motion path can also be preset in advance, and the casting can be photographed based on the preset relative motion path.

[0049] Specifically, this embodiment employs an improved ICP method. After photographing the casting from the first angle, the shooting path of the 3D dual-sided camera relative to the casting is recorded. When the casting is adjusted to other angles, it is photographed using the same relative motion path. This yields point cloud data of the casting photographed from different angles using the same relative motion path. Point cloud registration is performed on these point cloud data from multiple angles to obtain the complete point cloud data of the casting. Since the spatial coordinate systems of the multiple point cloud data acquired from multiple angles are not unified, this embodiment uses point cloud data obtained through a unified relative motion path to unify the multiple point cloud data into the same coordinate system, thus obtaining the complete point cloud data of the casting. The registration efficiency is high, perfectly solving the problem that the traditional ICP algorithm has high requirements for the initial position and is prone to local optima, thus failing to match successfully. Its registration accuracy and efficiency have been greatly improved. Combined with the casting cleaning robot in this embodiment, in actual application, a large number of castings are placed on the robot positioner. Compared with traditional industrial production, it has a certain degree of path order. The positioner of the casting cleaning robot adjusts the pose of the castings, and the angle and order of the 3D binocular camera to capture a large number of castings are fixed. In this way, the process of registering point cloud data is consistent each time. This method is suitable for the need of registering large batches of casting point cloud data in this invention, greatly reducing the complexity of registration and improving the efficiency of point cloud processing.

[0050] Step 7: Filter the complete point cloud data and divide it into six regions to obtain point cloud data for each region. Use the PCL point cloud processing library to identify and measure the areas to be cleaned in each of the six regions, obtaining the contour and location information of these areas. Specifically, identifying and measuring the features of impurities to be cleaned in the overall casting is quite difficult. Dividing it into six regions ensures that each region contains different impurity features, facilitating separate processing of these features. The division steps are as follows: First, analyze the two points with the largest distance in the point cloud, which correspond to the two endpoints of the casting. Then, divide the line segment connecting these two endpoints into three equal parts. Divide the region through the trisection points perpendicular to the line connecting the two endpoints, ultimately dividing it into six regions.

[0051] Preferably, the method for dividing the filtered complete point cloud data into six regions is as follows: Figure 4 As shown,

[0052] 1) Obtain the two points with the largest distance from the complete point cloud data and connect them to form a line segment;

[0053] 2) Take two equal division points a and b on the line segment, and divide the line segment into three equal parts;

[0054] 3) Obtain vertical planes A and B that pass through the points of equal division and are perpendicular to the line segment;

[0055] 4) Obtain the plane C passing through the line segment, and the plane C, together with the vertical plane A and the vertical plane, divide the complete point cloud data into six regions. The reference feature of the casting to be cleaned of impurities is located in any of the six regions. Specifically, in the casting industry, during casting and mold demolding, there will be reference features based on the mold.

[0056] In one embodiment of the present invention, the reference feature on the casting is a pin hole 1. Specifically, when the two points with the largest distance in the complete point cloud data are obtained and connected to form a line segment, and then the data is divided into six regions, if the reference feature cannot be placed in any of the six regions, then the adjacent points of the two points with the largest distance in the complete point cloud data are obtained and connected to form a line segment. The initial position of the line segment is adjusted until the reference feature is placed in any of the six regions, and two equally divided points are selected on the obtained line segment.

[0057] Specifically, the point cloud data processing module, based on the PCL point cloud processing algorithm, performs filtering, feature recognition, surface reconstruction, and point cloud visualization on the point cloud data of the casting to be cleaned in the six regions defined above.

[0058] Specifically, in order to measure the contour and position information of the parts to be cleaned in the casting with high precision, the method of casting regional feature recognition is adopted, that is, the complete point cloud data of the casting is divided into six regions, and the parts to be cleaned in these six regions are identified and measured respectively.

[0059] Step 8: Based on the contour information of the parts to be cleaned in each region, perform surface reconstruction to obtain the model of the casting to be cleaned, and further compare it with the standard digital model file of the casting to obtain the final contour information and position information of the parts to be cleaned; complete the visual measurement of the casting to be cleaned.

[0060] Specifically, the model obtained by PCL processing of the point cloud is further compared with the STL digital model file, and the final difference is the part of the impurity to be cleaned.

[0061] Preferably, the 3D vision measurement method for a casting cleaning robot in this embodiment further includes: using the visualization interface of the Qt graphical user interface, based on the signal and slot mechanism of Qt software, to realize the visualization of casting shooting results from multiple angles, point cloud data from multiple angles, complete point cloud data, regional point cloud data, contour information and position information of the parts to be cleaned in each region, the model of the casting, and the final visualization of the impurities to be cleaned.

[0062] Specifically, this embodiment utilizes the Qt graphical user interface application development framework to design a visual interface that integrates the PCL function for processing point cloud data. Through the Qt software's signal and slot mechanism, clicking the function buttons on the interface can complete operations such as calling the camera, processing point clouds, and visualizing results.

[0063] Specifically, the human-computer interaction interface module is designed using the Qt graphical user interface application development framework to obtain a visual interface. It integrates the PCL function to process point cloud data. Through the "signal and slot" mechanism of Qt software, clicking the function buttons on the interface can complete operations such as calling the camera, point cloud processing, and result visualization.

[0064] Specifically, the software used in this embodiment was developed under the Windows 10 system. Programming and the design of the host computer software interface were carried out using C++ and Qt in Visual Studio 2015. The OpenCV vision open-source library and the PCL point cloud library were combined to calibrate the 3D binocular camera, process images, and process the 3D point cloud data of the casting. 3D vision and point cloud data processing are used to achieve rapid and accurate measurement of the parts of the casting to be cleaned. In this embodiment, the 3D binocular camera, digital grating projector, host computer, and 3D vision measurement software used are all existing products in the field; this embodiment only uses them to implement the proposed 3D vision measurement method.

[0065] This embodiment also discloses a measurement system for a 3D vision measurement method for a casting cleaning robot, including a data acquisition module, a point cloud data processing module, and a human-machine interface module; such as Figure 2 ,

[0066] The data acquisition module includes a 3D binocular camera and a digital grating projector, which are used in conjunction with the positioner of the casting cleaning robot to take pictures of the casting to be cleaned from multiple angles using a relative motion path, and to obtain point cloud data of the casting to be cleaned from multiple angles.

[0067] Specifically, the data acquisition module is used to adjust the pose of the casting to be cleaned by the positioner of the casting cleaning robot, so as to assist the 3D binocular camera to take multi-angle pictures of the casting to be cleaned and obtain multi-angle point cloud data of the casting to be cleaned.

[0068] The point cloud data processing module is used to perform point cloud registration based on point cloud data from multiple angles to obtain complete point cloud data of the casting; the complete point cloud data is filtered and divided into six regions to obtain point cloud data for six regions; the parts to be cleaned in the point cloud data of the six regions are identified and measured to obtain the contour information and position information of the parts to be cleaned in each region; surface reconstruction is performed based on the contour information of the parts to be cleaned in each region to obtain a model of the casting to be cleaned; and the model is further compared with the standard digital model file of the casting to obtain the final contour information and position information of the parts to be cleaned; the visual measurement of the casting to be cleaned is completed.

[0069] The human-computer interaction interface module is used to utilize the visualization interface of the Qt graphical user interface, based on the signal and slot mechanism of Qt software, to realize the visualization of casting shooting results from multiple angles, point cloud data from multiple angles, complete point cloud data, regional point cloud data, contour information and position information of the parts to be cleaned in each region, casting model, and finally the visualization of the impurities to be cleaned.

[0070] Preferably, the 3D binocular camera and digital grating projector are disposed inside the encapsulation housing for encapsulating and protecting the 3D binocular camera and digital grating projector.

[0071] Specifically, to adapt to the working environment of cleaning large castings, a metal casing is used to encapsulate the 3D binocular camera and digital grating projector; the metal casing plays a protective role and greatly extends the service life of the binocular camera equipment.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A 3D vision measurement method for a casting cleaning robot, characterized in that, Includes the following steps, Step 1: Install the casting to be cleaned of impurities onto the positioner fixture of the casting cleaning robot, so that the casting to be cleaned of impurities is fixed at a first angle; the first angle is the angle at which the casting to be cleaned of impurities is positioned when it is installed on the positioner. Step 2: Use Zhang Zhengyou's calibration method to complete the hand-eye calibration of the casting cleaning robot and the 3D binocular camera; Step 3: Use a digital grating projector and the 3D binocular camera to take pictures of the casting to be cleaned from the first angle to obtain point cloud data from the first angle, and record the relative motion path used by the 3D binocular camera relative to the casting to be cleaned when taking pictures. Step 4: Adjust the position of the casting to be cleaned by the positioner of the casting cleaning robot, so that the casting to be cleaned is fixed at the 2nd angle, the 3rd angle, ..., the nth angle; the 1st angle, the 2nd angle, ..., the nth angle are all different. Step 5: Based on the relative motion path, a digital grating projector and the 3D binocular camera are used to take pictures of the casting to be cleaned of impurities at the 2nd angle, the 3rd angle, ..., the nth angle, respectively, so as to obtain point cloud data at the 2nd angle, the 3rd angle, ..., the nth angle; Step 6: Based on the point cloud data of the first angle, the second angle, the third angle, ..., the nth angle, perform point cloud registration to obtain the complete point cloud data of the casting to be cleaned of impurities. Step 7: Filter the complete point cloud data and divide the filtered complete point cloud data into six regions to obtain point cloud data for six regions. Identify and measure the parts to be cleaned in the point cloud data of the six regions respectively, and obtain the contour information and position information of the parts to be cleaned in each region. Step 8: Based on the contour information of the part to be cleaned corresponding to each region, perform surface reconstruction to obtain the model of the casting to be cleaned. Then, compare the model of the casting to be cleaned with the standard digital model file of the casting to obtain the final contour information and position information of the part to be cleaned. Based on the final contour information and position information of the part to be cleaned, control the casting robot to clean the casting to be cleaned and complete the visual measurement of the casting to be cleaned.

2. The 3D vision measurement method for a casting cleaning robot according to claim 1, characterized in that, The method for dividing the filtered complete point cloud data into six regions is as follows: 1) Obtain the two points with the largest distance from the complete point cloud data and connect them to form a line segment; 2) Divide the line segment into three equal parts by taking two equal division points on the line segment; 3) Obtain the vertical plane that passes through the dividing point and is perpendicular to the line segment; 4) Obtain the plane passing through the line segment, and the plane and the vertical plane divide the complete point cloud data into six regions, such that the reference feature of the casting to be cleaned of impurities is located in any of the six regions.

3. The 3D vision measurement method for a casting cleaning robot according to claim 1, characterized in that, In step 3, the method for obtaining the point cloud data at the first angle is as follows: 1) Using OpenCV technology, a digital raster projector is used to project a structural light pattern onto the casting to be cleaned of impurities; 2) Use a 3D vision camera to photograph the casting to be cleaned of impurities, which has a structured light pattern projected onto it. Based on the outline of the casting to be cleaned of impurities, obtain the three-dimensional surface information of the casting to be cleaned of impurities, and then obtain the point cloud data of the casting to be cleaned of impurities.

4. The 3D vision measurement method for a casting cleaning robot according to claim 1, characterized in that, Step 5 is followed by: using the Qt graphical user interface and based on the Qt software's signal and slot mechanism, to realize the casting imaging results at the 1st, 2nd, 3rd, ..., nth angles, point cloud data at the 1st, 2nd, 3rd, ..., nth angles, complete point cloud data, regional point cloud data, contour information and position information of the parts to be cleaned in each region, the casting model, and the final visualization of the impurities to be cleaned.

5. The measurement system for a 3D vision measurement method for a casting cleaning robot according to any one of claims 1-4, characterized in that, It includes a data acquisition module, a point cloud data processing module, and a human-computer interaction interface module; The data acquisition module includes a 3D binocular camera and a digital grating projector, which are used in conjunction with the positioner of the casting cleaning robot to take pictures of the casting at the first angle, the second angle, the third angle, ..., the nth angle using a relative motion path, and obtain point cloud data of the casting to be cleaned at the first angle, the second angle, the third angle, ..., the nth angle. The point cloud data processing module is used to perform point cloud registration based on point cloud data from the 1st angle, 2nd angle, 3rd angle, ..., nth angle to obtain complete point cloud data of the casting; to filter the complete point cloud data and divide the filtered complete point cloud data into six regions to obtain point cloud data of six regions; to identify and measure the parts to be cleaned in the point cloud data of the six regions respectively, and to obtain the contour information and position information of the parts to be cleaned in each region; to perform surface reconstruction based on the contour information of the parts to be cleaned in each region to obtain a model of the casting to be cleaned, and to further compare it with the standard digital model file of the casting to obtain the final contour information and position information of the parts to be cleaned; and to complete the visual measurement of the casting to be cleaned. The human-computer interaction interface module is used to utilize the visualization interface of the Qt graphical user interface, based on the signal and slot mechanism of Qt software, to realize the casting shooting results from the 1st angle, the 2nd angle, the 3rd angle, ..., the nth angle, the point cloud data from the 1st angle, the 2nd angle, the 3rd angle, ..., the nth angle, the complete point cloud data, the regional point cloud data, the contour information and position information of the parts to be cleaned in each region, the model of the casting, and the visualization of the final parts of the impurities to be cleaned.

Images