Three-coordinate measurement system and method based on three-dimensional visual guidance
Through three-dimensional visual guidance technology based on artificial intelligence, the automated measurement of the three-coordinate measuring machine is realized, which solves the problem that traditional three-coordinate measuring machine relies on human factors, and improves the intelligence and accuracy of measurement.
Patent Information
- Application Number
- CN202411284665.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-13
- Publication Date
- 2024-11-12
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The operation process of the traditional three-coordinate measuring machine relies on experienced operators, which makes the measurement time-consuming, susceptible to human factors, and has low degree of automation, making it difficult to adapt to the modern manufacturing industry's demand for rapid response and flexible production.
Adopting three-dimensional visual guidance technology based on artificial intelligence, through the embedding encoding, local and global correlation of three-dimensional point cloud data of workpieces, the multi-scale fusion features are used to intelligently identify the position and position data of the workpiece, and the coordinate system is built to realize the automation of the three-coordinate measuring machine. Measurement.
It reduces errors and uncertainties caused by human factors, adapts to different types and sizes of workpiece measurements, and improves the intelligence and accuracy of measurements.
Smart Images

Figure CN118936316A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of intelligent measurement, and more specifically, to a three-dimensional coordinate measurement system and method based on three-dimensional vision guidance. Background Art
[0002] With the rapid development of the manufacturing industry and the continuous innovation of technology, the industry's requirements for production efficiency and product accuracy have reached an unprecedented level. The three-dimensional coordinate measuring machine is a geometric measurement instrument that integrates multidisciplinary technologies and plays an important role in the field of industrial manufacturing. It can accurately measure various geometric dimensions and errors of workpieces, and has the advantages of high measurement accuracy, large measurement range, and wide application range. This kind of measurement equipment integrates mechanical, measurement and control technologies, and can provide accurate geometric measurement services for traditional industrial fields such as aviation, aerospace, automobiles, and ships.
[0003] Although coordinate measuring machines (CMMs) provide high-precision measurement solutions in the field of mechanical manufacturing, their traditional operating procedures have some defects and disadvantages. Specifically, the traditional CMM measurement process often relies on experienced operators to manually position and calibrate the workpiece to ensure that the workpiece is in the correct coordinate system before measurement. This method is not only time-consuming, but also easily affected by human factors, resulting in uncertainty and poor repeatability of the measurement results. Especially for workpieces with complex shapes or large sizes, manual positioning becomes more difficult, further increasing the difficulty and potential errors of measurement. In addition, the traditional three-coordinate measurement system has a low degree of automation and is difficult to adapt to the needs of modern manufacturing for rapid response and flexible production, limiting its widespread application in the field of intelligent manufacturing.
[0004] Therefore, an optimized three-coordinate measurement solution is desired. Summary of the invention
[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a three-dimensional vision-guided three-dimensional coordinate measurement system and method, which adopts artificial intelligence-based data processing technology to embed and encode the three-dimensional point cloud data of the workpiece, and respectively associates the embedded and encoded data locally and globally, so as to intelligently obtain the position data recognition results of the position data and posture data of the workpiece to be identified on the workbench according to the multi-scale fusion characteristics between the global and local three-dimensional point cloud data of the workpiece, and use these recognition results to construct the coordinate system of the workpiece to guide the three-coordinate measuring machine to automatically complete the measurement of the workpiece. In this way, it can adapt to the measurement of workpieces of different types and sizes, and can reduce the errors caused by human factors during the workpiece measurement process, thereby greatly improving the intelligence level of workpiece measurement.
[0006] According to one aspect of the present application, a three-dimensional coordinate measurement system based on three-dimensional vision guidance is provided, which includes:
[0007] A workpiece three-dimensional point cloud data acquisition module is used to obtain a workpiece three-dimensional point cloud of the workpiece to be identified collected by a binocular structured light 3D camera, wherein each data point in the workpiece three-dimensional point cloud includes an R value, a G value, a B value and a depth value;
[0008] A workpiece three-dimensional point cloud data embedding coding module is used to embed and code each data point in the workpiece three-dimensional point cloud to obtain a low-dimensional embedded coded workpiece three-dimensional point cloud;
[0009] A workpiece three-dimensional point cloud local association module is used to input the low-dimensional embedded coded workpiece three-dimensional point cloud into a point cloud local region implicit association feature extractor to obtain a workpiece three-dimensional point cloud local association feature map;
[0010] A workpiece three-dimensional point cloud global association module is used to input the workpiece three-dimensional point cloud local association feature map into a global point cloud association feature map extractor to obtain a workpiece three-dimensional point cloud global association feature map;
[0011] A workpiece three-dimensional point cloud associated multi-scale joint module is used to input the workpiece three-dimensional point cloud global associated feature map and the workpiece three-dimensional point cloud local associated feature map into an attention-driven feature joint saliency perception network to obtain a workpiece three-dimensional point cloud associated multi-scale joint perception feature map;
[0012] A position data recognition result generation module is used to obtain a position data recognition result based on the three-dimensional point cloud of the workpiece and the multi-scale joint perception feature map, wherein the position data recognition result includes the position data and posture data of the workpiece to be recognized on the workbench;
[0013] The workpiece automatic measurement module is used to construct a workpiece coordinate system based on the position data recognition result, and guide the three-dimensional coordinate measuring machine to complete the automatic measurement of the workpiece to be identified based on the workpiece coordinate system.
[0014] According to another aspect of the present application, a three-dimensional coordinate measurement method based on three-dimensional vision guidance is provided, which includes:
[0015] Acquire a three-dimensional point cloud of the workpiece to be identified collected by a binocular structured light 3D camera, wherein each data point in the three-dimensional point cloud of the workpiece includes an R value, a G value, a B value and a depth value;
[0016] Embedding and encoding each data point in the three-dimensional point cloud of the workpiece to obtain a low-dimensional embedded coded three-dimensional point cloud of the workpiece;
[0017] Inputting the low-dimensional embedded coded workpiece three-dimensional point cloud into a point cloud local region implicit correlation feature extractor to obtain a workpiece three-dimensional point cloud local correlation feature map;
[0018] Inputting the workpiece three-dimensional point cloud local correlation feature map into a global point cloud correlation feature map extractor to obtain a workpiece three-dimensional point cloud global correlation feature map;
[0019] Inputting the workpiece 3D point cloud global correlation feature map and the workpiece 3D point cloud local correlation feature map into an attention-driven feature joint saliency perception network to obtain a workpiece 3D point cloud correlation multi-scale joint perception feature map;
[0020] Based on the workpiece three-dimensional point cloud associated multi-scale joint perception feature map, a position data recognition result is obtained, wherein the position data recognition result includes the position data and posture data of the workpiece to be recognized on the workbench;
[0021] Based on the position data recognition result, a workpiece coordinate system is constructed, and based on the workpiece coordinate system, a three-dimensional coordinate measuring machine is guided to complete automatic measurement of the workpiece to be recognized.
[0022] Compared with the prior art, the three-dimensional vision-guided three-dimensional coordinate measurement system and method provided by the present application adopts artificial intelligence-based data processing technology to embed and encode the three-dimensional point cloud data of the workpiece, and respectively associates the embedded and encoded data locally and globally, so as to intelligently obtain the position data recognition results of the position data and posture data of the workpiece to be identified on the workbench according to the multi-scale fusion characteristics between the global and local three-dimensional point cloud data of the workpiece, and use these recognition results to construct the coordinate system of the workpiece to guide the three-dimensional coordinate measuring machine to automatically complete the measurement of the workpiece. In this way, it can adapt to the measurement of workpieces of different types and sizes, and can reduce the errors caused by human factors during the workpiece measurement process, thereby greatly improving the intelligent level of workpiece measurement. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:
[0024] Figure 1 4 is a system block diagram of a three-dimensional coordinate measurement system based on three-dimensional vision guidance according to an embodiment of the present application.
[0025] Figure 2 Schematic diagram of data flow of a three-dimensional coordinate measurement system based on three-dimensional vision guidance according to an embodiment of the present application.
[0026] Figure 3 It is a block diagram of a multi-scale joint module for associating a three-dimensional point cloud of a workpiece in a three-dimensional coordinate measurement system guided by three-dimensional vision according to an embodiment of the present application.
[0027] Figure 4 It is a flow chart of a three-coordinate measurement method based on three-dimensional vision guidance according to an embodiment of the present application. DETAILED DESCRIPTION
[0028] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.
[0029] Against the backdrop of rapid development of the manufacturing industry and continuous technological innovation, the industry's requirements for production efficiency and product accuracy have reached unprecedented heights. As a geometric measurement instrument that integrates multidisciplinary technologies, the three-dimensional coordinate measuring machine plays a vital role in the field of industrial manufacturing. It can accurately measure various geometric dimensions and errors of workpieces, and has the advantages of high measurement accuracy, large range, and wide applicability. This measuring equipment integrates mechanical, measurement and control technologies, and can provide accurate geometric measurement services for traditional industrial fields such as aviation, aerospace, automobiles, and ships.
[0030] However, although the three-dimensional coordinate measuring machine provides a high-level measurement solution in the field of mechanical manufacturing, the traditional operation process has certain defects and disadvantages. Specifically, the traditional three-dimensional coordinate measuring machine measurement process usually requires experienced operators to manually position and calibrate the workpiece to ensure that the workpiece is in the correct coordinate system before measurement. This method is not only time-consuming, but also easily affected by human factors, resulting in uncertainty and poor repeatability of the measurement results. Especially for workpieces with complex shapes or large sizes, manual positioning is more difficult, which further increases the difficulty and potential errors of measurement. In addition, the traditional three-dimensional coordinate measurement system has a low degree of automation, which is difficult to meet the needs of modern manufacturing for rapid response and flexible production, thus limiting its wide application in the field of intelligent manufacturing.
[0031] It should be understood that 3D vision guidance is an advanced technology that uses 3D image processing and analysis to achieve precise positioning and tracking of objects. This technology can identify and measure the size, shape and position of objects by capturing and analyzing image data in three-dimensional space, thereby providing key visual information for automation systems. In the field of measurement, the application of 3D vision guidance is crucial, which can improve the accuracy and efficiency of measurement, especially in complex industrial environments. For example, in the manufacturing industry, 3D vision guidance systems can be used for automated quality inspection to ensure that product dimensions meet design specifications by accurately measuring them. In addition, it can also be used for precise topographic mapping and structural analysis in the fields of construction and engineering, providing accurate data support for construction and design. The application of 3D vision guidance technology not only improves the accuracy of measurement, but also provides strong technical support for automated and intelligent production.
[0032] Based on this, the present application proposes a three-dimensional coordinate measurement solution based on three-dimensional vision guidance, which obtains the three-dimensional point cloud of the workpiece to be identified collected by a binocular structured light 3D camera, wherein each data point in the three-dimensional point cloud of the workpiece contains R value, G value, B value and depth value, and uses artificial intelligence-based data processing and analysis technology to embed the three-dimensional point cloud of the workpiece, and respectively associates the embedded encoded data locally and globally, so as to intelligently obtain the position data recognition result of the position data and posture data of the workpiece to be identified on the workbench according to the multi-scale fusion characteristics between the global and local three-dimensional point cloud data of the workpiece, and use these recognition results to construct the coordinate system of the workpiece to guide the three-dimensional coordinate measuring machine to automatically complete the measurement of the workpiece. In this way, the process of automatically acquiring and processing the three-dimensional point cloud data of the workpiece reduces the dependence on manual operation, thereby reducing the error and uncertainty caused by human factors. At the same time, it can adapt to workpieces of different types and sizes, solve the difficulty of manual positioning, and enable the system to maintain high precision when processing diversified workpieces, thereby comprehensively improving the degree of intelligence of measurement.
[0033] Figure 1 4 is a system block diagram of a three-dimensional coordinate measurement system based on three-dimensional vision guidance according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of a three-dimensional coordinate measurement system based on three-dimensional vision guidance according to an embodiment of the present application. Figure 1 and Figure 2As shown, in the three-dimensional vision-guided three-dimensional coordinate measurement system 100, it includes: a workpiece three-dimensional point cloud data acquisition module 110, which is used to obtain a workpiece three-dimensional point cloud of a workpiece to be identified collected by a binocular structured light 3D camera, wherein each data point in the workpiece three-dimensional point cloud includes an R value, a G value, a B value and a depth value; a workpiece three-dimensional point cloud data embedding coding module 120, which is used to embed and code each data point in the workpiece three-dimensional point cloud to obtain a low-dimensional embedded coded workpiece three-dimensional point cloud; a workpiece three-dimensional point cloud local association module 130, which is used to input the low-dimensional embedded coded workpiece three-dimensional point cloud into a point cloud local area implicit association feature extractor to obtain a workpiece three-dimensional point cloud local association feature map; a workpiece three-dimensional point cloud global association module 140, which is used to input the workpiece three-dimensional point cloud local association feature map into a global point cloud A correlation feature map extractor is used to obtain a global correlation feature map of a three-dimensional point cloud of a workpiece; a three-dimensional point cloud correlation multi-scale joint module 150 is used to input the global correlation feature map of the three-dimensional point cloud of the workpiece and the local correlation feature map of the three-dimensional point cloud of the workpiece into an attention-driven feature joint saliency perception network to obtain a three-dimensional point cloud correlation multi-scale joint perception feature map; a position data recognition result generation module 160 is used to obtain a position data recognition result based on the three-dimensional point cloud correlation multi-scale joint perception feature map of the workpiece, and the position data recognition result includes the position data and posture data of the workpiece to be identified on the workbench; a workpiece automatic measurement module 170 is used to construct a workpiece coordinate system based on the position data recognition result, and guide a three-coordinate measuring machine to complete the automatic measurement of the workpiece to be identified based on the workpiece coordinate system.
[0034] In an embodiment of the present application, the workpiece three-dimensional point cloud data acquisition module 110 is used to obtain the workpiece three-dimensional point cloud of the workpiece to be identified collected by the binocular structured light 3D camera, wherein each data point in the workpiece three-dimensional point cloud includes an R value, a G value, a B value and a depth value. It should be understood that the binocular structured light 3D camera is a three-dimensional imaging device that captures and analyzes by emitting structured light (usually stripe light or dot light) and using two cameras. It combines the advantages of structured light and stereo vision and can quickly acquire high-precision three-dimensional point cloud data. The three-dimensional point cloud data of the workpiece to be identified collected by the binocular structured light 3D camera provides the necessary geometric basis for the identification of the workpiece position and posture. By analyzing the collected workpiece three-dimensional point cloud data, the exact position and posture of the workpiece in three-dimensional space can be calculated. This information is the key to guiding the three-dimensional coordinate measuring machine to perform automatic measurement. In particular, in a specific implementable embodiment of the present application, the following steps are used to acquire the three-dimensional point cloud data of the workpiece to be identified: a. Start the structured light source to project a light pattern onto the surface of the workpiece; b. Use two cameras at the same time to capture the light image reflected from the surface of the workpiece; c. Process the captured image, identify the deformation of the light pattern, and calculate the three-dimensional coordinates (X, Y, Z) of each point; d. Combine all the calculated three-dimensional coordinates to form a complete three-dimensional point cloud data of the workpiece.
[0035] In the embodiment of the present application, the workpiece three-dimensional point cloud data embedding coding module 120 is used to embed and code each data point in the workpiece three-dimensional point cloud to obtain a low-dimensional embedded coded workpiece three-dimensional point cloud. Specifically, in the embodiment of the present application, the workpiece three-dimensional point cloud data embedding coding module is used to: use the point cloud embedding coding matrix to embed and code each data point in the workpiece three-dimensional point cloud to obtain the low-dimensional embedded coded workpiece three-dimensional point cloud. It should be understood that the original workpiece three-dimensional point cloud data contains a large amount of information and has a high dimension. If the original high-dimensional point cloud data is directly processed and analyzed, it will bring high computational complexity and memory overhead, and the workpiece three-dimensional point cloud data contains key feature information. In order to reduce the complexity of high-dimensional point cloud data while retaining key information, in the technical solution of the present application, the point cloud embedding coding matrix is used to embed and code each data point in the workpiece three-dimensional point cloud to obtain a low-dimensional embedded coded workpiece three-dimensional point cloud. It should be understood that the point cloud embedding coding matrix here is essentially an embedding coding matrix, that is, by embedding and encoding the individual data points through the point cloud embedding coding matrix, a large amount of point cloud data can be mapped to a low-dimensional space to be compressed into a more compact representation, while retaining the important information of the original data as much as possible for more effective analysis and processing.
[0036] In the embodiment of the present application, the workpiece three-dimensional point cloud local association module 130 is used to input the low-dimensional embedded coded workpiece three-dimensional point cloud into the point cloud local region implicit association feature extractor to obtain the workpiece three-dimensional point cloud local association feature map. Specifically, in the embodiment of the present application, the workpiece three-dimensional point cloud local association module is used to: input the low-dimensional embedded coded workpiece three-dimensional point cloud into the point cloud local region implicit association feature extractor based on the three-dimensional convolutional neural network model to obtain the workpiece three-dimensional point cloud local association feature map. It should be understood that the low-dimensional embedded coded workpiece three-dimensional point cloud contains rich local geometric information, such as edges, surfaces, etc., and the three-dimensional convolutional neural network can capture the local geometric features and shape information in the point cloud data, and can effectively model the spatial structure information of the point cloud data, which is very important for understanding the local structure and surface characteristics of the workpiece. Based on this, in the technical solution of the present application, the low-dimensional embedded coded workpiece three-dimensional point cloud is input into the point cloud local region implicit association feature extractor based on the three-dimensional convolutional neural network model to capture and mine the three-dimensional local feature information of the workpiece three-dimensional point cloud to obtain the workpiece three-dimensional point cloud local association feature map. In particular, the point cloud local area implicit correlation feature extractor based on the three-dimensional convolutional neural network model is essentially a three-dimensional convolutional neural network model.
[0037] In the embodiment of the present application, the workpiece three-dimensional point cloud global association module 140 is used to input the workpiece three-dimensional point cloud local association feature map into the global point cloud association feature map extractor to obtain the workpiece three-dimensional point cloud global association feature map. Specifically, in the embodiment of the present application, the workpiece three-dimensional point cloud global association module is used to: input the workpiece three-dimensional point cloud local association feature map into the global point cloud association feature map extractor based on the transformer structure to obtain the workpiece three-dimensional point cloud global association feature map. It should be understood that the three-dimensional point cloud data contains rich global association information, such as overall structure, topological relationship, etc., and the transformer (Transformer) structure can effectively capture the global dependency of the entire point cloud data, which is crucial for understanding the overall geometric structure and spatial layout of the workpiece. Based on this, in the technical solution of the present application, the workpiece three-dimensional point cloud local association feature map is input into the global point cloud association feature map extractor based on the transformer structure to effectively model the global correlation of the point cloud data, capture higher-level semantic features, and obtain the workpiece three-dimensional point cloud global association feature map.
[0038] In the embodiment of the present application, the workpiece three-dimensional point cloud associated multi-scale joint module 150 is used to input the workpiece three-dimensional point cloud global associated feature map and the workpiece three-dimensional point cloud local associated feature map into the attention-driven feature joint saliency perception network to obtain the workpiece three-dimensional point cloud associated multi-scale joint perception feature map. Accordingly, considering that the workpiece three-dimensional point cloud global associated feature map provides information about the overall shape and structure of the workpiece, the workpiece three-dimensional point cloud local associated feature map provides information about the details and tiny structures of the workpiece, and there is mutual dependence and association between the two, and in order to better interact with the two to jointly constitute a comprehensive description of the workpiece, in the technical solution of the present application, an attention-driven feature joint saliency perception network is introduced to perform joint perception processing on the workpiece three-dimensional point cloud global associated feature map and the workpiece three-dimensional point cloud local associated feature map to obtain the workpiece three-dimensional point cloud associated multi-scale joint perception feature map. It is worth mentioning that the attention-driven feature joint saliency perception network uses dual interactive attention to enhance the features of each scale, thereby improving the overall feature expression capability.
[0039] In detail, first, the global correlation feature map of the workpiece three-dimensional point cloud and the local correlation feature map of the workpiece three-dimensional point cloud are reshaped to obtain the shape reshaping matrix of the global correlation feature of the workpiece three-dimensional point cloud and the shape reshaping matrix of the local correlation feature of the workpiece three-dimensional point cloud. It should be understood that the shape reshaping can unify the format of the data, adapt it to the subsequent attention mechanism processing flow, and retain the integrity of the information of the features in space and channels. Then, the feature channel-by-channel interactive perception module is used to calculate the dependency of the two reshaped feature matrices, so as to reveal the intrinsic connection and mutual dependence between the three-dimensional point cloud features of the workpiece at different scales, which is crucial to understanding the complexity of the data, thereby generating the global-local dependency matrix of the workpiece three-dimensional point cloud and the local-global dependency matrix of the workpiece three-dimensional point cloud, providing accurate data for the subsequent attention-based optimization. Then, the above two dependency matrices are input into the random dropout module for pruning operation. In particular, random dropout can simulate the robustness of the network in the face of partial information loss, and can better handle noise and abnormal data while retaining more informative features. The pruning operation prunes unimportant connections across domains to simplify the network structure. Then, the pruned features are matrix-multiplied with the reshaped feature matrix to strengthen the role of important or more discriminative features in the model, and the dependency-optimized workpiece 3D point cloud local association feature matrix and the dependency-optimized workpiece 3D point cloud global association layer feature matrix are obtained. Finally, the optimized features are reshaped to restore them to the form of feature maps and weighted fused to obtain the workpiece 3D point cloud associated multi-scale joint perception feature map. In this way, the workpiece 3D point cloud multi-scale joint perception features that combine local details and global context can more effectively reflect the key attributes of the workpiece to form a more comprehensive understanding of the workpiece.
[0040] Specifically, Figure 3 FIG. 1 is a block diagram of a multi-scale joint module for associating a three-dimensional point cloud of a workpiece in a three-dimensional coordinate measurement system guided by three-dimensional vision according to an embodiment of the present application. Figure 3As shown, the workpiece three-dimensional point cloud association multi-scale joint module 150 includes: a workpiece three-dimensional point cloud association feature map shape reshaping unit 151, which is used to reshape the feature shape of the workpiece three-dimensional point cloud global association feature map and the workpiece three-dimensional point cloud local association feature map to obtain the workpiece three-dimensional point cloud global association feature shape reshaping matrix and the workpiece three-dimensional point cloud local association feature shape reshaping matrix; a workpiece three-dimensional point cloud global-local channel-by-channel interactive perception unit 152, which is used to input the workpiece three-dimensional point cloud global association feature shape reshaping matrix and the workpiece three-dimensional point cloud local association feature shape reshaping matrix into the feature channel-by-channel interactive perception module to obtain the workpiece three-dimensional point cloud global-local dependency matrix and the workpiece three-dimensional point cloud local-global dependency matrix; a workpiece three-dimensional point cloud dependency matrix random deactivation processing unit 153, which is used to input the workpiece three-dimensional point cloud global-local dependency matrix and the workpiece three-dimensional point cloud local-global dependency matrix into the random deactivation module to obtain the pruned workpiece three-dimensional point cloud global-local dependency matrix and the pruned workpiece three-dimensional point cloud local-global dependency matrix; The three-dimensional point cloud local dependency optimization unit 154 is used to perform dependency optimization on the workpiece three-dimensional point cloud local association feature shape reshaping matrix and the workpiece three-dimensional point cloud global association feature shape reshaping matrix based on the pruned workpiece three-dimensional point cloud global-local dependency matrix and the pruned workpiece three-dimensional point cloud local-global dependency matrix to obtain a dependency optimized workpiece three-dimensional point cloud local association feature matrix and a dependency optimized workpiece three-dimensional point cloud global association feature matrix; the dependency optimized workpiece three-dimensional point cloud association matrix shape reshaping unit 155 is used to perform feature shape reshaping on the dependency optimized workpiece three-dimensional point cloud global association feature matrix and the dependency optimized workpiece three-dimensional point cloud local association feature matrix to obtain an optimized workpiece three-dimensional point cloud global association feature map and an optimized workpiece three-dimensional point cloud local association feature map; the workpiece three-dimensional point cloud associated multi-scale joint perception feature generation unit 156 is used to calculate the weighted sum of the optimized workpiece three-dimensional point cloud global association feature map and the optimized workpiece three-dimensional point cloud local association feature map to obtain the workpiece three-dimensional point cloud associated multi-scale joint perception feature map.
[0041] More specifically, in an embodiment of the present application, the workpiece three-dimensional point cloud global-local channel-by-channel interactive perception unit is used to: calculate the matrix product between the transposed matrix of the workpiece three-dimensional point cloud global associated feature shape reshaping matrix and the workpiece three-dimensional point cloud local associated feature shape reshaping matrix to obtain a first related feature matrix; calculate the positional division of the scale of the first related feature matrix and the first related feature matrix to obtain a first related feature modulation matrix; calculate the matrix product between the transposed matrix of the workpiece three-dimensional point cloud local associated feature shape reshaping matrix and the workpiece three-dimensional point cloud global associated feature shape reshaping matrix to obtain a second related feature matrix; calculate the positional division of the scale of the second related feature matrix and the second related feature matrix to obtain a second related feature modulation matrix; input the first related feature modulation matrix and the second related feature modulation matrix into the Softmax function respectively to obtain the workpiece three-dimensional point cloud global-local dependency matrix and the workpiece three-dimensional point cloud local-global dependency matrix.
[0042] More specifically, in an embodiment of the present application, the workpiece three-dimensional point cloud local dependency optimization unit is used to: perform matrix multiplication of the pruned workpiece three-dimensional point cloud global-local dependency matrix with the workpiece three-dimensional point cloud local association feature shape reshaping matrix to obtain the dependency optimized workpiece three-dimensional point cloud local association feature matrix; perform matrix multiplication of the pruned workpiece three-dimensional point cloud local-global dependency matrix with the workpiece three-dimensional point cloud global association feature shape reshaping matrix to obtain the dependency optimized workpiece three-dimensional point cloud global association feature matrix.
[0043] In the embodiment of the present application, specifically, the workpiece three-dimensional point cloud associated multi-scale joint module is used to: input the workpiece three-dimensional point cloud global associated feature map and the workpiece three-dimensional point cloud local associated feature map into the attention-driven feature joint saliency perception network, and process them with the following joint saliency perception formula to obtain the workpiece three-dimensional point cloud associated multi-scale joint perception feature map; wherein the joint saliency perception formula is:
[0044] M a =Reshape(F a )
[0045] M b =Reshape(F b )
[0046]
[0047] W ab = Dropout(M ab )
[0048] W ba= Dropout(M ba )
[0049]
[0050] F a ′=Reshape(M a ′)
[0051] F b ′=Reshape(M b ′)
[0052] F c =αF a ′+βF b '
[0053] Among them, F a and F b are the global correlation feature map of the workpiece 3D point cloud and the local correlation feature map of the workpiece 3D point cloud, Reshape(·) is the shape reshaping operation, M a and M b are respectively the workpiece 3D point cloud global associated feature shape reshaping matrix and the workpiece 3D point cloud local associated feature shape reshaping matrix, is the matrix product, M a T and M b T M a and M b The transposed matrix of , S1 and S2 are the scales of the first correlation feature matrix and the second correlation feature matrix, Softmax(·) is the Softmax function, M ab and M ba is the global-local dependency matrix of the workpiece 3D point cloud and the local-global dependency matrix of the workpiece 3D point cloud, Dropout(·) is the pruning process, W ab and W ba are the pruned workpiece 3D point cloud global-local dependency matrix and the pruned workpiece 3D point cloud local-global dependency matrix, M b ' and M a 'respectively the local correlation feature matrix of the three-dimensional point cloud of the workpiece optimized by the dependency relationship and the global correlation feature matrix of the three-dimensional point cloud of the workpiece optimized by the dependency relationship, F a ' and F b ' are the global correlation feature map of the optimized workpiece 3D point cloud and the local correlation feature map of the optimized workpiece 3D point cloud, α and β are weight hyperparameters, F c A multi-scale joint perception feature map is associated with the three-dimensional point cloud of the workpiece.
[0054] In the embodiment of the present application, the position data recognition result generation module 160 and the workpiece automatic measurement module 170 are respectively used to obtain the position data recognition result including the position data and posture data of the workpiece to be identified on the workbench based on the multi-scale joint perception feature map associated with the three-dimensional point cloud of the workpiece, and to construct a workpiece coordinate system based on the position data recognition result, and guide the three-coordinate measuring machine to complete the automatic measurement of the workpiece to be identified based on the workpiece coordinate system. Specifically, in the embodiment of the present application, the position data recognition result generation module is used to: input the multi-scale joint perception feature map associated with the three-dimensional point cloud of the workpiece into the decoder-based position data identifier to obtain the position data recognition result, and the position data recognition result includes the position data and posture data of the workpiece to be identified on the workbench. That is, the workpiece three-dimensional point cloud associated multi-scale joint perception feature map obtained by multi-scale joint perception using the global associated feature map of the workpiece three-dimensional point cloud and the local associated feature map of the workpiece three-dimensional point cloud is decoded and processed, so as to intelligently obtain the position data recognition result of the position data and posture data of the workpiece to be identified on the workbench, and use these recognition results to construct the coordinate system of the workpiece to guide the three-dimensional coordinate measuring machine to automatically complete the measurement of the workpiece. In this way, the process of automatically acquiring and processing the workpiece three-dimensional point cloud data reduces the dependence on manual operation, thereby reducing the errors and uncertainties caused by human factors. At the same time, it can adapt to workpieces of different types and sizes, solve the difficulties of manual positioning, and enable the system to maintain high precision when processing diversified workpieces, thereby comprehensively improving the degree of intelligent measurement.
[0055] Preferably, considering that the workpiece three-dimensional point cloud global association feature map and the workpiece three-dimensional point cloud local association feature map respectively represent the global image semantic association features and local image semantic association features of the low-dimensional embedded coding workpiece three-dimensional point cloud, when performing attention-driven feature joint saliency perception, the joint saliency perception attention difference caused by the local-global distribution difference of the image semantics will lead to the local-global image semantic joint perception aggregation fine-grained alignment conflict, thereby causing the loss of aggregation key information, affecting the expression effect of the workpiece three-dimensional point cloud associated multi-scale joint perception feature map.
[0056] Based on this, in the preferred example, when the workpiece three-dimensional point cloud associated multi-scale joint perception feature map is input into the decoder-based position data identifier to obtain the position data identification result, the workpiece three-dimensional point cloud associated multi-scale joint perception feature map is optimized. Specifically, the optimization process includes the following steps: expanding the workpiece three-dimensional point cloud associated multi-scale joint perception feature map into a workpiece three-dimensional point cloud associated multi-scale joint perception feature vector; clustering all eigenvalues of the workpiece three-dimensional point cloud associated multi-scale joint perception feature vector based on the distance between eigenvalues, such as the difference absolute value distance, and determining the ratio of the number of features of the clustering features to the number of features of the workpiece three-dimensional point cloud associated multi-scale joint perception feature vector; arranging the clustering features into a workpiece three-dimensional point cloud associated multi-scale joint perception clustering vector, and dividing the second norm of the workpiece three-dimensional point cloud associated multi-scale joint perception clustering vector by the second norm of the workpiece three-dimensional point cloud associated multi-scale joint perception feature vector to obtain a first workpiece three-dimensional point cloud associated multi-scale joint perception clustering related weight value; and dividing the workpiece three-dimensional point cloud associated multi-scale joint perception clustering vector by the second norm of the workpiece three-dimensional point cloud associated multi-scale joint perception feature vector to obtain a first workpiece three-dimensional point cloud associated multi-scale joint perception clustering related weight value; A first power value of a norm of a quantity with the scale value as an exponent is divided by a second power value of a norm of the workpiece three-dimensional point cloud associated multi-scale joint perception feature vector with the scale value as an exponent to obtain a second workpiece three-dimensional point cloud associated multi-scale joint perception clustering related weight value; for each eigenvalue of the workpiece three-dimensional point cloud associated multi-scale joint perception clustering vector, multiply it by the reciprocal of the difference between the first and second workpiece three-dimensional point cloud associated multi-scale joint perception clustering related weight values to obtain the optimized eigenvalue of the workpiece three-dimensional point cloud associated multi-scale joint perception clustering vector; for each eigenvalue outside the cluster in the workpiece three-dimensional point cloud associated multi-scale joint perception feature vector, multiply it by the reciprocal of the sum of the first and second workpiece three-dimensional point cloud associated multi-scale joint perception clustering related weight values to obtain the optimized out-of-class eigenvalue of the workpiece three-dimensional point cloud associated multi-scale joint perception feature vector; the optimized eigenvalue of the workpiece three-dimensional point cloud associated multi-scale joint perception clustering vector and the optimized out-of-class eigenvalue of the workpiece three-dimensional point cloud associated multi-scale joint perception feature vector are combined to form an optimized workpiece three-dimensional point cloud associated multi-scale joint perception feature vector. Then, the optimized workpiece three-dimensional point cloud associated multi-scale joint perception feature vector is input into the decoder-based position data identifier to obtain the position data recognition result. Then, the optimized workpiece three-dimensional point cloud associated multi-scale joint perception feature vector is input into the decoder-based position data identifier to obtain the position data recognition result
[0057] That is, the optimization process is expressed as:
[0058]
[0059] Wherein, f is each eigenvalue of the multi-scale joint perception clustering vector associated with the three-dimensional point cloud of the workpiece, V n is the multi-scale joint perception feature vector associated with the workpiece three-dimensional point cloud, is the multi-scale joint perception clustering vector associated with the workpiece three-dimensional point cloud, n is the number of features of the multi-scale joint perception feature vector associated with the workpiece three-dimensional point cloud, k is the number of features of the multi-scale joint perception clustering vector associated with the workpiece three-dimensional point cloud, λ is the ratio of the number of features of the multi-scale joint perception clustering vector associated with the workpiece three-dimensional point cloud to the multi-scale joint perception feature vector associated with the workpiece three-dimensional point cloud, represents the clustering feature set corresponding to the multi-scale joint perception clustering vector associated with the three-dimensional point cloud of the workpiece, ||·||2 and They respectively represent the λth power of the second norm and the first norm of the vector, and f' is each eigenvalue of the optimized workpiece three-dimensional point cloud associated multi-scale joint perception feature vector.
[0060] Here, in order to avoid the loss of key suffix semantic information relative to the original feature set as a whole due to aggregation conflict caused by the multi-scale joint perception feature map associated with the workpiece three-dimensional point cloud being based on aggregated features, the clustering ratio of the number of features of the multi-scale joint perception clustering vector associated with the workpiece three-dimensional point cloud relative to the number of features of the multi-scale joint perception feature vector associated with the workpiece three-dimensional point cloud is used as a decision function to perform an adversarial decision on the set absolute representation of the one-norm of the multi-scale joint perception clustering vector associated with the workpiece three-dimensional point cloud and the multi-scale joint perception feature vector associated with the workpiece three-dimensional point cloud, and the adversarial decision on the set absolute representation of the one-norm of the multi-scale joint perception clustering vector associated with the workpiece three-dimensional point cloud and the multi-scale joint perception feature vector associated with the workpiece three-dimensional point cloud is respectively performed. The clustering intrinsic conflict representation of the two norms of the three-dimensional point cloud associated multi-scale joint perception feature vector is positively and negatively interacted to construct the optimized workpiece three-dimensional point cloud associated multi-scale joint perception feature vector based on the aggregated features and the original feature set as a whole. The solid alignment guardrail is thereby realized based on the aggregated risk transferability of the optimized workpiece three-dimensional point cloud associated multi-scale joint perception feature vector. The harmful information loss intention of the optimized workpiece three-dimensional point cloud associated multi-scale joint perception feature vector is mitigated, and the expression effect of the optimized workpiece three-dimensional point cloud associated multi-scale joint perception feature vector is improved, thereby improving the accuracy of the position data recognition result obtained by the workpiece three-dimensional point cloud associated multi-scale joint perception feature map input into the decoder-based position data identifier. In this way, the process of automatically acquiring and processing the workpiece three-dimensional point cloud data reduces the dependence on manual operation, thereby reducing the errors and uncertainties caused by human factors. At the same time, it can adapt to workpieces of different types and sizes, solve the difficulties of manual positioning, and enable the system to maintain high accuracy when processing diversified workpieces, thereby comprehensively improving the degree of intelligent measurement.
[0061] In summary, the three-dimensional vision-guided three-dimensional coordinate measurement system 100 based on the embodiment of the present application is explained, which uses artificial intelligence-based data processing technology to embed and encode the three-dimensional point cloud data of the workpiece, and respectively associates the embedded and encoded data locally and globally, so as to intelligently obtain the position data recognition results of the position data and posture data of the workpiece to be identified on the workbench according to the multi-scale fusion characteristics between the global and local three-dimensional point cloud data of the workpiece, and use these recognition results to construct the coordinate system of the workpiece to guide the three-dimensional coordinate measuring machine to automatically complete the measurement of the workpiece. In this way, it can adapt to the measurement of workpieces of different types and sizes, and can reduce the errors caused by human factors during the workpiece measurement process, thereby greatly improving the intelligence of workpiece measurement.
[0062] As described above, the three-dimensional vision-guided three-dimensional coordinate measurement system 100 according to the embodiment of the present application can be implemented in various terminal devices, such as a server for three-dimensional vision-guided three-dimensional coordinate measurement. In one example, the three-dimensional vision-guided three-dimensional coordinate measurement system 100 according to the embodiment of the present application can be integrated into the terminal device as a software module and / or a hardware module. For example, the three-dimensional vision-guided three-dimensional coordinate measurement system 100 can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the three-dimensional vision-guided three-dimensional coordinate measurement system 100 can also be one of the many hardware modules of the terminal device.
[0063] Alternatively, in another example, the three-dimensional vision-guided three-coordinate measurement system 100 and the terminal device may also be separate devices, and the three-dimensional vision-guided three-coordinate measurement system 100 may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.
[0064] Figure 4 FIG. 1 is a flow chart of a three-dimensional coordinate measurement method based on three-dimensional vision guidance according to an embodiment of the present application. Figure 4As shown, in the three-dimensional coordinate measurement method based on three-dimensional vision guidance, it includes: S110, obtaining a three-dimensional point cloud of the workpiece to be identified collected by a binocular structured light 3D camera, wherein each data point in the workpiece three-dimensional point cloud contains an R value, a G value, a B value and a depth value; S120, respectively embedding and encoding each data point in the workpiece three-dimensional point cloud to obtain a low-dimensional embedded coded workpiece three-dimensional point cloud; S130, inputting the low-dimensional embedded coded workpiece three-dimensional point cloud into a point cloud local area implicit association feature extractor to obtain a workpiece three-dimensional point cloud local association feature map; S140, inputting the workpiece three-dimensional point cloud local association feature map into a global point cloud association feature map extractor to obtain to the global correlation feature map of the workpiece three-dimensional point cloud; S150, inputting the global correlation feature map of the workpiece three-dimensional point cloud and the local correlation feature map of the workpiece three-dimensional point cloud into the attention-driven feature joint saliency perception network to obtain the workpiece three-dimensional point cloud associated multi-scale joint perception feature map; S160, based on the workpiece three-dimensional point cloud associated multi-scale joint perception feature map, obtaining a position data recognition result, the position data recognition result includes the position data and posture data of the workpiece to be identified on the workbench; S170, based on the position data recognition result, constructing a workpiece coordinate system, and guiding the three-dimensional coordinate measuring machine to complete the automatic measurement of the workpiece to be identified based on the workpiece coordinate system.
[0065] Here, those skilled in the art can understand that the specific operations of each step in the above three-dimensional vision-guided three-dimensional coordinate measurement method have been described in the above reference. Figures 1 to 3 It has been introduced in detail in the description of the three-dimensional vision-guided three-dimensional coordinate measurement system, and therefore, its repeated description will be omitted.
[0066] In summary, the three-dimensional vision-guided three-dimensional coordinate measurement method based on the embodiment of the present application is explained, which uses artificial intelligence-based data processing technology to embed and encode the three-dimensional point cloud data of the workpiece, and respectively associates the embedded and encoded data locally and globally, so as to intelligently obtain the position data recognition results of the position data and posture data of the workpiece to be identified on the workbench according to the multi-scale fusion characteristics between the global and local three-dimensional point cloud data of the workpiece, and use these recognition results to construct the coordinate system of the workpiece to guide the three-dimensional coordinate measuring machine to automatically complete the measurement of the workpiece. In this way, it can adapt to the measurement of workpieces of different types and sizes, and can reduce the errors caused by human factors during the workpiece measurement process, thereby greatly improving the intelligence level of workpiece measurement.
[0067] In the several embodiments provided in this application, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are only schematic, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.
[0068] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0069] In addition, each functional module in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of hardware plus software functional modules.
[0070] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and that the present application can be implemented in other specific forms without departing from the spirit or essential features of the present application.
Claims
1. A three-dimensional coordinate measurement system based on three-dimensional vision guidance, It is characterized in that include: A workpiece three-dimensional point cloud data acquisition module is used to obtain a workpiece three-dimensional point cloud of the workpiece to be identified collected by a binocular structured light 3D camera, wherein each data point in the workpiece three-dimensional point cloud includes an R value, a G value, a B value and a depth value; A workpiece three-dimensional point cloud data embedding coding module is used to embed and code each data point in the workpiece three-dimensional point cloud to obtain a low-dimensional embedded coded workpiece three-dimensional point cloud; A workpiece three-dimensional point cloud local association module is used to input the low-dimensional embedded coded workpiece three-dimensional point cloud into a point cloud local region implicit association feature extractor to obtain a workpiece three-dimensional point cloud local association feature map; A workpiece three-dimensional point cloud global association module is used to input the workpiece three-dimensional point cloud local association feature map into a global point cloud association feature map extractor to obtain a workpiece three-dimensional point cloud global association feature map; A workpiece three-dimensional point cloud associated multi-scale joint module is used to input the workpiece three-dimensional point cloud global associated feature map and the workpiece three-dimensional point cloud local associated feature map into an attention-driven feature joint saliency perception network to obtain a workpiece three-dimensional point cloud associated multi-scale joint perception feature map; A position data recognition result generation module is used to obtain a position data recognition result based on the three-dimensional point cloud of the workpiece and the multi-scale joint perception feature map, wherein the position data recognition result includes the position data and posture data of the workpiece to be recognized on the workbench; The workpiece automatic measurement module is used to construct a workpiece coordinate system based on the position data recognition result, and guide the three-dimensional coordinate measuring machine to complete the automatic measurement of the workpiece to be identified based on the workpiece coordinate system.
2. The three-dimensional coordinate measurement system based on three-dimensional vision guidance according to claim 1, It is characterized in that The workpiece three-dimensional point cloud data embedding coding module is used to: use the point cloud embedding coding matrix to embed and code each data point in the workpiece three-dimensional point cloud to obtain the low-dimensional embedded coding workpiece three-dimensional point cloud.
3. The three-dimensional coordinate measurement system based on three-dimensional vision guidance according to claim 2, It is characterized in that The workpiece three-dimensional point cloud local association module is used to: input the low-dimensional embedded coded workpiece three-dimensional point cloud into a point cloud local area implicit association feature extractor based on a three-dimensional convolutional neural network model to obtain the workpiece three-dimensional point cloud local association feature map.
4. The three-dimensional coordinate measurement system based on three-dimensional vision guidance according to claim 3, It is characterized in that The workpiece three-dimensional point cloud global association module is used to: input the workpiece three-dimensional point cloud local association feature map into the global point cloud association feature map extractor based on the converter structure to obtain the workpiece three-dimensional point cloud global association feature map.
5. The three-dimensional coordinate measurement system based on three-dimensional vision guidance according to claim 4, It is characterized in that The workpiece three-dimensional point cloud association multi-scale joint module includes: A workpiece three-dimensional point cloud associated feature map shape reshaping unit is used to perform feature shape reshaping on the workpiece three-dimensional point cloud global associated feature map and the workpiece three-dimensional point cloud local associated feature map to obtain a workpiece three-dimensional point cloud global associated feature shape reshaping matrix and a workpiece three-dimensional point cloud local associated feature shape reshaping matrix; A workpiece three-dimensional point cloud global-local channel-by-channel interactive perception unit is used to input the workpiece three-dimensional point cloud global associated feature shape reshaping matrix and the workpiece three-dimensional point cloud local associated feature shape reshaping matrix into a feature channel-by-channel interactive perception module to obtain a workpiece three-dimensional point cloud global-local dependency matrix and a workpiece three-dimensional point cloud local-global dependency matrix; A workpiece three-dimensional point cloud dependency matrix random deactivation processing unit is used to input the workpiece three-dimensional point cloud global-local dependency matrix and the workpiece three-dimensional point cloud local-global dependency matrix into a random deactivation module to obtain a pruned workpiece three-dimensional point cloud global-local dependency matrix and a pruned workpiece three-dimensional point cloud local-global dependency matrix; A workpiece three-dimensional point cloud local dependency optimization unit is used to perform dependency optimization on the workpiece three-dimensional point cloud local association feature shape reshaping matrix and the workpiece three-dimensional point cloud global association feature shape reshaping matrix based on the pruned workpiece three-dimensional point cloud global-local dependency matrix and the pruned workpiece three-dimensional point cloud local-global dependency matrix to obtain a dependency-optimized workpiece three-dimensional point cloud local association feature matrix and a dependency-optimized workpiece three-dimensional point cloud global association feature matrix; A dependency relationship optimized workpiece three-dimensional point cloud association matrix shape reshaping unit is used to perform feature shape reshaping on the dependency relationship optimized workpiece three-dimensional point cloud global association feature matrix and the dependency relationship optimized workpiece three-dimensional point cloud local association feature matrix to obtain an optimized workpiece three-dimensional point cloud global association feature map and an optimized workpiece three-dimensional point cloud local association feature map; The workpiece 3D point cloud associated multi-scale joint perception feature generation unit is used to calculate the weighted sum of the optimized workpiece 3D point cloud global association feature map and the optimized workpiece 3D point cloud local association feature map to obtain the workpiece 3D point cloud associated multi-scale joint perception feature map.
6. The three-dimensional coordinate measurement system based on three-dimensional vision guidance according to claim 5, It is characterized in that The workpiece three-dimensional point cloud global-local channel-by-channel interactive perception unit is used to: Calculating the matrix product between the transposed matrix of the workpiece three-dimensional point cloud global correlation feature shape reshaping matrix and the workpiece three-dimensional point cloud local correlation feature shape reshaping matrix to obtain a first correlation feature matrix; calculating a position-wise division of the first correlation feature matrix and the scale of the first correlation feature matrix to obtain a first correlation feature modulation matrix; Calculating the matrix product between the transposed matrix of the workpiece three-dimensional point cloud local associated feature shape reshaping matrix and the workpiece three-dimensional point cloud global associated feature shape reshaping matrix to obtain a second associated feature matrix; calculating a position-wise division of the scale of the second correlation feature matrix and the second correlation feature matrix to obtain a second correlation feature modulation matrix; The first correlation feature modulation matrix and the second correlation feature modulation matrix are respectively input into the Softmax function to obtain the workpiece three-dimensional point cloud global-local dependency matrix and the workpiece three-dimensional point cloud local-global dependency matrix.
7. The three-dimensional coordinate measurement system based on three-dimensional vision guidance according to claim 6, It is characterized in that The workpiece three-dimensional point cloud local dependency optimization unit is used to: Performing matrix multiplication of the pruned workpiece three-dimensional point cloud global-local dependency matrix and the workpiece three-dimensional point cloud local associated feature shape reshaping matrix to obtain the dependency optimized workpiece three-dimensional point cloud local associated feature matrix; The pruned workpiece three-dimensional point cloud local-global dependency matrix is matrix-multiplied with the workpiece three-dimensional point cloud global associated feature shape reshaping matrix to obtain the dependency-optimized workpiece three-dimensional point cloud global associated feature matrix.
8. The three-dimensional coordinate measurement system based on three-dimensional vision guidance according to claim 7, It is characterized in that The position data recognition result generation module is used to: input the workpiece three-dimensional point cloud associated multi-scale joint perception feature map into the decoder-based position data identifier to obtain the position data recognition result, and the position data recognition result includes the position data and posture data of the workpiece to be identified on the workbench.
9. A three-dimensional coordinate measurement method based on three-dimensional vision guidance, It is characterized in that include: Acquire a three-dimensional point cloud of the workpiece to be identified collected by a binocular structured light 3D camera, wherein each data point in the three-dimensional point cloud of the workpiece includes an R value, a G value, a B value and a depth value; Embedding and encoding each data point in the three-dimensional point cloud of the workpiece to obtain a low-dimensional embedded coded three-dimensional point cloud of the workpiece; Inputting the low-dimensional embedded coded workpiece three-dimensional point cloud into a point cloud local region implicit correlation feature extractor to obtain a workpiece three-dimensional point cloud local correlation feature map; Inputting the workpiece three-dimensional point cloud local correlation feature map into a global point cloud correlation feature map extractor to obtain a workpiece three-dimensional point cloud global correlation feature map; Inputting the workpiece 3D point cloud global correlation feature map and the workpiece 3D point cloud local correlation feature map into an attention-driven feature joint saliency perception network to obtain a workpiece 3D point cloud correlation multi-scale joint perception feature map; Based on the workpiece three-dimensional point cloud associated multi-scale joint perception feature map, a position data recognition result is obtained, wherein the position data recognition result includes the position data and posture data of the workpiece to be recognized on the workbench; Based on the position data recognition result, a workpiece coordinate system is constructed, and based on the workpiece coordinate system, a three-dimensional coordinate measuring machine is guided to complete automatic measurement of the workpiece to be recognized.
10. The three-dimensional coordinate measurement method based on three-dimensional vision guidance according to claim 9, It is characterized in that Each data point in the workpiece three-dimensional point cloud is embedded and encoded respectively to obtain a low-dimensional embedded coded workpiece three-dimensional point cloud, including: using a point cloud embedding coding matrix to embed and encode each data point in the workpiece three-dimensional point cloud to obtain the low-dimensional embedded coded workpiece three-dimensional point cloud.
Citation Information
Cited By
Health data management platform based on cloud computing
CN119742048A
Large-scale point cloud data feature extraction method and system based on artificial intelligence
CN119785045A
Special-shaped curved surface weight reduction processing method based on binocular structured light 3D measurement
CN120095722A
Method for surface weight reduction machining of special-shaped curved surfaces based on binocular structured light 3D measurement
CN120095722B
Three-coordinate measurement method and system based on virtual simulation
CN120953517A
