A machine vision-based air spring quality detection system and method

Abstract

The application belongs to the field of quality detection, and provides an air spring quality detection system and method based on machine vision, which comprises the following steps: obtaining a standard three-dimensional model of an air spring under different pressures; applying a first pressure to a to-be-detected air spring; obtaining two-dimensional image data of the to-be-detected air spring at multiple angles under the first pressure; projecting a light fringe pattern onto the surface of the to-be-detected air spring, and obtaining three-dimensional shape data of the surface of the to-be-detected air spring under the first pressure according to the deformation of the light fringe; fusing the two-dimensional image data and the three-dimensional shape data to generate a first three-dimensional model of the air spring; obtaining a first standard three-dimensional model under the first pressure from a template module, comparing the texture and geometric information of the first three-dimensional model and the first standard three-dimensional model, and determining that the to-be-detected air spring is unqualified if the difference exceeds a preset threshold.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of quality detection, and specifically relates to an air spring quality detection system and method based on machine vision. BACKGROUND

[0002] Air springs, as a common damping device, are widely used in various fields such as automobiles, trains, and buildings. Its main principle is to use compressed air as an elastic medium, and to change the elastic coefficient by adjusting the air pressure, so as to effectively reduce vibration and impact. The performance of air springs directly affects the overall effect of the damping system, therefore, in the production and use process, the quality of air springs needs to be strictly detected and evaluated.

[0003] Traditional air spring quality detection methods usually rely on manual visual inspection and physical measurement, which is not only low in efficiency, but also difficult to ensure accuracy and consistency. With the development of technology, machine vision technology has been gradually introduced into the field of industrial detection, which can efficiently and accurately identify defects on the surface and shape of objects through cameras and image processing algorithms.

[0004] In existing three-dimensional vision solutions, although some progress has been made, there are still some inherent limitations, especially in texture detection. Three-dimensional vision technology usually focuses on the acquisition of object geometry and depth information, but has weak defect detection capability for details rich and dependent on texture information. This is because the fine features of surface texture are often lost in the three-dimensional reconstruction process, especially in the case of complex or uneven texture information in the image. The texture itself may appear aliasing, blurring or distortion in the image, resulting in inaccurate texture mapping of the three-dimensional model, which affects the identification and analysis of defects.

[0005] In addition, air springs (or gas springs) as a common elastic element, its deformation characteristics are directly related to external pressure. When the pressure of the air spring changes, its shape will change accordingly, usually showing compression or expansion. Due to the nonlinear relationship between air pressure and deformation, two-dimensional images will be affected by distortion when capturing these changes. Specifically, as deformation occurs, the original geometric structure and texture distribution in the image will be affected, and even severe deformation may occur, which is not conducive to image matching, making effective defect detection in two-dimensional images complex and challenging. SUMMARY

[0006] In order to solve the problems in the prior art, the present application provides an air spring quality detection system based on machine vision, comprising the following modules:

[0007] a template module for storing standard three-dimensional models of the air spring under different pressures, the standard three-dimensional models including surface textures and geometric information;

[0008] a pressurization module for applying a first pressure to the air spring to be tested;

[0009] an image acquisition module for acquiring two-dimensional image data of the air spring to be tested at multiple angles under the first pressure;

[0010] a structured light module for acquiring three-dimensional shape data of the surface of the air spring to be tested under the first pressure by projecting a light fringe pattern onto the surface of the air spring to be tested and according to the deformation of the light fringe;

[0011] a data fusion processing module for fusing the two-dimensional image data and the three-dimensional shape data to generate a first three-dimensional model of the air spring, the first three-dimensional model including surface textures and geometric information of the air spring to be tested under the first pressure;

[0012] a defect detection module for acquiring a first standard three-dimensional model under the first pressure from the template module, comparing the textures and geometric information of the first three-dimensional model and the first standard three-dimensional model, and determining that the air spring to be tested is unqualified if the difference exceeds a preset threshold.

[0013] Further, the three-dimensional shape data of the surface of the air spring to be tested under the first pressure is acquired by projecting a light fringe pattern onto the surface of the air spring to be tested and according to the deformation of the light fringe, which includes:

[0014] a standard cylindrical model is constructed in advance, the diameter and height of the standard cylindrical model correspond to the maximum diameter and maximum height of the air spring respectively; a first projected light fringe pattern is acquired through the standard cylindrical model, the first projected light fringe pattern is deformed in accordance with the geometric shape of the air spring to be tested, and the three-dimensional shape data of the surface of the air spring is acquired according to the deformation information of the light fringe after the first projected light fringe pattern passes through the surface of the air spring to be tested.

[0015] Further, the fusing of the two-dimensional image data and the three-dimensional shape data includes:

[0016] the deformation of the first projected light fringe pattern is registered with the two-dimensional image, and the two-dimensional image is mapped to the three-dimensional shape model according to the registration result.

[0017] Further, the two-dimensional image data at multiple angles completely covers the surface of the air spring to be tested.

[0018] Further, in the defect detection process, the texture difference and the geometric difference are compared, a weighted calculation is used, the weight of the texture difference and the geometric difference in quality judgment is considered, a difference value is obtained, if the difference exceeds a preset threshold, it is determined that the quality of the air spring to be tested is unqualified, otherwise, it is determined to be qualified.

[0019] The application further provides an air spring quality detection method based on machine vision, comprising the following steps:

[0020] Obtaining a standard three-dimensional model of the air spring under different pressures, saving the standard three-dimensional model of the air spring under different pressures, and the standard three-dimensional model comprising surface texture and geometric information;

[0021] Applying a first pressure to the air spring to be tested;

[0022] Obtaining two-dimensional image data of the air spring to be tested at multiple angles under the first pressure;

[0023] Projecting a light stripe pattern onto the surface of the air spring to be tested, and obtaining three-dimensional shape data of the surface of the air spring to be tested under the first pressure according to the deformation of the light stripe;

[0024] Fusing the two-dimensional image data and the three-dimensional shape data to generate a first three-dimensional model of the air spring, and the first three-dimensional model comprising surface texture and geometric information of the air spring to be tested under the first pressure;

[0025] Obtaining a first standard three-dimensional model under the first pressure from a template module, comparing the texture and geometric information of the first three-dimensional model and the first standard three-dimensional model, and determining that the quality of the air spring to be tested is unqualified if the difference exceeds a preset threshold.

[0026] Further, the three-dimensional shape data of the surface of the air spring to be tested under the first pressure is obtained by projecting a light stripe pattern onto the surface of the air spring to be tested, and the three-dimensional shape data of the surface of the air spring to be tested under the first pressure is obtained according to the deformation of the light stripe.

[0027] A standard cylindrical model is constructed in advance, the diameter and height of the standard cylindrical model correspond to the maximum diameter and maximum height of the air spring respectively, a first projected light stripe pattern is obtained through the standard cylindrical model, the first projected light stripe pattern is deformed in adaptation to the geometric shape of the air spring to be tested, and the three-dimensional shape data of the surface of the air spring is obtained according to the deformation information of the light stripe after the first projected light stripe pattern passes through the surface of the air spring to be tested.

[0028] Further, the fusing of the two-dimensional image data and the three-dimensional shape data comprises:

[0029] The deformation of the first projected light stripe pattern is registered with the two-dimensional image, and the two-dimensional image is mapped to a three-dimensional shape model according to the registration result.

[0030] Further, the two-dimensional image data of the plurality of angles completely covers the surface of the air spring to be tested.

[0031] Further, in the defect detection process, the texture difference and the geometric difference are compared, and a weighted calculation is used to consider the weight of the texture difference and the geometric difference in quality judgment to obtain a difference value. If the difference exceeds a preset threshold, it is determined that the quality of the air spring to be tested is unqualified, otherwise, it is determined to be qualified.

[0032] The present application effectively solves the problems in the background art by designing an air spring quality detection system and method based on machine vision. The specific beneficial effects are as follows:

[0033] Through the fusion of multi-dimensional information, especially the joint processing of three-dimensional images and texture images, the shortcomings of traditional three-dimensional vision schemes in texture recognition can be compensated. This scheme can more accurately capture the detailed texture of the object surface, thereby improving the accuracy of defect detection, especially in complex texture information conditions.

[0034] Considering the deformation characteristics of the air spring under pressure changes, the system can dynamically adjust the image acquisition and analysis method to cope with image distortion caused by deformation. This adaptability enables the system to process deformation data under different pressure conditions in real time, ensuring accurate detection of possible defects during pressure changes.

[0035] By combining the distortion of the three-dimensional surface, the system can match the two-dimensional image under complex deformation and distortion, thereby facilitating accurate identification of surface defects. This improved accuracy is particularly important in automated production and quality inspection, helping to improve the quality control level of the production process and reduce human error. BRIEF DESCRIPTION OF DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments. Obviously, the drawings described below are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0037] Figure 1 is a structural diagram of the system of the present application. DETAILED DESCRIPTION

[0038] Next, the preferred description of the application will be made in combination with the drawings and specific embodiments.

[0039] The present embodiment solves the above problems by the following steps:

[0040] In one embodiment, with reference to Figure 1 The present application provides a machine vision-based air spring quality detection system, aiming to realize quality evaluation and defect detection of air springs under different working conditions through efficient and accurate visual analysis means.

[0041] In the present application, an air spring is a damping device that uses compressed air as an elastic medium, widely used in damping systems of automobiles, trains, buildings, and other industries. The air spring is characterized by its flexibility and adjustability, which can adjust its elastic coefficient according to changes in load or compressed air pressure to meet the needs of different working conditions.

[0042] Specifically, the system of the present application includes the following modules:

[0043] A template module for saving standard three-dimensional models of air springs under different pressures, said standard three-dimensional models including surface texture and geometric information.

[0044] The template module in the present application refers to a module specifically used to store and manage standard three-dimensional models of air springs corresponding to different inflation pressures or working conditions. The template module can provide standard reference models for comparison and detection during the quality detection process of air springs. The standard three-dimensional model consists of two parts, namely surface texture information and geometric information, which together constitute the ideal state or reference form of the air spring under specific pressure conditions.

[0045] The main function of the template module is to store and provide standard three-dimensional models, which represent the standard shape of air springs under various working pressures.

[0046] The template module saves a set of standard three-dimensional models obtained through experiments and measurements, which can accurately reflect the ideal form of air springs under different working pressures. The standard model is designed and measured according to the working conditions of air springs in actual use, such as different loads, air pressures, etc.

[0047] During the quality detection process of air springs, the standard three-dimensional model provided by the template module serves as a comparison reference for comparing and analyzing the actual detected three-dimensional data of air springs, thereby identifying the parts that do not meet the standards, such as shape changes, defects, deformations, etc.

[0048] The template module not only saves standard models under a single state, but also includes standard three-dimensional models of air springs under different inflation pressures, loads, or working environmental conditions, so that the system can automatically select the corresponding reference model for comparison according to the actual working conditions during detection.

[0049] The standard three-dimensional model includes surface texture and geometric information. Surface texture information refers to the two-dimensional image data of the surface details of the air spring, including visual features such as color, surface texture, scratches, and wear. In the model, these surface information are attached to the three-dimensional surface through texture mapping technology, so that during the inspection process, not only the geometric shape of the spring can be obtained, but also subtle changes on its surface can be identified. Geometric information refers to the three-dimensional data that describes the shape of the air spring, including key structural features such as surface curvature, edge profile, size, and shape. Geometric information is generally obtained through three-dimensional scanning technology (such as laser scanning, structured light scanning, etc.) and expressed through point cloud data, three-dimensional meshes, etc. Geometric information provides key morphological data for quality inspection, ensuring that the inspection system can identify whether the size and shape of the air spring meet the standards.

[0050] The standard three-dimensional model can be stored in the template module using different technologies according to actual needs, such as STL, OBJ, etc. The specific storage method and form do not affect the main scheme of the present invention and are not specifically limited in this embodiment.

[0051] The boost module is used to apply a first pressure to the air spring to be tested.

[0052] The boost module in this context is a device specifically designed to apply a certain pressure to the air spring under test during air spring quality testing. Its primary function is to simulate the load or operating conditions of the air spring under specific operating conditions by applying a first pressure during testing, ensuring accurate assessment of the performance and quality of the air spring under test.

[0053] The boost module applies a first pressure to the air spring under test. This first pressure is set based on the air spring's operating requirements, test standards, or the specific task. This first pressure is typically the typical operating pressure of the air spring under normal use, or simulates a load pressure under specific conditions.

[0054] By applying a first pressure, the boost module simulates the operating conditions an air spring might experience during actual use, ensuring accurate detection of key parameters such as morphological and dimensional changes under varying pressures. This initial pressure helps to reveal deformation or surface defects in the air spring, providing a basis for subsequent quality testing.

[0055] The pressurizing module can not only apply a fixed first pressure, but also realize adjustable control of the pressure, so that the system can adjust the applied pressure value according to the detection requirement. By applying different pressures for multiple times, performance data of the air spring under different working conditions can be obtained, so that more comprehensive quality evaluation can be provided for the detection system.

[0056] The pressurizing module can realize the function of applying pressure in various ways, such as air pump, hydraulic device, etc., as long as the set pressure can be applied, and the embodiment is not limited in the implementation manner.

[0057] The image acquisition module is used to acquire two-dimensional image data of the air spring to be detected at multiple angles under the first pressure.

[0058] The image acquisition module in the present application refers to a device module specially used for acquiring image data of the air spring to be detected under a specific pressure condition. The module captures two-dimensional images of the air spring under the action of the first pressure from different angles through the installation of multiple camera devices, so as to provide required image data for subsequent quality detection, three-dimensional modeling and defect analysis.

[0059] The main function of the image acquisition module is to capture the morphological changes and surface information of the air spring to be detected after the first pressure is applied through the camera device. The image acquisition module is used to acquire two-dimensional image data from multiple different angles when the air spring is under the first pressure. These image data include the surface texture, geometric morphology and possible defect characteristics of the air spring. In order to more comprehensively acquire the morphological information of the air spring, the image acquisition module adopts multiple camera devices to capture images from different angles. Through the image data acquired from multiple perspectives, more accurate quality analysis can be performed on the air spring, especially in terms of complex morphology, texture and the like of the spring surface.

[0060] The image acquisition module includes an image sensor and a camera device. The image sensor is the core component of the image acquisition module, which is responsible for converting the morphology of the air spring to be detected under a specific pressure into processable digital image data. The image sensor can be a CMOS sensor or a CCD sensor, which can provide high-resolution image quality to clearly record the surface information and possible defects of the air spring. The image acquisition module includes multiple camera devices arranged in the form of multiple camera arrays to capture image data of the air spring from different angles. Each camera device can adjust the shooting angle and distance according to the position and morphology of the air spring to be detected, so as to ensure the integrity and accuracy of the image data.

[0061] In order to ensure that all texture data is obtained, the two-dimensional image data of the multiple angles completely covers the surface of the air spring to be detected.

[0062] The structural light module is used for acquiring the three-dimensional shape data of the surface of the air spring under the first pressure by projecting a light fringe pattern to the surface of the air spring to be measured and acquiring the deformation of the light fringe.

[0063] The structural light module in the present application is a device module for acquiring three-dimensional shape data by using structural light technology. The structural light technology projects a light fringe pattern with a known pattern to the surface of an object, and then captures the deformation of the light fringe on the surface to calculate the three-dimensional shape information of the object. In the application of the present application, the structural light module is used to acquire the three-dimensional shape data of the surface of the air spring under the first pressure, so as to evaluate and detect the quality of the air spring.

[0064] The main function of the structural light module is to acquire the three-dimensional shape data of the surface of the air spring to be measured by projecting and deforming the light fringe pattern. The specific functions include:

[0065] The structural light module projects a specific light fringe pattern to the surface of the air spring to be measured by using a projection device. The shape, density and direction of these fringe patterns are known and are used for subsequent three-dimensional reconstruction.

[0066] The geometry and texture of the surface of the air spring to be measured will cause the deformation of the projected light fringe. By capturing the deformed light fringe pattern by a camera, the structural light module can calculate the three-dimensional surface shape of the air spring to be measured according to the deformation.

[0067] By analyzing the deformed light fringe pattern, combining the known optical principles and geometric algorithms, the structural light module can calculate and output the three-dimensional shape data of the surface of the air spring to be measured. The data not only contains the geometric profile of the object, but also contains the surface detail information.

[0068] The processing module is responsible for analyzing and calculating the captured light fringe deformation data. By using image processing algorithms, the deformation of the light fringe is first identified, and then the three-dimensional surface shape of the air spring to be measured is calculated by using the known geometric model and optical transformation relationship. The processing module can also output the results in the form of three-dimensional coordinate points, surface grid or depth map as needed.

[0069] Further, a standard cylindrical model is constructed in advance, the diameter and height of the standard cylindrical model correspond to the maximum diameter and maximum height of the air spring respectively; a first projected light fringe pattern is acquired through the standard cylindrical model, the first projected light fringe pattern is deformed in adaptation to the geometric shape of the air spring to be measured, and the three-dimensional shape data of the surface of the air spring is acquired according to the deformation information of the light fringe after the first projected light fringe pattern passes through the surface of the air spring to be measured.

[0070] The standard cylindrical model is constructed based on the maximum diameter and maximum height of the air spring. This model is used as a reference standard for projecting the light stripe pattern. By generating a standard light stripe pattern under ideal conditions and deforming it, the deformation of the air spring surface under different pressures can be simulated.

[0071] The structured light method is to project a known light stripe pattern onto the object surface and calculate the three-dimensional shape of the object according to the deformation of the light stripe caused by the geometry of the object surface. In this scheme, a light stripe pattern is first generated according to the geometric features of the standard cylindrical model, and then the pattern is projected onto the surface of the air spring to be measured. Since the surface features of the air spring will cause deformation of the light stripe, after capturing these deformation images by the camera device, the three-dimensional shape data of the measured object is calculated by the algorithm.

[0072] By analyzing the deformed light stripe pattern, the computer vision algorithm can accurately invert the three-dimensional data of the measured air spring. These three-dimensional data contain the surface geometric information and texture features of the air spring, which enable accurate detection of whether its shape meets the standard requirements.

[0073] This method can accurately capture the three-dimensional geometry of the surface of the air spring to be measured, with high measurement accuracy. By obtaining three-dimensional information through the deformation of the light stripe, the deformation characteristics of the air spring under different pressures can be accurately reflected. By pre-constructing a standard cylindrical model and adapting the projection of the light stripe, the distortion caused by changes in the surface shape can be effectively reduced, ensuring the accuracy of the three-dimensional data and improving the reliability of quality detection.

[0074] The data fusion processing module is used to fuse the two-dimensional image data and the three-dimensional shape data to generate a first three-dimensional model of the air spring, which includes the surface texture and geometric information of the air spring to be measured under the first pressure.

[0075] The data fusion processing module in the present application is a module for fusing multiple types of measurement data. It combines two-dimensional image data with three-dimensional shape data to generate a first three-dimensional model of the air spring to be measured. This three-dimensional model not only contains the geometric shape information of the air spring, but also includes the surface texture information, thereby achieving comprehensive and accurate modeling of the air spring under the first pressure. This process provides accurate data support for subsequent quality evaluation and performance analysis.

[0076] The main function of the data fusion processing module is to establish a connection between two-dimensional image data and three-dimensional morphology data, effectively fuse these two different types of data, and generate a unified three-dimensional model. Through this model, the surface morphology, texture characteristics, and geometric structure of the air spring under test at the first pressure can be comprehensively described. The specific functions are as follows:

[0077] Two-dimensional image data provides high-resolution information of the surface details of the air spring under test, including color, texture, and other surface features; while three-dimensional morphology data provides geometric structure information of the object surface, such as depth, contour, and shape. Through the data fusion processing module, two-dimensional images and three-dimensional morphology data can be combined in the same coordinate system to form a complete three-dimensional model.

[0078] During the fusion process, the data fusion processing module uses image processing algorithms and three-dimensional reconstruction algorithms to match the texture information of two-dimensional images with the geometric structure of three-dimensional morphology, generating a three-dimensional model that can contain both the geometric shape and surface texture of the air spring, reflecting its true state under the first pressure.

[0079] Through fusion, the texture information in the two-dimensional image is mapped onto the surface of the three-dimensional geometric model, enhancing the detail performance of the model. The geometric information in the model is obtained through three-dimensional morphology data, ensuring accurate representation of surface curvature, boundary, and volume, etc.

[0080] The data fusion processing module effectively combines data from different sources to improve the accuracy and detail display of the final three-dimensional model, especially in the combination of surface texture and geometric structure, greatly enhancing the realism and usability of the model.

[0081] The composition of the data fusion processing module mainly includes data reception and preprocessing unit, data alignment and fusion unit, and three-dimensional modeling unit. Each unit cooperates with each other to complete the whole process from data reception, preprocessing to final three-dimensional model generation.

[0082] The data reception and preprocessing unit obtains two-dimensional image data from the camera device, including surface images of the air spring under test at different angles. These images contain information such as color, texture, and light reflection of the surface, which are used for subsequent texture mapping. Three-dimensional morphology data obtained through devices such as structured light modules provides geometric information of the air spring under test, including surface contour, curvature, depth, etc. Three-dimensional morphology data is generally represented in the form of point cloud or mesh. After receiving these data, the preprocessing unit performs preliminary processing, including denoising, data completion, image enhancement, etc., to improve the quality and usability of the data.

[0083] The data alignment and fusion unit is responsible for aligning the two-dimensional image data and three-dimensional morphological data in the same coordinate system. The alignment algorithm matches the feature points (such as edges, corners, etc.) in the image and three-dimensional data, and aligns the two-dimensional image and three-dimensional morphological data in the same coordinate system. The aligned two-dimensional image data and three-dimensional morphological data are synthesized by a fusion algorithm. This process usually uses texture mapping technology to map the texture information of the two-dimensional image to the surface of the three-dimensional model. In addition, the geometry reconstruction algorithm ensures that the geometric features of the three-dimensional morphological data are correctly presented. Finally, the fusion result generates a complete three-dimensional model containing surface texture and geometric information.

[0084] The three-dimensional modeling unit is responsible for reconstructing the model according to the fused data. Through the fusion of point cloud data and texture images, the three-dimensional geometry of the air spring is generated. This process usually uses three-dimensional reconstruction algorithms (such as voxel method, surface reconstruction method, etc.) to accurately construct the geometric shape of the object surface. The texture information in the two-dimensional image data is mapped to the surface of the three-dimensional model. This process corresponds each pixel point in the image to the corresponding point on the surface of the three-dimensional model by calculating the coordinate relationship between the image and the three-dimensional data, achieving accurate texture mapping.

[0085] Further, the deformation of the first projected light stripe pattern is registered with the two-dimensional image, and the two-dimensional image is mapped to the three-dimensional morphological model according to the registration result.

[0086] In the foregoing steps, the structured light module projects the first projected light stripe pattern onto the surface of the air spring to be measured to obtain the reflected light stripe deformation data. In order to accurately combine the surface texture and the three-dimensional geometric morphological data, the system uses image registration technology to register the deformation of the first projected light stripe pattern with the two-dimensional image data. The registration process includes aligning the two-dimensional image from different perspectives with the three-dimensional model obtained by the light stripe deformation through feature extraction, matching, and geometric transformation algorithms. Specifically, the registration method can include feature point matching, edge detection, or full-image registration technology based on optimization algorithms to reduce errors caused by different perspectives or different shooting angles, thereby achieving high-precision alignment of two-dimensional images and three-dimensional morphological data. Finally, the registered two-dimensional image is mapped to the surface of the three-dimensional morphological model to form a complete three-dimensional model containing surface texture and geometric information, providing intuitive and accurate data support for subsequent quality detection.

[0087] Through precise registration technology, the texture data of two-dimensional images can be effectively combined with three-dimensional shape data, effectively solving the errors caused by shooting angles and object deformation, ensuring accurate mapping of textures and realistic reflection of geometric shapes. Through the way of light stripe deformation and two-dimensional image registration, texture distortion can be avoided, especially in the case of complex air spring surface deformation, the texture data can be accurately fitted to the three-dimensional model, enhancing the expression effect of the texture.

[0088] By fusing two-dimensional image data and three-dimensional shape data, the generated three-dimensional model not only has the geometric structure characteristics of the air spring, but also accurately presents its surface texture, with high detail performance. The data fusion processing module can effectively reconstruct the surface texture and geometric features, ensuring the accuracy and authenticity of the model, especially in the detection of surface defects and small deformations. The fused three-dimensional model has a high degree of visualization, accurately presenting the morphological changes of the air spring under the first pressure, facilitating subsequent analysis, evaluation and display.

[0089] The defect detection module is used to obtain the first standard three-dimensional model under the first pressure from the template module, compare the texture and geometric information of the first three-dimensional model and the first standard three-dimensional model, and if the difference exceeds the preset threshold, determine that the quality of the air spring to be tested is unqualified.

[0090] The defect detection module in the present application is an important component of the air spring quality detection system, and its main function is to compare the differences between the three-dimensional model of the air spring to be tested and the standard three-dimensional model, so as to determine whether the air spring to be tested meets the predetermined quality standard. Based on the setting of the difference threshold, the module accurately identifies possible surface defects or geometric errors, and then evaluates whether the quality of the air spring is qualified. Specifically, the defect detection module compares with the first standard three-dimensional model provided by the template module to ensure that the performance indicators of the air spring to be tested meet the quality standards and meet the product requirements.

[0091] The composition of the defect detection module mainly includes a standard three-dimensional model acquisition unit, a texture comparison unit, a geometric comparison unit, a difference calculation and determination unit, and a result output unit. Each unit works together to realize the whole process from comparison and analysis of standard models and test models to final quality determination through a series of data processing steps.

[0092] A standard three-dimensional model acquisition unit, the standard three-dimensional model is a three-dimensional model generated by the template module based on air spring design specifications and process requirements, serving as a reference for subsequent comparison. The surface texture and geometric information in the standard model are pre-set and have passed quality verification. The generation of the standard model is carried out under a specific environmental pressure (including the first pressure), ensuring that it represents the actual performance of the air spring under working conditions. The template module will provide three-dimensional model data under this pressure condition.

[0093] A texture comparison unit, the texture comparison unit extracts surface texture information from the three-dimensional model of the air spring to be tested, including the color, smoothness, defects (such as cracks, bubbles, scratches, etc.), and other surface features. The texture comparison unit compares the texture information of the air spring to be tested with the texture information in the standard three-dimensional model. Through image processing techniques (such as gray difference, edge detection, texture mapping, etc.), the differences between the two are calculated. If the texture of the test model deviates significantly from the standard model, it is considered to have defects.

[0094] A geometric comparison unit, the geometric comparison unit extracts geometric information from the three-dimensional model of the air spring to be tested, including surface profile, size, curvature, volume, etc. This unit extracts geometric features such as shape and boundary through point cloud or mesh model analysis. The geometric comparison unit compares the geometric information of the test model with the geometric information in the standard three-dimensional model. Through geometric difference detection algorithms, the differences in size, shape, and surface curvature between the two are calculated. If the difference exceeds the set tolerance range, it is determined to be unqualified.

[0095] A difference calculation and judgment unit, this unit integrates the difference data output by the texture comparison unit and the geometric comparison unit, and combines the preset difference threshold to determine whether the air spring to be tested meets the quality standards. Difference calculation usually uses a weighted calculation method, considering the weight of texture difference and geometric difference in quality judgment. If the difference exceeds the preset threshold, the quality of the air spring to be tested is determined to be unqualified. Otherwise, it is determined to be qualified. The judgment result will serve as the basis for subsequent processing (such as scrapping, repairing, retesting, etc.).

[0096] A result output unit, the defect detection module will feed back the quality judgment result to the user or control system through the result output unit. The output result can be displayed in the states of qualified, unqualified, or to be further detected, etc. The test results can be stored in the database and generate a test report, detailing the quality status of the air spring to be tested, the difference during the test process, and the final quality judgment.

[0097] On the other hand, the present application also provides an air spring quality detection method based on machine vision, comprising:

[0098] Obtaining standard three-dimensional models of the air spring under different pressures, saving the standard three-dimensional models of the air spring under different pressures, the standard three-dimensional models including surface texture and geometric information;

[0099] Applying a first pressure to the air spring to be measured;

[0100] Obtaining two-dimensional image data of the air spring to be measured under the first pressure from multiple angles;

[0101] Obtaining three-dimensional shape data of the surface of the air spring to be measured under the first pressure by projecting a light fringe pattern onto the surface of the air spring to be measured and according to the deformation of the light fringe;

[0102] Fusing the two-dimensional image data and the three-dimensional shape data to generate a first three-dimensional model of the air spring, the first three-dimensional model including surface texture and geometric information of the air spring to be measured under the first pressure;

[0103] Obtaining a first standard three-dimensional model under the first pressure from a template module, comparing the texture and geometric information of the first three-dimensional model and the first standard three-dimensional model, and determining that the air spring to be measured is unqualified if the difference exceeds a preset threshold.

[0104] The part of the modules of the present application which are not particularly clear are subject to the content of the prior art. The prior art mentioned in the foregoing background section and the specific embodiment section of the present application can be used as a part of the present application to understand the meaning of some technical features or parameters.

Claims

1. A machine vision based air spring quality detection system, characterized in that, The detection system comprises the following modules: a template module for storing standard three-dimensional models of air springs under different pressures, the standard three-dimensional models comprising surface textures and geometric information; a pressurization module for applying a first pressure to the air spring to be tested; an image acquisition module for acquiring two-dimensional image data of the air spring to be tested at multiple angles under the first pressure; a structured light module for acquiring three-dimensional shape data of the surface of the air spring to be tested under the first pressure by projecting a light fringe pattern onto the surface of the air spring to be tested and according to the deformation of the light fringe; a data fusion processing module for fusing the two-dimensional image data and the three-dimensional shape data to generate a first three-dimensional model of the air spring, the first three-dimensional model comprising surface textures and geometric information of the air spring to be tested under the first pressure; a defect detection module for acquiring a first standard three-dimensional model under the first pressure from the template module, comparing the first three-dimensional model and the first standard three-dimensional model in terms of textures and geometric information, and determining that the air spring to be tested is unqualified if the difference exceeds a preset threshold value; the structured light module comprises the following steps: a standard cylindrical model is constructed in advance, the diameter and height of the standard cylindrical model corresponding to the maximum diameter and maximum height of the air spring respectively; a first projected light fringe pattern is acquired through the standard cylindrical model, the first projected light fringe pattern being deformed in adaptation to the geometric shape of the air spring to be tested; and three-dimensional shape data of the surface of the air spring to be tested is acquired according to the deformation information of the light fringe after the first projected light fringe pattern passes through the surface of the air spring to be tested; the data fusion processing module comprises the following steps: the deformation of the first projected light fringe pattern is registered with the two-dimensional image, and the two-dimensional image is mapped to the three-dimensional shape model according to the registration result.

2. The machine vision-based air spring quality detection system of claim 1, wherein, The two-dimensional image data at multiple angles completely covers the surface of the air spring to be tested.

3. The machine vision-based air spring quality detection system of claim 1, wherein, In the defect detection process, the texture difference and the geometric difference are compared, a weighted calculation is used, the weights of the texture difference and the geometric difference in quality determination are considered, a difference value is obtained, and if the difference exceeds a preset threshold value, it is determined that the air spring to be tested is unqualified, otherwise, it is determined to be qualified.

4. A method for detecting the quality of an air spring based on machine vision, characterized in that, The method comprises the following steps: standard three-dimensional models of air springs under different pressures are acquired and stored, the standard three-dimensional models comprising surface textures and geometric information; a first pressure is applied to the air spring to be tested; two-dimensional image data of the air spring to be tested at multiple angles under the first pressure is acquired; three-dimensional shape data of the surface of the air spring to be tested under the first pressure is acquired by projecting a light fringe pattern onto the surface of the air spring to be tested and according to the deformation of the light fringe; and Fusing the two-dimensional image data and the three-dimensional shape data to generate a first three-dimensional model of the air spring, the first three-dimensional model including surface texture and geometric information of the air spring under the first pressure; Obtaining a first standard three-dimensional model under the first pressure from a template module, and comparing the texture and geometric information of the first three-dimensional model and the first standard three-dimensional model, if the difference exceeds a preset threshold, determining that the quality of the air spring under test is unqualified; The method for obtaining the three-dimensional shape data of the surface of the air spring under the first pressure by projecting a light fringe pattern onto the surface of the air spring under test includes: A standard cylindrical model is constructed in advance, the diameter and height of the standard cylindrical model corresponding to the maximum diameter and maximum height of the air spring respectively; a first projected light fringe pattern is obtained through the standard cylindrical model, the first projected light fringe pattern deforming in accordance with the geometric shape of the air spring under test, and the three-dimensional shape data of the surface of the air spring under test being obtained according to the deformation information of the light fringe after the first projected light fringe pattern passes through the surface of the air spring under test; The method for fusing the two-dimensional image data and the three-dimensional shape data includes: Registering the deformation of the first projected light fringe pattern with the two-dimensional image, and mapping the two-dimensional image to the three-dimensional shape model according to the registration result.

5. The machine vision-based air spring quality detection method of claim 4, wherein, The two-dimensional image data at the plurality of angles completely covers the surface of the air spring under test.

6. The machine vision-based air spring quality inspection method of claim 4, wherein, In the defect detection process, the texture difference and the geometric difference are compared, a weighted calculation is used, the weights of the texture difference and the geometric difference in quality determination are considered, a difference value is obtained, if the difference exceeds a preset threshold, it is determined that the quality of the air spring under test is unqualified, otherwise, it is determined to be qualified.

Images