Visual identification automatic correction assembly system for bearing retainer

Abstract

The invention provides a visual identification automatic correction assembly system for a bearing retainer, and relates to the technical field of data processing, and the method comprises the steps: collecting images of the retainer and an assembly station through a camera, carrying out the preprocessing of the images, obtaining multi-angle image data, extracting the feature points of the retainer and the assembly station from the multi-angle image data, and obtaining the feature points of the retainer and the assembly station; analyzing whether the retainer has deviation or not, constructing a three-dimensional assembly model according to the assembly stations, establishing a three-dimensional coordinate system, analyzing the positions of the feature points, generating a three-dimensional structure, constructing an entity assembly model, ensuring that the assembly stations of the entity assembly model and the three-dimensional assembly model coincide, and determining the coordinates of the entity assembly model; correction data are obtained by analyzing the deviation between the retainer and the assembly station, the retainer is corrected and fed back to the image processing module, and whether the retainer has deviation or not is detected. According to the invention, whether the bearing retainer has deviation or not can be visually identified, and the bearing retainer can be automatically corrected.

Description

Technical Field

[0001] The invention relates to the technical field of data processing, in particular to a visual recognition automatic correction and assembly system for a bearing retainer. Background Art

[0002] In the existing bearing cage assembly system, visual recognition technology is mainly used to identify the relative position of the cage and the bearing to achieve accurate assembly of the components. However, since the existing visual recognition technology is based on simple image feature recognition, it may not be able to identify slight deviations in the processed image in actual applications, and there may be problems with insufficient accuracy and stability.

[0003] For example, when the retainer reaches the assembly position, the position of the retainer often shifts slightly. Traditional visual recognition may not be able to accurately detect such slight shifts, resulting in deviations in the position of the retainer during the actual assembly process, causing the assembled bearing to wear and vibrate during operation, shortening the service life of the bearing. Summary of the invention

[0004] The purpose of the present invention is to provide a bearing retainer visual recognition automatic correction assembly system, aiming to solve the problems mentioned in the background technology.

[0005] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0006] A bearing retainer visual recognition automatic correction assembly system, the system comprising:

[0007] An image acquisition module is used to acquire images of the retainer and the assembly station through a camera to obtain multi-view image data;

[0008] An image processing module is used to pre-process the multi-view image data, remove noise and enhance edge features to obtain multi-angle image data;

[0009] The feature point module is used to extract the feature points of the cage and the assembly station according to the multi-angle image data, and analyze whether there is a deviation in the cage according to the feature points to obtain a feature point data set;

[0010] A three-dimensional coordinate system module is used to construct a three-dimensional assembly model according to the assembly station, and to establish a three-dimensional coordinate system according to the three-dimensional assembly model;

[0011] The model building module is used to analyze the positions of feature points in images at different angles, generate a three-dimensional structure, fill the three-dimensional structure, and build a solid assembly model;

[0012] A coordinate analysis module is used to superimpose the physical assembly model with the three-dimensional assembly model to ensure that the assembly stations of the physical assembly model and the three-dimensional assembly model coincide with each other, thereby obtaining a superimposed model, and determining the coordinates of the physical assembly model based on the superimposed model to obtain three-dimensional coordinate data;

[0013] A deviation detection module is used to analyze the deviation between the retainer and the assembly station according to the three-dimensional coordinate data to obtain correction data, the correction data including the center deviation value and the angle deviation value;

[0014] An automatic correction module is used to correct the holder according to the correction data to obtain a correction result image;

[0015] The correction inspection module is used to feed back the correction result image to the image processing module, and detect whether there is deviation in the cage through the feature point module. When there is deviation in the cage, the above steps are repeated. When there is no deviation in the cage, the assembly is completed.

[0016] Furthermore, the image processing module includes:

[0017] An image denoising unit, used to remove noise interference in the image according to the multi-view image data to obtain clear image data;

[0018] The edge enhancement unit is used to detect the gradient of the image edge in the horizontal and vertical directions according to the clear image data, determine the area with rapid gradient changes, and increase its pixel value to obtain edge enhanced image data;

[0019] The scale analysis unit is used to analyze and capture the detail features of the image at different scales based on the edge enhanced image data to obtain multi-angle image data.

[0020] Furthermore, the feature point module includes:

[0021] An image analysis unit, used to analyze the structure of the cage and the assembly station according to multi-angle image data to obtain a characteristic area;

[0022] A feature analysis unit is used to extract the center point and contour point of the cage and the assembly station according to the feature area, and obtain the feature point of the cage and the feature point of the assembly station;

[0023] The deviation analysis unit is used to analyze the relative position between the characteristic points of the cage and the characteristic points of the assembly station and determine whether there is a deviation in the cage based on the relative position. When the result is no, the assembly is completed. When the result is yes, the characteristic point data set is obtained based on the relative position between the characteristic points of the cage and the characteristic points of the assembly station, and the automatic correction function is turned on.

[0024] Furthermore, the three-dimensional coordinate system module includes:

[0025] The assembly model unit is used to construct a three-dimensional model according to the assembly station to obtain a three-dimensional assembly model;

[0026] The coordinate system unit is used to establish a three-dimensional coordinate system based on the three-dimensional assembly model, with the center point of the three-dimensional assembly model as the origin, the horizontal straight line passing through the center point of the three-dimensional assembly model as the x-axis, the vertical straight line passing through the center point of the three-dimensional assembly model as the y-axis, and the straight line passing through the center point of the three-dimensional assembly model and perpendicular to the plane formed by the x-axis and the y-axis as the z-axis.

[0027] Furthermore, the model building module includes:

[0028] A viewing angle analysis unit, used to analyze the viewing angle errors of the feature points of the cage in images at different angles according to the feature point data set, and obtain a parallax adjustment factor;

[0029] A structure generation unit, used for adjusting the position of the characteristic points of the cage according to the parallax adjustment factor, and spatially connecting the adjusted characteristic points of the cage and the characteristic points of the assembly station to generate a three-dimensional structure;

[0030] A detail filling unit is used to obtain a preliminary solid assembly model by adding filling points in the space area between the feature points in the three-dimensional structure;

[0031] The model optimization unit is used to smooth the preliminary solid assembly model, eliminate the abrupt edges on the surface of the preliminary solid assembly model, and obtain the solid assembly model.

[0032] Furthermore, the calculation formula of the parallax adjustment factor is: ;

[0033] in, is the parallax adjustment factor, is the number of cameras, and is the index of the camera, and , For the Camera and The angle between the imaging points of the cameras, For the The focal length of the camera, For the The focal length of the camera, For the The distance from the camera to the holder and assembly station, For the The distance from the camera to the holder and assembly station, is a constant.

[0034] Furthermore, the coordinate analysis module includes:

[0035] A proportion adjustment unit is used to adjust the physical assembly model to ensure that the proportion of the physical assembly model is consistent with that of the three-dimensional assembly model, thereby obtaining model data;

[0036] A model matching unit is used to superimpose the physical assembly model and the three-dimensional assembly model according to the model data, so as to ensure that the assembly stations of the physical assembly model and the three-dimensional assembly model coincide with each other, and obtain a superimposed model;

[0037] The entity coordinate unit is used to identify the coordinates of the entity assembly model in the three-dimensional coordinate system according to the superposition model to obtain the three-dimensional coordinate data.

[0038] Furthermore, the deviation detection module includes:

[0039] The center deviation unit is used to analyze whether the center point coordinates of the solid assembly model are located at the origin of the three-dimensional coordinate system according to the three-dimensional coordinate data. When the result is yes, the solid assembly model does not have a center deviation. When the result is no, the solid assembly model has a center deviation, and the center deviation value is calculated according to the center point coordinates of the solid assembly model.

[0040] The angle deviation unit is used to analyze whether there is an angle between the retaining frame and the assembly station in the superposition model according to the three-dimensional coordinate data. When the result is no, there is no angle deviation in the physical assembly model. When the result is yes, there is an angle deviation in the physical assembly model, and the angle deviation value is calculated.

[0041] Furthermore, the calculation formula of the angle deviation value is: ;

[0042] in, is the angle deviation value, They are respectively the entity assembly model The horizontal, vertical and vertical coordinates of the contour points are They are the three-dimensional assembly models The horizontal, vertical and vertical coordinates of the contour points are Assemble the model for the entity, For a three-dimensional assembly model, is the total number of contour points, is the index of the contour point, For the The coefficients of the contour points.

[0043] Furthermore, the automatic correction module includes:

[0044] A center deviation analysis unit, used to determine a translation direction vector according to a center deviation value, and normalize the translation direction vector to obtain a translation unit vector;

[0045] A center correction unit is used to determine the translation amount of the cage in the three-dimensional coordinate system according to the center deviation value and the translation unit vector, and obtain the translation correction vector;

[0046] An angle deviation analysis unit, used to determine the rotation direction vector according to the angle deviation value, and normalize the rotation direction vector to obtain a rotation unit vector;

[0047] An angle correction unit is used to determine the rotation amount of the cage on each axis according to the angle deviation value and the rotation unit vector to obtain a rotation correction vector;

[0048] The correction execution unit is used to perform translation correction and rotation correction on the retaining frame according to the translation correction vector and the rotation correction vector to obtain a correction result image.

[0049] The above solution of the present invention includes at least the following beneficial effects:

[0050] The present invention uses denoising and edge enhancement, and the image processing module can effectively improve the quality of the image, thereby extracting clearer edge features. A common problem in the prior art is that noise affects the recognition results, and this function eliminates noise interference through a variety of preprocessing methods to ensure the reliability of subsequent analysis. At the same time, edge feature enhancement increases the value of the pixel gradient change area, making it easier for the system to identify complex structures. In particular, during the assembly process, when there are tiny bumps on the surface of the cage or workstation, it ensures that these detailed features can be accurately captured, greatly improving the robustness and detection accuracy of the system.

[0051] The present invention significantly improves the ability to judge the position and direction of the cage by extracting feature points from multi-angle images and analyzing their deviations. Unlike traditional methods, this function not only detects the position of feature points, but also analyzes the relative relationship between feature points, so as to more accurately evaluate the matching degree between the cage and the assembly station. This feature point analysis method is particularly suitable for processing cages with complex shapes, so that the system can be compatible with components of various specifications, thereby improving the breadth of application. The feature point-based analysis method also reduces the dependence on physical alignment accuracy, indirectly improving production efficiency.

[0052] The present invention provides a unified spatial reference for the entire system by constructing a three-dimensional coordinate system based on the assembly station. Compared with two-dimensional plane recognition, the three-dimensional coordinate system can more accurately reflect the actual position and posture of the retaining bracket, thereby improving the accuracy of correction. The introduction of the three-dimensional coordinate system allows the entire assembly process to be dynamically adjusted in space, avoiding the accumulation of deviations caused by the lack of spatial reference in traditional methods. This three-dimensional space calibration method also has good scalability and can adapt to different assembly scenarios, providing technical support for the functional upgrade and expansion of the system.

[0053] The present invention truly restores the structure and relationship of the retaining frame and the assembly station by generating a three-dimensional solid model. Compared with the simple image superposition method, the three-dimensional model can accurately calculate the position and shape of the retaining frame. Especially when there is parallax in multi-view data, the error is eliminated by filling points and optimization processing, thereby improving the accuracy of the solid model. This process provides more reliable data support for deviation analysis and correction, and also improves the adaptability of the entire system to complex assembly tasks.

[0054] The present invention automatically corrects the retaining frame through deviation data, thus overcoming the limitations of manual correction in traditional methods. Through the combined operation of translation and rotation, it ensures that the position and posture of the retaining frame in space fully meet the design requirements. The correction process is combined with a feedback mechanism to continuously repeat the correction steps until the goal of precise assembly is achieved. This adaptive correction capability not only improves assembly efficiency, but also reduces errors that may be introduced by human operation, greatly enhancing the intelligence level and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a flowchart of a bearing retainer visual recognition automatic correction assembly system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0057] like Figure 1 As shown, an embodiment of the present invention provides a bearing retainer visual recognition automatic correction assembly system, the system comprising:

[0058] An image acquisition module is used to acquire images of the retainer and the assembly station through a camera to obtain multi-view image data;

[0059] An image processing module is used to pre-process the multi-view image data, remove noise and enhance edge features to obtain multi-angle image data;

[0060] The feature point module is used to extract the feature points of the cage and the assembly station according to the multi-angle image data, and analyze whether there is a deviation in the cage according to the feature points to obtain a feature point data set;

[0061] A three-dimensional coordinate system module is used to construct a three-dimensional assembly model according to the assembly station, and to establish a three-dimensional coordinate system according to the three-dimensional assembly model;

[0062] The model building module is used to analyze the positions of feature points in images at different angles, generate a three-dimensional structure, fill the three-dimensional structure, and build a solid assembly model;

[0063] A coordinate analysis module is used to superimpose the physical assembly model with the three-dimensional assembly model to ensure that the assembly stations of the physical assembly model and the three-dimensional assembly model coincide with each other, thereby obtaining a superimposed model, and determining the coordinates of the physical assembly model based on the superimposed model to obtain three-dimensional coordinate data;

[0064] A deviation detection module is used to analyze the deviation between the retainer and the assembly station according to the three-dimensional coordinate data to obtain correction data, the correction data including the center deviation value and the angle deviation value;

[0065] An automatic correction module is used to correct the holder according to the correction data to obtain a correction result image;

[0066] The correction inspection module is used to feed back the correction result image to the image processing module, and detect whether there is deviation in the cage through the feature point module. When there is deviation in the cage, the above steps are repeated. When there is no deviation in the cage, the assembly is completed.

[0067] In an embodiment of the present invention, images of the retaining frame and the assembly station are collected by a camera to obtain multi-view image data, acquire comprehensive visual information, and provide high-quality, multi-angle original data for subsequent modules; the multi-view image data is preprocessed to remove noise and enhance edge features to obtain multi-angle image data, thereby improving the availability and accuracy of the image, thereby ensuring that subsequent feature extraction and deviation detection are more accurate; based on the multi-angle image data, feature points of the retaining frame and the assembly station are extracted, and based on the feature points, whether there is any deviation in the retaining frame is analyzed to obtain a feature point data set, which provides accurate geometric information for the subsequent construction of a three-dimensional coordinate system and deviation correction, and by analyzing the deviation between the retaining frame and the station, lays the foundation for the system's automatic correction function.

[0068] A three-dimensional assembly model is constructed according to the assembly station, and a three-dimensional coordinate system is established based on the three-dimensional assembly model to provide a unified spatial reference for the assembly process, so that the position and direction of all components can be described in the form of a three-dimensional coordinate system; the position of feature points in images at different angles is analyzed to generate a three-dimensional structure, and the three-dimensional structure is filled to construct a solid assembly model, and the real spatial relationship between the retaining frame and the assembly station is restored from multi-angle images to provide an accurate geometric reference for deviation detection and automatic correction, while improving the reliability of correction through the optimization of the solid model; the solid assembly model is superimposed on the three-dimensional assembly model to ensure that the assembly stations of the solid assembly model and the three-dimensional assembly model coincide with each other to obtain a superimposed model, and the coordinates of the solid assembly model are determined based on the superimposed model to obtain three-dimensional coordinate data, and the position and direction of the retaining frame are accurately located through superimposed analysis to provide a basis for the deviation detection module, while ensuring that the corrected retaining frame can perfectly fit the assembly station.

[0069] According to the three-dimensional coordinate data, the deviation between the cage and the assembly station is analyzed to obtain the correction data, which provides detailed deviation data for the subsequent automatic correction module. At the same time, accurate deviation detection is used to ensure that no cumulative error occurs during the assembly process. The cage is corrected according to the correction data to obtain a correction result image, and automatic alignment of the cage is achieved, greatly improving assembly efficiency and accuracy. The correction result image is fed back to the image processing module, and the feature point module is used to detect whether there is any deviation in the cage. The stability and intelligence level of the system are further enhanced through the cyclic correction mechanism.

[0070] The image acquisition module is used to acquire images of the retainer and the assembly station through a camera to obtain multi-view image data, specifically including:

[0071] First, locate the installation position of the camera around the assembly station to ensure that it can cover the entire area of ​​the retaining frame and the station. Then the camera continuously captures images at a preset frame rate to obtain multi-view image data, and transmits the multi-view image data to the image processing module in real time through the data transmission interface. This multi-view acquisition method can effectively avoid the information blind spots caused by a single view and improve the system's adaptability to complex assembly environments.

[0072] In a preferred embodiment of the present invention, the image processing module includes:

[0073] An image denoising unit, used to remove noise interference in the image according to the multi-view image data to obtain clear image data;

[0074] The edge enhancement unit is used to detect the gradient of the image edge in the horizontal and vertical directions according to the clear image data, determine the area with rapid gradient changes, and increase its pixel value to obtain edge enhanced image data;

[0075] The scale analysis unit is used to analyze and capture the detail features of the image at different scales based on the edge enhanced image data to obtain multi-angle image data.

[0076] In an embodiment of the present invention, based on multi-view image data, noise interference in the image is removed to obtain clear image data, improve the clarity and stability of the image, and provide high-quality data input for the subsequent edge enhancement unit and feature extraction module, thereby reducing the risk of false detection and missed detection; based on the clear image data, the gradient of the image edge in the horizontal and vertical directions is detected, the rapidly changing area of ​​the gradient is determined, and its pixel value is increased to obtain edge enhanced image data, and by strengthening the component boundaries and feature points, the system can more accurately capture the spatial relationship between the retaining frame and the assembly station; based on the edge enhanced image data, the detailed features of the image at different scales are analyzed and the detailed features are captured to obtain multi-angle image data, thereby improving the robustness and flexibility of the system, so that it can maintain a high level of recognition accuracy when the image resolution changes or the target size changes significantly.

[0077] The scale analysis unit is used to analyze the detailed features of the image at different scales and capture the detailed features according to the edge enhanced image data to obtain multi-angle image data, specifically including:

[0078] First, the edge-enhanced image data is scaled down to form a multi-level image through multi-resolution image technology. Then, local detail features are extracted from each level of the image, and the scale invariance of the feature points is calculated to ensure that the feature points remain consistent at different resolutions. Finally, the multi-level feature points are fused to generate multi-angle image data containing rich detail information.

[0079] In a preferred embodiment of the present invention, the feature point module includes:

[0080] An image analysis unit, used to analyze the structure of the cage and the assembly station according to multi-angle image data to obtain a characteristic area;

[0081] A feature analysis unit is used to extract the center point and contour point of the cage and the assembly station according to the feature area, and obtain the feature point of the cage and the feature point of the assembly station;

[0082] The deviation analysis unit is used to analyze the relative position between the characteristic points of the cage and the characteristic points of the assembly station and determine whether there is a deviation in the cage based on the relative position. When the result is no, the assembly is completed. When the result is yes, the characteristic point data set is obtained based on the relative position between the characteristic points of the cage and the characteristic points of the assembly station, and the automatic correction function is turned on.

[0083] In an embodiment of the present invention, the structures of the retainer and the assembly station are analyzed based on multi-angle image data to obtain feature areas, and the boundaries and regional distribution of the retainer and the assembly station are clarified, thereby providing accurate input data for feature point extraction; based on the feature areas, the center points and contour points of the retainer and the assembly station are extracted to obtain retainer feature points and assembly station feature points, thereby providing specific and comprehensive geometric information for subsequent deviation analysis, and at the same time improving the adaptability to complex retainer structures through a multidimensional feature extraction method; the relative positions between the retainer feature points and the assembly station feature points are analyzed and it is determined whether there is a deviation in the retainer based on the relative positions, and the displacement and angle errors that may occur during the assembly process are quantitatively described, thereby providing an accurate basis for the automatic correction module, and at the same time ensuring that the system can monitor the assembly status in real time to avoid error accumulation affecting the final assembly effect.

[0084] In a preferred embodiment of the present invention, the three-dimensional coordinate system module includes:

[0085] The assembly model unit is used to construct a three-dimensional model according to the assembly station to obtain a three-dimensional assembly model;

[0086] The coordinate system unit is used to establish a three-dimensional coordinate system based on the three-dimensional assembly model, with the center point of the three-dimensional assembly model as the origin, the horizontal straight line passing through the center point of the three-dimensional assembly model as the x-axis, the vertical straight line passing through the center point of the three-dimensional assembly model as the y-axis, and the straight line passing through the center point of the three-dimensional assembly model and perpendicular to the plane formed by the x-axis and the y-axis as the z-axis.

[0087] In an embodiment of the present invention, a three-dimensional model is constructed based on the assembly station to obtain a three-dimensional assembly model, which provides an accurate geometric reference for the entire assembly system, so that the spatial relationship between the retaining frame and the station can be accurately described in subsequent steps, and at the same time provides a reference standard for deviation analysis and correction; based on the three-dimensional assembly model, a three-dimensional coordinate system is established, and accurate spatial reference is provided through a unified coordinate system, so that subsequent deviation detection and correction can be efficiently calculated and executed based on this coordinate system.

[0088] The assembly model unit is used to construct a three-dimensional model according to the assembly station to obtain a three-dimensional assembly model, which specifically includes:

[0089] First, the geometric parameters and physical size data of the assembly station are collected, such as the width, height, groove depth, etc. of the station surface. Then, the geometric shape of the station is digitized through modeling tools such as CAD to generate a three-dimensional model. Finally, the model is refined and the edge transition area is optimized to ensure that the model is completely consistent with the actual station.

[0090] In a preferred embodiment of the present invention, the model building module includes:

[0091] A viewing angle analysis unit, used to analyze the viewing angle errors of the feature points of the cage in images at different angles according to the feature point data set, and obtain a parallax adjustment factor;

[0092] A structure generation unit, used for adjusting the position of the characteristic points of the cage according to the parallax adjustment factor, and spatially connecting the adjusted characteristic points of the cage and the characteristic points of the assembly station to generate a three-dimensional structure;

[0093] A detail filling unit is used to obtain a preliminary solid assembly model by adding filling points in the space area between the feature points in the three-dimensional structure;

[0094] The model optimization unit is used to smooth the preliminary solid assembly model, eliminate the abrupt edges on the surface of the preliminary solid assembly model, and obtain the solid assembly model.

[0095] In an embodiment of the present invention, according to a feature point data set, the viewing angle error of the feature points of the retainer in images at different angles is analyzed to obtain a parallax adjustment factor, thereby eliminating the feature point position error caused by parallax in the multi-view images, and providing high-precision data support for the subsequent generation of a three-dimensional structure; according to the parallax adjustment factor, the position of the feature points of the retainer is adjusted, and the adjusted feature points of the retainer and the feature points of the assembly station are spatially connected to generate a three-dimensional structure, and a true three-dimensional relationship between the retainer and the assembly station is constructed from the multi-angle images to ensure that the subsequent assembly model can accurately reflect the actual working conditions; by adding filling points in the spatial area between the feature points in the three-dimensional structure, a preliminary entity assembly model is obtained, the integrity and usability of the three-dimensional model are improved, a high-quality geometric basis is provided for subsequent correction and assembly steps, and the error of the model in subsequent operations is reduced; the preliminary entity assembly model is smoothed to eliminate the abrupt edges on the surface of the preliminary entity assembly model to obtain a physical assembly model, and the adaptability and accuracy of the physical assembly model are improved through meticulous optimization processing to provide a guarantee for the final assembly effect, while significantly reducing the problems caused by model errors during the correction process.

[0096] The structure generation unit is used to adjust the position of the cage feature point according to the parallax adjustment factor, and spatially connect the adjusted cage feature point and the assembly station feature point to generate a three-dimensional structure, which specifically includes:

[0097] First, the position of the feature points of the cage is corrected by the parallax adjustment factor to ensure the consistency of the feature points in the three-dimensional space. Then, based on the position data of these feature points, these feature points are connected to generate a three-dimensional point cloud structure. Finally, the point cloud structure is improved by the topology optimization method. According to the preset constraints, the point cloud structure after each iteration is optimized to the target result, forming a preliminary three-dimensional structure. Among them, the detail filling unit is used to obtain a preliminary entity assembly model by adding filling points in the space area between the feature points in the three-dimensional structure, which specifically includes:

[0098] First, the spatial distribution of feature points is analyzed to identify blank areas and unconnected areas. Then, filling points are generated through an interpolation algorithm to supplement the integrity of the three-dimensional structure. Finally, the density of the filled three-dimensional model is optimized to ensure that the generated solid assembly model has a smooth and continuous surface.

[0099] In a preferred embodiment of the present invention, the calculation formula of the parallax adjustment factor is: ;

[0100] in, is the parallax adjustment factor, is the number of cameras, and is the index of the camera, and , For the Camera and The angle between the imaging points of the cameras, For the The focal length of the camera, For the The focal length of the camera, For the The distance from the camera to the holder and assembly station, For the The distance from the camera to the holder and assembly station, is a constant.

[0101] In an embodiment of the present invention, the geometric relationship of camera imaging is analyzed, the projection difference of feature points in different viewing angles is derived using trigonometric functions, and the focal length and distance parameters are substituted into the formula to calculate the parallax adjustment factor of the feature points in three-dimensional space to ensure its applicability to the overall model. The formula can unify multi-view image data into a consistent three-dimensional coordinate system, providing a mathematical basis for the subsequent construction and correction of the three-dimensional model, while ensuring the accuracy of the parallax calculation.

[0102] In a preferred embodiment of the present invention, the coordinate analysis module includes:

[0103] A proportion adjustment unit is used to adjust the physical assembly model to ensure that the proportion of the physical assembly model is consistent with that of the three-dimensional assembly model, thereby obtaining model data;

[0104] A model matching unit is used to superimpose the physical assembly model and the three-dimensional assembly model according to the model data, so as to ensure that the assembly stations of the physical assembly model and the three-dimensional assembly model coincide with each other, and obtain a superimposed model;

[0105] The entity coordinate unit is used to identify the coordinates of the entity assembly model in the three-dimensional coordinate system according to the superposition model to obtain the three-dimensional coordinate data.

[0106] In an embodiment of the present invention, the physical assembly model is adjusted to ensure that the proportions of the physical assembly model and the three-dimensional assembly model are consistent, and model data is obtained to solve the problem of proportion deviation between the physical model and the three-dimensional model that may be caused by different sources, and the accuracy of model superposition and the accuracy of subsequent assembly processes are enhanced through precise proportion matching; according to the model data, the physical assembly model and the three-dimensional assembly model are superimposed to ensure that the assembly stations of the physical assembly model and the three-dimensional assembly model coincide with each other, and a superimposed model is obtained. Through high-precision model superposition, it is ensured that the assembly stations of the physical model and the three-dimensional assembly model are completely consistent, providing reliable spatial data support for subsequent deviation detection and correction; according to the superimposed model, the coordinates of the physical assembly model in the three-dimensional coordinate system are identified to obtain three-dimensional coordinate data, and provide accurate spatial position data for the deviation detection module.

[0107] In a preferred embodiment of the present invention, the deviation detection module includes:

[0108] The center deviation unit is used to analyze whether the center point coordinates of the solid assembly model are located at the origin of the three-dimensional coordinate system according to the three-dimensional coordinate data. When the result is yes, the solid assembly model does not have a center deviation. When the result is no, the solid assembly model has a center deviation, and the center deviation value is calculated according to the center point coordinates of the solid assembly model.

[0109] The angle deviation unit is used to analyze whether there is an angle between the retaining frame and the assembly station in the superposition model according to the three-dimensional coordinate data. When the result is no, there is no angle deviation in the physical assembly model. When the result is yes, there is an angle deviation in the physical assembly model, and the angle deviation value is calculated.

[0110] In an embodiment of the present invention, based on the three-dimensional coordinate data, it is analyzed whether the center point coordinates of the physical assembly model are located at the origin of the three-dimensional coordinate system, and through precise center deviation analysis, it is ensured that the center position of the retaining frame during the assembly process is completely consistent with the design requirements, thereby avoiding assembly errors caused by center offset; based on the three-dimensional coordinate data, it is analyzed whether there is an angle between the retaining frame and the assembly station in the superposition model, thereby ensuring that the relative posture between the retaining frame and the station during the assembly process meets the design requirements, thereby avoiding reduced operating efficiency or increased component wear due to angle deviation.

[0111] In a preferred embodiment of the present invention, the calculation formula of the angle deviation value is: ;

[0112] in, is the angle deviation value, They are respectively the entity assembly model The horizontal, vertical and vertical coordinates of the contour points are They are the three-dimensional assembly models The horizontal, vertical and vertical coordinates of the contour points are Assemble the model for the entity, For a three-dimensional assembly model, is the total number of contour points, is the index of the contour point, For the The coefficients of the contour points.

[0113] In an embodiment of the present invention, the three-dimensional coordinates of all contour points are extracted, and the angle difference of each pair of contour points is calculated. The angle difference data are then summed and the average value is obtained through the total number of points to ensure that the calculation result is more in line with the assembly requirements, improve the accuracy of the angle deviation value calculation, and reduce the error accumulation problem that may occur in complex scenarios.

[0114] In a preferred embodiment of the present invention, the automatic correction module includes:

[0115] A center deviation analysis unit, used to determine a translation direction vector according to a center deviation value, and normalize the translation direction vector to obtain a translation unit vector;

[0116] A center correction unit is used to determine the translation amount of the cage in the three-dimensional coordinate system according to the center deviation value and the translation unit vector, and obtain the translation correction vector;

[0117] An angle deviation analysis unit, used to determine the rotation direction vector according to the angle deviation value, and normalize the rotation direction vector to obtain a rotation unit vector;

[0118] An angle correction unit is used to determine the rotation amount of the cage on each axis according to the angle deviation value and the rotation unit vector to obtain a rotation correction vector;

[0119] The correction execution unit is used to perform translation correction and rotation correction on the retaining frame according to the translation correction vector and the rotation correction vector to obtain a correction result image.

[0120] In an embodiment of the present invention, the translation direction vector is determined based on the center deviation value, and the translation direction vector is normalized to obtain a translation unit vector, which accurately calibrates the offset direction of the retaining bracket and provides a directional reference for translation correction; the rotation direction vector is determined based on the angle deviation value, and the rotation direction vector is normalized to obtain a rotation unit vector, which provides a specific numerical value that the retaining bracket needs to move, ensuring that the correction operation is both accurate and efficient, and significantly improving the automation and accuracy of the assembly process.

[0121] According to the angle deviation value, the rotation direction vector is determined, and the rotation direction vector is normalized to obtain the rotation unit vector, which provides the posture adjustment direction and angle required for the cage correction, ensuring that the cage can quickly return to the correct angle position during the assembly process; according to the angle deviation value and the rotation unit vector, the rotation amount of the cage on each axis is determined to obtain the rotation correction vector. Through accurate angle correction amount calculation, it is ensured that the rotation correction operation of the cage in multiple axes meets the design requirements, thereby improving the assembly accuracy; according to the translation correction vector and the rotation correction vector, the cage is subjected to translation correction and rotation correction to obtain the correction result image, and the position and angle of the cage are adjusted to the design requirements through automated correction operations, thereby ensuring the accuracy of the assembly process.

[0122] The center deviation analysis unit is used to determine the translation direction vector according to the center deviation value, and normalize the translation direction vector to obtain the translation unit vector, which specifically includes:

[0123] First, the coordinate difference between the center point of the cage and the origin of the three-dimensional coordinate system is extracted from the center deviation value, and then the translation direction vector is calculated by the coordinate difference, such as the translation direction vector ,in They are the offsets of the center point in the x-axis, y-axis, and z-axis directions respectively. Finally, the translation direction vector is normalized to obtain the translation unit vector to ensure that the direction is clear and suitable for subsequent correction operations.

[0124] The center correction unit is used to determine the translation amount of the cage in the three-dimensional coordinate system according to the center deviation value and the translation unit vector to obtain the translation correction vector, which specifically includes:

[0125] First, the correction direction of the cage is determined according to the translation unit vector and the center deviation value, and then the center deviation value is decomposed into components in the x-axis, y-axis and z-axis directions to control the precise movement of the cage in each direction and obtain the translation correction vector.

[0126] The correction execution unit is used to perform translation correction and rotation correction on the holder according to the translation correction vector and the rotation correction vector to obtain a correction result image, which specifically includes:

[0127] First, the translation correction vector and rotation correction vector are received and parsed as the control input of the correction operation. Then, the robotic arm is controlled to perform precise translation and rotation adjustments to the holder according to the instructions of the correction vector. Finally, the correction result image is acquired and input into the correction verification module to verify whether the correction is successful.

[0128] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A bearing cage visual recognition automatic correction assembly system, characterized in that: The system comprises: An image acquisition module is used to acquire images of the retainer and the assembly station through a camera to obtain multi-view image data; An image processing module is used to pre-process the multi-view image data, remove noise and enhance edge features to obtain multi-angle image data; The feature point module is used to extract the feature points of the cage and the assembly station according to the multi-angle image data, and analyze whether there is a deviation in the cage according to the feature points to obtain a feature point data set; A three-dimensional coordinate system module is used to construct a three-dimensional assembly model according to the assembly station, and to establish a three-dimensional coordinate system according to the three-dimensional assembly model; The model building module is used to analyze the positions of feature points in images at different angles, generate a three-dimensional structure, fill the three-dimensional structure, and build a solid assembly model; A coordinate analysis module is used to superimpose the physical assembly model with the three-dimensional assembly model to ensure that the assembly stations of the physical assembly model and the three-dimensional assembly model coincide with each other, thereby obtaining a superimposed model, and determining the coordinates of the physical assembly model based on the superimposed model to obtain three-dimensional coordinate data; A deviation detection module is used to analyze the deviation between the retainer and the assembly station according to the three-dimensional coordinate data to obtain correction data, the correction data including the center deviation value and the angle deviation value; An automatic correction module is used to correct the holder according to the correction data to obtain a correction result image; The correction inspection module is used to feed back the correction result image to the image processing module, and detect whether there is deviation in the cage through the feature point module. When there is deviation in the cage, the above steps are repeated. When there is no deviation in the cage, the assembly is completed.

2. A bearing retainer visual recognition automatic correction assembly system according to claim 1, characterized in that: The image processing module comprises: An image denoising unit, used to remove noise interference in the image according to the multi-view image data to obtain clear image data; The edge enhancement unit is used to detect the gradient of the image edge in the horizontal and vertical directions according to the clear image data, determine the area with rapid gradient changes, and increase its pixel value to obtain edge enhanced image data; The scale analysis unit is used to analyze and capture the detail features of the image at different scales based on the edge enhanced image data to obtain multi-angle image data.

3. A bearing retainer visual recognition automatic correction assembly system according to claim 2, characterized in that: The feature point module includes: An image analysis unit, used to analyze the structure of the cage and the assembly station according to multi-angle image data to obtain a characteristic area; A feature analysis unit is used to extract the center point and contour point of the cage and the assembly station according to the feature area, and obtain the feature point of the cage and the feature point of the assembly station; The deviation analysis unit is used to analyze the relative position between the characteristic points of the cage and the characteristic points of the assembly station and determine whether there is a deviation in the cage based on the relative position. When the result is no, the assembly is completed. When the result is yes, the characteristic point data set is obtained based on the relative position between the characteristic points of the cage and the characteristic points of the assembly station, and the automatic correction function is turned on.

4. A bearing retainer visual recognition automatic correction assembly system according to claim 3, characterized in that: The three-dimensional coordinate system module includes: The assembly model unit is used to construct a three-dimensional model according to the assembly station to obtain a three-dimensional assembly model; The coordinate system unit is used to establish a three-dimensional coordinate system based on the three-dimensional assembly model, with the center point of the three-dimensional assembly model as the origin, the horizontal straight line passing through the center point of the three-dimensional assembly model as the x-axis, the vertical straight line passing through the center point of the three-dimensional assembly model as the y-axis, and the straight line passing through the center point of the three-dimensional assembly model and perpendicular to the plane formed by the x-axis and the y-axis as the z-axis.

5. A bearing retainer visual recognition automatic correction assembly system according to claim 4, characterized in that: The model building module includes: A viewing angle analysis unit, used to analyze the viewing angle errors of the feature points of the cage in images at different angles according to the feature point data set, and obtain a parallax adjustment factor; A structure generation unit, used for adjusting the position of the characteristic points of the cage according to the parallax adjustment factor, and spatially connecting the adjusted characteristic points of the cage and the characteristic points of the assembly station to generate a three-dimensional structure; A detail filling unit is used to obtain a preliminary solid assembly model by adding filling points in the space area between the feature points in the three-dimensional structure; The model optimization unit is used to smooth the preliminary solid assembly model, eliminate the abrupt edges on the surface of the preliminary solid assembly model, and obtain the solid assembly model.

6. A bearing retainer visual recognition automatic correction assembly system according to claim 5, characterized in that: The calculation formula of the parallax adjustment factor is: ； in, is the parallax adjustment factor, is the number of cameras, and is the index of the camera, and , For the Camera and The angle between the imaging points of the cameras, For the The focal length of the camera, For the The focal length of the camera, For the The distance from the camera to the holder and assembly station, For the The distance from the camera to the holder and assembly station, is a constant.

7. A bearing retainer visual recognition automatic correction assembly system according to claim 6, characterized in that: The coordinate analysis module comprises: A proportion adjustment unit is used to adjust the physical assembly model to ensure that the proportion of the physical assembly model is consistent with that of the three-dimensional assembly model, thereby obtaining model data; A model matching unit is used to superimpose the physical assembly model and the three-dimensional assembly model according to the model data, so as to ensure that the assembly stations of the physical assembly model and the three-dimensional assembly model coincide with each other, and obtain a superimposed model; The entity coordinate unit is used to identify the coordinates of the entity assembly model in the three-dimensional coordinate system according to the superposition model to obtain the three-dimensional coordinate data.

8. The bearing retainer visual recognition automatic correction assembly system according to claim 7, characterized in that: The deviation detection module comprises: The center deviation unit is used to analyze whether the center point coordinates of the solid assembly model are located at the origin of the three-dimensional coordinate system according to the three-dimensional coordinate data. When the result is yes, the solid assembly model does not have a center deviation. When the result is no, the solid assembly model has a center deviation, and the center deviation value is calculated according to the center point coordinates of the solid assembly model. The angle deviation unit is used to analyze whether there is an angle between the retaining frame and the assembly station in the superposition model according to the three-dimensional coordinate data. When the result is no, there is no angle deviation in the physical assembly model. When the result is yes, there is an angle deviation in the physical assembly model, and the angle deviation value is calculated.

9. A bearing retainer visual recognition automatic correction assembly system according to claim 8, characterized in that: The calculation formula of the angle deviation value is: ； in, is the angle deviation value, They are respectively the entity assembly model The horizontal, vertical and horizontal coordinates of the contour points are They are the three-dimensional assembly model The horizontal, vertical and horizontal coordinates of the contour points are Assemble the model for the entity, For a three-dimensional assembly model, is the total number of contour points, is the index of the contour point, For the The coefficients of the contour points.

10. A bearing retainer visual recognition automatic correction assembly system according to claim 9, characterized in that: The automatic correction module comprises: A center deviation analysis unit, used to determine a translation direction vector according to a center deviation value, and normalize the translation direction vector to obtain a translation unit vector; A center correction unit is used to determine the translation amount of the cage in the three-dimensional coordinate system according to the center deviation value and the translation unit vector, and obtain the translation correction vector; An angle deviation analysis unit, used to determine the rotation direction vector according to the angle deviation value, and normalize the rotation direction vector to obtain a rotation unit vector; An angle correction unit is used to determine the rotation amount of the cage on each axis according to the angle deviation value and the rotation unit vector to obtain a rotation correction vector; The correction execution unit is used to perform translation correction and rotation correction on the retaining frame according to the translation correction vector and the rotation correction vector to obtain a correction result image.

Images