Part batch size detection method and device based on structured light projection

Abstract

The application provides a part batch size detection method and device based on structured light projection, and relates to the technical field of visual detection, and comprises the following steps: a structured light source and an image acquisition device are pre-disposed in a detection area; when a first part to be detected is conveyed to a detection position of the detection area, the structured light source projects a preset structured light sequence to the first part to be detected in time, and the image acquisition device acquires images of the first part to be detected, and structured light deformation information is extracted according to the image acquisition result; three-dimensional coordinate conversion is carried out based on a triangulation method, and three-dimensional point cloud data of the first part to be detected is generated; and size detection is carried out based on the three-dimensional point cloud data, and first size information is acquired. The application solves the technical problem that the part size detection method in the prior art usually depends on physical measurement tools, cannot quickly and effectively detect a large number of parts accurately, and leads to low production efficiency.

Description

Technical Field

[0001] This invention relates to the field of visual inspection technology, and more specifically to a method and apparatus for batch dimensional inspection of parts based on structured light projection. Background Technology

[0002] In precision manufacturing and mass production, the dimensional accuracy of parts directly affects product performance and safety. Traditional dimensional inspection methods typically rely on physical measuring tools such as calipers, micrometers, and laser measuring instruments. While these tools offer high accuracy when measuring individual parts, they present numerous challenges in mass production. For large-scale production, traditional measuring tools require manual operation and are slow, failing to meet the demands for rapid inspection. Furthermore, the unavoidable nature of manual operation easily introduces human error, reducing the accuracy and consistency of inspections. Consequently, under the requirements of precision manufacturing and efficient production, traditional dimensional inspection methods cannot quickly and effectively perform accurate inspections on large batches of parts, impacting production efficiency and increasing the difficulty of quality control. Summary of the Invention

[0003] This application provides a method and apparatus for batch size inspection of parts based on structured light projection, which aims to solve the technical problem that existing part size inspection methods usually rely on physical measuring tools, which cannot quickly and effectively inspect large batches of parts accurately, resulting in reduced production efficiency.

[0004] The first aspect disclosed in this application provides a method for batch dimensional inspection of parts based on structured light projection. The method includes: pre-deploying a structured light source and an image acquisition device in a detection area, wherein the image acquisition device has preset calibration parameters; when a first part to be inspected is transported to the detection position in the detection area, the structured light source projects a preset structured light sequence onto the first part to be inspected in a time-division manner, while the image acquisition device acquires an image of the first part to be inspected, and extracts structured light deformation information based on the image acquisition results; based on the structured light deformation information and the preset calibration parameters, performing a three-dimensional coordinate transformation based on triangulation to generate three-dimensional point cloud data of the first part to be inspected; and performing dimensional inspection based on the three-dimensional point cloud data to obtain first dimensional information of the first part to be inspected.

[0005] The second aspect of this application discloses a batch size inspection device for parts based on structured light projection. The device is used in the aforementioned batch size inspection method for parts based on structured light projection. The device includes: a device deployment module for pre-deploying a structured light source and an image acquisition device in the inspection area, wherein the image acquisition device has preset calibration parameters; an image acquisition module for, when a first part to be inspected is transferred to the inspection position in the inspection area, for the structured light source to project a preset structured light sequence onto the first part to be inspected in a time-division manner, while the image acquisition device simultaneously acquires an image of the first part to be inspected, and extracts structured light deformation information based on the image acquisition results; a three-dimensional coordinate transformation module for performing three-dimensional coordinate transformation based on the structured light deformation information and the preset calibration parameters, generating three-dimensional point cloud data of the first part to be inspected; and a size inspection module for performing size inspection based on the three-dimensional point cloud data to obtain first size information of the first part to be inspected.

[0006] One or more technical solutions provided in this application have at least the following beneficial effects:

[0007] By pre-deploying structured light sources and image acquisition devices in the inspection area and setting preset calibration parameters, the accuracy of the entire inspection system can be ensured. These calibrations guarantee that the light from the source accurately illuminates the part surface, and that the image acquisition device precisely captures the deformed stripe pattern, reducing measurement inaccuracies caused by equipment errors or viewing angle issues. The structured light source projects a preset structured light sequence onto the part to be inspected in a sequential manner. The deformation of the stripe pattern is closely related to the surface shape of the part. By capturing these deformed stripes in real time and extracting the structured light deformation information, the image acquisition device can accurately obtain the geometric information of each point on the part surface. Through three... The triangulation method, based on structured light deformation information and preset calibration parameters, accurately converts two-dimensional information in an image into three-dimensional coordinates, enabling the acquisition of complete three-dimensional shape data of a part from the deformed striped image. Triangulation effectively avoids angular errors or viewing angle limitations that may occur in traditional two-dimensional measurement methods, providing higher precision three-dimensional spatial coordinates. By generating three-dimensional point cloud data, it comprehensively reflects the three-dimensional structural information of the part's surface, thus enabling precise dimensional inspection. Compared with traditional dimensional inspection methods, point cloud-based dimensional inspection better captures the complex geometry of the part's surface and ensures the comprehensiveness and accuracy of dimensional inspection. Overall, this method allows for efficient, non-contact batch inspection, avoiding errors or damage that may result from physical contact, and achieving high-precision visual inspection.

[0008] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0009] Figure 1 This is a schematic flowchart of a batch size detection method for parts based on structured light projection, provided in an embodiment of this application.

[0010] Figure 2 A schematic diagram of the structure of the part batch size detection device based on structured light projection provided in the embodiments of this application.

[0011] Explanation of reference numerals in the attached diagram: Equipment deployment module 10, image acquisition module 20, 3D coordinate transformation module 30, dimension detection module 40. Detailed Implementation

[0012] This application provides a method and apparatus for batch size inspection of parts based on structured light projection, which solves the technical problem that existing part size inspection methods usually rely on physical measuring tools, which cannot quickly and effectively inspect large batches of parts accurately, resulting in reduced production efficiency.

[0013] After introducing the basic principles of this application, various non-limiting embodiments of this application will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0014] Example 1, as Figure 1 As shown in the embodiment of this application, a method for batch dimensional inspection of parts based on structured light projection is provided, the method comprising:

[0015] A structured light source and an image acquisition device are pre-deployed in the detection area, wherein the image acquisition device has preset calibration parameters.

[0016] A structured light source is deployed within the inspection area to project a pre-defined structured light sequence onto the part under inspection, such as sinusoidal or diagonal stripes. The changes in the pre-defined structured light sequence are closely related to the surface shape of the object and are used for subsequent 3D measurement and dimensional inspection. The image acquisition device is typically a camera, fixed in an appropriate position to acquire the structured light pattern on the part under inspection in real time. To ensure the accuracy of image acquisition, the image acquisition device has been calibrated using pre-defined calibration parameters to ensure accurate coordination between the structured light source and the image acquisition device in terms of spatial position and optical characteristics. These pre-defined calibration parameters include: intrinsic parameters, such as focal length and optical center position, which describe the camera's geometric characteristics; extrinsic parameters, such as the relative position and angle between the camera and the light source; and distortion parameters.

[0017] When the first part to be inspected is transported to the inspection position of the inspection area, the structured light source projects a preset structured light sequence onto the first part to be inspected in a time-division manner. At the same time, the image acquisition device acquires images of the first part to be inspected and extracts structured light deformation information based on the image acquisition results.

[0018] The first part to be inspected is transported to the predetermined inspection position via a conveyor system, such as a conveyor belt or robotic arm, ensuring that the part is within the observation range of the structured light source and image acquisition equipment. Once the part arrives at the inspection position, the structured light source begins to project multiple preset structured light sequences in a time-division manner (i.e., at different times) according to a preset sequence. These structured light sequences are light patterns generated through programming, such as sine stripes or other waveforms. There is a certain phase shift between each structured light pattern, which allows for better extraction of structured light deformation information in subsequent analysis.

[0019] While the structured light source projects the light pattern, the image acquisition device captures an image of the first part to be inspected in real time. After each light pattern is projected, the image acquisition device captures the deformation of the pattern projected onto the part surface. Since the stripes of the structured light are affected by the shape and surface texture on the part surface, the stripes in the image captured by the image acquisition device will be deformed. This deformation information contains the geometric information of the part surface.

[0020] By processing the image acquisition results, structured light deformation information is extracted. This includes preprocessing the image, such as denoising, grayscale adjustment, and image enhancement, to ensure that the stripes in the image are clearer and easier to identify. Based on the deformed stripe pattern, the phase information of the structured light stripes in the image is obtained through phase retrieval methods. This process uses Gaussian filtering, grayscale gradient analysis, and other methods to obtain the specific deformation of the stripes, unpack the phase information, and obtain high-precision structured light deformation information through subsequent calculations.

[0021] Based on the structured light deformation information and the preset calibration parameters, a three-dimensional coordinate transformation is performed using triangulation to generate three-dimensional point cloud data of the first part to be inspected.

[0022] Triangulation is a geometric method that calculates the three-dimensional coordinates of points on the surface of a part by using the known positions of the structured light source and the image acquisition device, combined with the deformation information of the projected structured light. This method assumes that there is a certain geometric relationship between the projection of the structured light in three-dimensional space and the observation angle of the image acquisition device. Through these known relationships, namely the positions and angles of the structured light source and the image acquisition device, as well as the deformation of the light pattern, the spatial position of each point on the surface of the part to be inspected can be calculated.

[0023] In images captured by image acquisition devices, the structured light stripes are deformed due to the irregularities of the part's surface, such as unevenness. These deformations are caused by the relative positions of surface points and thus contain important information about the surface geometry. By extracting the phase information of each pixel position in the image, the degree of deformation of the structured light is calculated. This information is used to infer the spatial coordinates of each point on the object's surface. Preset calibration parameters include the relative positions of the structured light source and the image acquisition device, camera intrinsic parameters, and the projection mode of the light source. These parameters are used to accurately establish the geometric relationship between the structured light source, the image acquisition device, and the part, ensuring that the information in the two-dimensional image can be accurately converted into three-dimensional spatial coordinates during the measurement process.

[0024] Using known geometric relationships and phase information, and employing the principle of triangulation, the two-dimensional image coordinates corresponding to each pixel are mapped to three-dimensional spatial coordinates. This process involves the transformation between the camera coordinate system and the world coordinate system. Each calculated three-dimensional coordinate point (X, Y, Z) corresponds to a point on the surface of the part. All these points are combined to form three-dimensional point cloud data, which provides spatial location information for each surface point.

[0025] Based on the three-dimensional point cloud data, size detection is performed to obtain the first size information of the first part to be inspected.

[0026] The generated 3D point cloud data contains the spatial position information of each point on the surface of the part. For dimensional inspection, the 3D point cloud data is first processed and analyzed to extract the geometric features of the part, including its contour, aperture, edges, diameter, curvature, etc. Based on the extracted geometric features, the dimensions are calculated. For example, if the diameter of the part is to be detected, the diameter of the circle is calculated by analyzing points in the point cloud data related to the circular contour. For complex shapes, a fitting algorithm is used to fit the point cloud data into a geometric model, such as a circle, ellipse, or straight line. Then, the dimensions are calculated based on the fitting results. Through geometric feature extraction and dimension calculation, the first dimensional information is obtained, including key dimensions such as the length, width, thickness, aperture, and diameter of the part.

[0027] Furthermore, it also includes:

[0028] Before the first part to be inspected is transferred to the detection area, it first passes through the part identification area; in the part identification area, the basic information of the first part to be inspected is retrieved locally, wherein the basic information of the first part to be inspected includes the first standard size information and tolerance constraint information; the structured light source and image acquisition device are deployed and calibrated according to the first standard size information, and the preset calibration parameters are generated.

[0029] The part identification area is a preprocessing area before parts enter the inspection area. Its main function is to identify the parts to be inspected and obtain their basic information, providing accurate reference data for subsequent dimensional inspection. This part identification area is equipped with high-precision image recognition systems, laser scanners, barcode or QR code scanners, etc., which can quickly read the part's markings and appearance features. If the part has markings, such as barcodes or QR codes, the image recognition system first reads the markings to obtain the part's basic information, such as model, specifications, batch number, etc.

[0030] After part identification, the basic information of the first part to be inspected is retrieved. The first standard dimension information describes the standard dimension requirements of the part in the design or production process, including key dimensions such as the length, width, thickness, diameter, and hole diameter of the part. The standard dimensions are used as the basis for subsequent dimension inspection and are used to compare with the actual measured values. The tolerance constraint information specifies the allowable error range of each dimension of the part, that is, the allowable dimensional deviation of the part in the production process. The tolerance constraint ensures that the part can work normally and meet the requirements even if there are certain errors in actual production.

[0031] Based on the retrieved first standard dimension information, the settings of the structured light source and image acquisition device are adjusted. This step mainly optimizes parameters such as the illumination angle and distance of the light source according to the actual size of the part, and ensures that the image acquisition device can accurately capture the deformation of the structured light pattern. For example, if the part is large, the projection range of the structured light source is increased, or the position of the image acquisition device is adjusted to ensure that its field of view covers the entire part; if the part is small, the light source and image acquisition device need to be closer together. The calibration process includes: light source calibration, ensuring that the output light pattern of the structured light source can precisely match the capture angle of the image acquisition device. Calibration can ensure that the geometric relationship between the light source and the image acquisition device is optimized, thereby improving the accuracy of the data; camera calibration, using a standard object of known size, such as a calibration board, to calibrate the intrinsic and extrinsic parameters of the image acquisition device, including the camera's focal length, optical center, distortion coefficient, etc., so that the structured light fringes in the subsequent images can accurately reflect the three-dimensional shape of the part. Through the calibration of the light source and image acquisition device, a set of preset calibration parameters is generated, including the position and angle of the light source, the parameters of the image acquisition device, the projection method of the light pattern, etc.

[0032] Furthermore, the preset structured light sequence consists of four sinusoidal stripes generated by a four-step phase-shifting method based on the first standard size information, wherein there is a preset phase shift between every two sinusoidal stripes.

[0033] The four-step phase-shifting method obtains the phase information of each pixel by projecting four sinusoidal fringe patterns with known phase offsets. Through four different phase offsets, it overcomes the periodic blurring problem caused by a single fringe pattern, making phase extraction more accurate. The generation of the four sinusoidal fringes is based on a first standard size information. This means that the fringe parameters, such as frequency, period, wavelength, and phase offset, are customized according to the actual size of the part, ensuring that the fringes cover the surface of the part and match its geometry. For example, larger parts use lower frequency fringes, and smaller parts use higher frequency fringes. The four fringes are sinusoidal in shape, with periodic brightness changes. The shape, frequency, and phase parameters of the fringes affect the accuracy of subsequent phase recovery and 3D reconstruction. Phase offset refers to the phase difference between each fringe and the previous fringe. In the four-step phase-shifting method, there is a preset phase offset between every two sinusoidal fringes, usually set to 90 degrees, or π / 2 radians. This offset is achieved by adjusting the projection light source or through electronic signal control.

[0034] Furthermore, the method includes:

[0035] Before the first part to be inspected is conveyed to the inspection area, the structured light source projects the four sinusoidal stripes onto the inspection position at preset time intervals, and the image acquisition device acquires images of the inspection position to obtain the original stripe image sequence; after the first part to be inspected is conveyed to the inspection position in the inspection area, the structured light source projects the four sinusoidal stripes onto the first part to be inspected at preset time intervals, and the image acquisition device acquires images of the first part to be inspected to obtain the first stripe image sequence.

[0036] Before the first part to be inspected reaches the inspection area, the structured light source projects four sinusoidal fringes sequentially onto the inspection position according to the four-step phase-shifting method. The projection time interval of the four sinusoidal fringes needs to be precisely controlled to ensure that the images captured at different time points of each fringe projection do not overlap. These time intervals are typically on the order of milliseconds, ensuring that the image acquisition device can capture information of different fringe patterns frame by frame. After each fringe is projected, the image acquisition device captures an image of the inspection position in real time, obtaining the original fringe image sequence. These original fringe images are unaffected by the part and represent the original fringe patterns.

[0037] When the first part to be inspected is transported to the inspection area and reaches the inspection position, the structured light source projects four sinusoidal stripes onto the surface of the part at preset time intervals. At this time, the surface of the part will be illuminated by the structured light stripe pattern, and the stripes will deform according to the geometry of the part (such as protrusions, depressions, etc.). The image acquisition device takes real-time pictures of the projection of each stripe. Since the surface morphology of the part will affect the shape of the stripes, the image acquisition device will capture these deformation information and obtain the first stripe image sequence. This image sequence contains the deformation of each stripe on the surface of the part, reflecting the surface geometric features of the part.

[0038] Furthermore, the extraction of structured light deformation information based on the image acquisition results includes:

[0039] After Gaussian filtering the first stripe image sequence, the first stripe information sequence is obtained through grayscale gradient analysis; the first stripe information sequence is phase recovered to obtain the first phase map sequence; a phase constraint interval is obtained according to a preset phase offset, and the first phase map sequence is phase unpacked according to the phase constraint interval to obtain the first continuous phase; the original continuous phase of the original stripe image sequence is obtained by analogy; the difference between the first continuous phase and the original continuous phase is calculated to obtain the first phase difference, which is used as the structured light deformation information.

[0040] Gaussian filtering is an image denoising technique used to reduce noise and smooth images, thereby improving the accuracy of subsequent processing. Especially in structured light images, unwanted noise may exist in the original image due to ambient light, sensor noise, etc., affecting the extraction of stripe information. Gaussian filtering of the first stripe image sequence typically involves convolving the image with a Gaussian kernel function to smooth details and remove noise. This process helps improve the continuity and clarity of the stripe pattern, providing more accurate image data for subsequent phase calculations. After Gaussian filtering, stripe information is extracted through grayscale gradient analysis. Grayscale gradient analysis detects edges or stripes in the image by calculating changes in grayscale values. Specifically, it extracts stripe edge information by calculating the grayscale differences between adjacent regions of each pixel. This method extracts the stripe position and intensity changes of each pixel in the image, thus obtaining the first stripe information sequence. This stripe information contains information about the geometric deformation of the part's surface.

[0041] Phase retrieval is the process of extracting phase information from a fringe image. In structured light measurement, the phase information of each pixel can be calculated by analyzing the shape deformation of the fringe. Specifically, the phase map of the fringe is recovered by using mathematical models, such as Fourier transform and phase unrolling, through grayscale values ​​or fringe displacement information. Since the projection of the fringe pattern in space is affected by the shape of the part surface, the recovered phase map reflects the geometric features of the part surface. The first phase map sequence obtained through phase retrieval contains the phase information corresponding to each fringe image. The phase value of each pixel in these phase maps represents the spatial information of the corresponding position on the part surface.

[0042] In the four-step phase shifting method, due to the preset phase offset between every two stripes, such as π / 2 radians, the phase will undergo multiple periodic changes. During the phase unpacking process, in order to eliminate periodic phase jumps, such as the loop problem between 0 and 2π, a phase constraint interval is defined so that the unpacked phase is continuous and does not cross the periodic boundary. Phase unpacking refers to converting the recovered phase information into continuous phase data. According to the phase constraint interval, the unpacked phase data will not jump across 2π and maintain continuity. The unpacked phase data is called the first continuous phase, which represents the true phase information of the part surface in three-dimensional space and is the basis for subsequent three-dimensional reconstruction and dimensional inspection.

[0043] The original stripe image sequence consists of stripe patterns that are not deformed against a background without any parts. These stripe patterns only reflect the ideal state of the structured light and have no surface deformation. Therefore, they represent the preset phase between the structured light source and the camera. By performing a similar phase recovery and phase unpacking process on the original stripe image sequence, the corresponding original continuous phase is obtained, which represents the phase information of the original stripe pattern.

[0044] The difference between the first continuous phase and the original continuous phase is calculated. Since the first fringe image sequence and the original fringe image sequence correspond to the fringe deformation of the part surface and the background (without part), respectively, the phase difference reflects the degree of deformation of the part surface relative to the ideal state. The first phase difference is the structured light deformation information, which describes the deformation of the part surface. By analyzing the phase difference, the three-dimensional geometry of the part surface can be deduced. This deformation information is used for subsequent three-dimensional reconstruction, dimensional inspection, and quality assessment.

[0045] Furthermore, the step of performing size detection based on the three-dimensional point cloud data to obtain the first size information of the first part to be inspected includes:

[0046] Geometric features of the part are extracted based on the three-dimensional point cloud data, wherein the geometric features of the part include edges, apertures, diameters, curvatures, and contour lines; the dimensions are calculated based on the geometric features of the part to obtain the first dimension information.

[0047] 3D point cloud data is obtained through methods such as structured light projection and triangulation. The coordinates of each point contain the position of the part's surface in 3D space. This point cloud data records the surface shape and geometric features of the part. Extracting geometric features from 3D point cloud data involves extracting the following: edges represent the boundaries or abrupt changes in the part's surface, usually caused by sudden shape changes (such as corners or breaks). In 3D point cloud data, edges are identified by extracting gradient changes at points, typically using edge detection algorithms to find high-gradient regions. Apertures refer to the size and shape of holes on the part. In point clouds, hole extraction is achieved by analyzing the density and shape of the point cloud, identifying circular or elliptical regions, and combining this with the normal direction and geometric constraints in the point cloud to obtain the diameter or size of the hole. Diameter refers to the size of circular or cylindrical portions on the part. By fitting circular or cylindrical regions in point cloud data and using fitting algorithms such as least squares, the diameter of these regions can be accurately calculated. Curvature is a geometric feature that describes the degree of surface curvature. In point cloud data, curvature is obtained by calculating the change between the normal vector of each point in the point cloud and the surrounding points. For example, by fitting neighboring points, the curvature value of the point is obtained, which describes whether the surface is smooth or has a large curvature. Contour lines are curves that describe the surface contour of a part. In point cloud data, they are extracted by analyzing the relationship between edge points and adjacent points. For example, curve fitting is performed on the point cloud, and representative points are extracted at the surface contour of the part.

[0048] Dimensions are calculated based on the extracted geometric features of the parts. Specifically, for the edges of the parts, the length, angle, or relative position of the edges are calculated; for example, the total length of the outer contour of the part is calculated, or the distance between two edges is calculated. For the aperture, the diameter or other dimensions of the aperture are calculated by fitting circular or elliptical regions identified in the point cloud data. For example, the diameter of the aperture is obtained by fitting the equation of a circle using the least squares method, or the major and minor axes of an ellipse are calculated by fitting an ellipse. Diameter calculation is accomplished by fitting circular or cylindrical regions. By extracting points in the point cloud that conform to the characteristics of a circle, the diameter of the circle is calculated using a fitting algorithm. For cylindrical parts, the diameter and length of the cylinder can be calculated. Curvature calculation is accomplished by fitting local curved surfaces. The calculated curvature values ​​are used to evaluate the surface quality of the parts. The dimensions of the contour lines are obtained by calculating the total length of the contour lines, the maximum height of the contour, or the minimum depth of the contour. Based on the calculation results, the first dimensional information obtained includes edge length, aperture, diameter, curvature, and contour lines. This dimensional information serves as the core data for part inspection and is provided to subsequent quality control and acceptance judgment stages.

[0049] Furthermore, the method also includes:

[0050] Based on the first standard size information, a size deviation is calculated on the first size information to obtain a first size deviation; if the first size deviation meets the tolerance constraint information, the first size detection result is output as "detection passed"; if the first size deviation does not meet the tolerance constraint information, the first size detection result is output as "detection failed".

[0051] Dimensional deviation refers to the difference between the actual measured dimension and the standard design dimension. Ideally, the actual dimension should be as close as possible to the design dimension; the smaller the deviation, the better the quality of the part. The first dimensional information is compared with the first standard dimensional information, and the difference between the two is calculated to obtain the first dimensional deviation. If the first dimensional deviation is positive, the actual dimension is larger than the standard dimension; if it is negative, the actual dimension is smaller than the standard dimension; if the deviation is zero, it means that the actual dimension is completely consistent with the design dimension.

[0052] Tolerance constraints refer to the allowable dimensional variation range of a part during manufacturing. A tolerance sets a maximum permissible deviation range; the actual size of the part can fluctuate within this range, but it is still considered acceptable. Deviations exceeding this range are considered unacceptable. Tolerance constraints are specified by design standards, industry specifications, or product requirements, including upper and lower tolerance limits. The first dimensional deviation is compared with the tolerance constraints. Based on the preset tolerance range, it is determined whether the actual size meets the requirements. If the dimensional deviation is within the tolerance range, the size is considered acceptable; otherwise, it is considered unacceptable. If the first dimensional deviation meets the tolerance constraints, the first dimensional inspection result is output as "inspection passed," indicating that the part meets the dimensional requirements and can pass quality control inspection.

[0053] If the calculated first dimensional deviation exceeds the range of the tolerance constraint information, that is, the deviation value is greater than the upper limit tolerance or less than the lower limit tolerance, then the size of the part does not meet the design requirements. In this case, the output first dimensional inspection result is inspection failure, indicating that the part does not meet the quality requirements and needs to be corrected or reworked.

[0054] Furthermore, the method also includes:

[0055] Obtain the dimensional inspection results of N parts to be inspected in the same batch. Calculate the number of parts that fail inspection based on the N dimensional inspection results, where N is the batch inspection quantity. Calculate the percentage of the number of parts that fail inspection in the batch inspection quantity to obtain the batch non-conforming rate. If the batch non-conforming rate is greater than or equal to the batch non-conforming rate threshold, then transfer the N parts to be inspected to the batch rework area. If the batch non-conforming rate is less than the batch non-conforming rate threshold, then transfer the N parts to the batch conforming area.

[0056] On the production line, batch inspection is carried out on parts of the same batch (quantity N). The dimensions of each part undergo the corresponding inspection process, and the dimensional inspection results are obtained, including whether the inspection passes or fails. After the batch inspection, the inspection results of all parts are counted, and the number of parts that fail the inspection is counted, i.e., the number of unqualified parts.

[0057] The non-conforming rate represents the proportion of defective parts in a batch of parts. Specifically, it is the ratio of the number of parts that fail inspection to the number of parts inspected in the batch. The non-conforming rate reflects the quality level of the parts production process. A lower non-conforming rate means that the production process is stable and the quality control is good; a higher non-conforming rate means that there are production defects or process problems.

[0058] The batch non-conforming rate threshold is a pre-set standard used to determine whether parts in the same batch need to be reworked. This threshold is set based on production processes, quality standards, or customer requirements, representing an acceptable upper limit for the non-conforming rate. If the batch non-conforming rate is greater than or equal to the set threshold, the parts in that batch need to be reworked. These parts are sent to the batch rework area for correction. In the batch rework area, non-conforming parts are repaired or reprocessed to ensure that they meet design requirements and are thus qualified for shipment.

[0059] If the batch non-conforming rate is less than the set batch non-conforming rate threshold, it means that the quality of the parts in that batch is within the allowable range and meets the quality standards. This means that most parts have passed the dimensional inspection, meet the design requirements, and have not exceeded the tolerance range. In this case, all the parts to be inspected are sent to the batch qualified area. These parts will be considered qualified and ready to enter the production or shipping process. The batch qualified area is the area where qualified parts that have passed the inspection are finally stored. Qualified parts will be subsequently assembled, packaged, or distributed here.

[0060] Furthermore, the method also includes:

[0061] Before transferring the N parts to be inspected to the batch rework area, extract N dimensional information of the N parts to be inspected; based on the N dimensional information, perform dimensional concentration evaluation and dimensional change trend analysis; generate batch rework feedback information based on the dimensional concentration evaluation results and dimensional change trend analysis results, and perform batch rework management based on the batch rework feedback information.

[0062] Before batch rework, extract N dimensional information from N parts to be inspected. This dimensional information is obtained through dimensional inspection, including key dimensions such as length, diameter, hole diameter, and thickness, which serve as the basis for subsequent analysis.

[0063] Dimensional concentration is an indicator of the dimensional consistency of parts within a batch. It reflects whether the dimensional distribution of parts is close to standard dimensions or design requirements. High dimensional concentration means that the dimensions of the parts are relatively consistent. Dimensional concentration is evaluated by calculating the standard deviation of the part dimensions; the smaller the standard deviation, the more concentrated the dimensions of the parts and the less variation. Dimensional variation trend analysis analyzes the trend of dimensional changes of parts within a batch over time or at different stages of the production process. This helps determine if there are systemic problems in the production process. It is done by plotting time series or trend graphs of dimensions to identify whether the dimensions show a gradual increase or decrease trend across different batches. Combining dimensional concentration evaluation and dimensional variation trend analysis provides a comprehensive assessment of the quality of the batch production process. If the concentration is low or the variation trend is abnormal, it indicates problems in the production process, such as equipment failure, improper operation, or material issues.

[0064] Based on the dimensional concentration evaluation results and dimensional change trend analysis results, batch rework feedback information is generated. This feedback provides valuable guidance to production managers, helping them determine which parts or production processes require improvement. For example, based on dimensional deviations and defect rates, it identifies which parts need rework; based on part dimensional concentration evaluation and change trends, it indicates which parts have the most serious dimensional problems and require priority rework; if the dimensional change trend shows a systemic problem in a certain production stage or equipment, the feedback information will offer relevant suggestions on how to improve the production process. Through batch rework management, the production line can gradually reduce the defect rate and improve overall product quality.

[0065] In summary, the batch size detection method for parts based on structured light projection provided in this application has the following technical effects:

[0066] By pre-deploying structured light sources and image acquisition devices in the inspection area and setting preset calibration parameters, the accuracy of the entire inspection system can be ensured. These calibrations guarantee that the light from the source accurately illuminates the part surface, and that the image acquisition device precisely captures the deformed stripe pattern, reducing measurement inaccuracies caused by equipment errors or viewing angle issues. The structured light source projects a preset structured light sequence onto the part to be inspected in a sequential manner. The deformation of the stripe pattern is closely related to the surface shape of the part. By capturing these deformed stripes in real time and extracting the structured light deformation information, the image acquisition device can accurately obtain the geometric information of each point on the part surface. Through three... The triangulation method, based on structured light deformation information and preset calibration parameters, accurately converts two-dimensional information in an image into three-dimensional coordinates, enabling the acquisition of complete three-dimensional shape data of a part from the deformed striped image. Triangulation effectively avoids angular errors or viewing angle limitations that may occur in traditional two-dimensional measurement methods, providing higher precision three-dimensional spatial coordinates. By generating three-dimensional point cloud data, it comprehensively reflects the three-dimensional structural information of the part's surface, thus enabling precise dimensional inspection. Compared with traditional dimensional inspection methods, point cloud-based dimensional inspection better captures the complex geometry of the part's surface and ensures the comprehensiveness and accuracy of dimensional inspection. Overall, this method allows for efficient, non-contact batch inspection, avoiding errors or damage that may result from physical contact, and achieving high-precision visual inspection.

[0067] Example 2, based on the same inventive concept as the structured light projection-based batch part size detection method in the foregoing examples, such as... Figure 2 As shown in the figure, this application provides a batch size inspection device for parts based on structured light projection. The device includes:

[0068] The equipment deployment module 10 is used to pre-deploy a structured light source and an image acquisition device in the detection area, wherein the image acquisition device has preset calibration parameters; the image acquisition module 20 is used to, when the first part to be inspected is transferred to the detection position in the detection area, project a preset structured light sequence onto the first part to be inspected in a time-division manner by the structured light source, and simultaneously acquire images of the first part to be inspected by the image acquisition device, and extract structured light deformation information based on the image acquisition results; the three-dimensional coordinate transformation module 30 is used to perform three-dimensional coordinate transformation based on the structured light deformation information and the preset calibration parameters, and generate three-dimensional point cloud data of the first part to be inspected; the size detection module 40 is used to perform size detection based on the three-dimensional point cloud data, and obtain the first size information of the first part to be inspected.

[0069] Furthermore, the device deployment module 10 is used to perform the following operation steps:

[0070] Before the first part to be inspected is transferred to the detection area, it first passes through the part identification area; in the part identification area, the basic information of the first part to be inspected is retrieved locally, wherein the basic information of the first part to be inspected includes the first standard size information and tolerance constraint information; the structured light source and image acquisition device are deployed and calibrated according to the first standard size information, and the preset calibration parameters are generated.

[0071] Furthermore, the preset structured light sequence consists of four sinusoidal stripes generated by a four-step phase-shifting method based on the first standard size information, wherein there is a preset phase shift between every two sinusoidal stripes.

[0072] Furthermore, the image acquisition module 20 is used to perform the following operation steps:

[0073] Before the first part to be inspected is conveyed to the inspection area, the structured light source projects the four sinusoidal stripes onto the inspection position at preset time intervals, and the image acquisition device acquires images of the inspection position to obtain the original stripe image sequence; after the first part to be inspected is conveyed to the inspection position in the inspection area, the structured light source projects the four sinusoidal stripes onto the first part to be inspected at preset time intervals, and the image acquisition device acquires images of the first part to be inspected to obtain the first stripe image sequence.

[0074] Furthermore, the image acquisition module 20 is used to perform the following operation steps:

[0075] After Gaussian filtering the first stripe image sequence, the first stripe information sequence is obtained through grayscale gradient analysis; the first stripe information sequence is phase recovered to obtain the first phase map sequence; a phase constraint interval is obtained according to a preset phase offset, and the first phase map sequence is phase unpacked according to the phase constraint interval to obtain the first continuous phase; the original continuous phase of the original stripe image sequence is obtained by analogy; the difference between the first continuous phase and the original continuous phase is calculated to obtain the first phase difference, which is used as the structured light deformation information.

[0076] Furthermore, the size detection module 40 is used to perform the following operation steps:

[0077] Geometric features of the part are extracted based on the three-dimensional point cloud data, wherein the geometric features of the part include edges, apertures, diameters, curvatures, and contour lines; the dimensions are calculated based on the geometric features of the part to obtain the first dimension information.

[0078] Furthermore, the system also includes a quality inspection module for performing the following steps:

[0079] Based on the first standard size information, a size deviation is calculated on the first size information to obtain a first size deviation; if the first size deviation meets the tolerance constraint information, the first size detection result is output as "detection passed"; if the first size deviation does not meet the tolerance constraint information, the first size detection result is output as "detection failed".

[0080] Furthermore, the quality inspection module is used to perform the following operation steps:

[0081] Obtain the dimensional inspection results of N parts to be inspected in the same batch. Calculate the number of parts that fail inspection based on the N dimensional inspection results, where N is the batch inspection quantity. Calculate the percentage of the number of parts that fail inspection in the batch inspection quantity to obtain the batch non-conforming rate. If the batch non-conforming rate is greater than or equal to the batch non-conforming rate threshold, then transfer the N parts to be inspected to the batch rework area. If the batch non-conforming rate is less than the batch non-conforming rate threshold, then transfer the N parts to the batch conforming area.

[0082] Furthermore, the quality inspection module is used to perform the following operation steps:

[0083] Before transferring the N parts to be inspected to the batch rework area, extract N dimensional information of the N parts to be inspected; based on the N dimensional information, perform dimensional concentration evaluation and dimensional change trend analysis; generate batch rework feedback information based on the dimensional concentration evaluation results and dimensional change trend analysis results, and perform batch rework management based on the batch rework feedback information.

[0084] Through the foregoing detailed description of the method for batch size inspection of parts based on structured light projection, those skilled in the art can clearly understand the batch size inspection device for parts based on structured light projection in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to in the method section.

[0085] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting the dimensions of a part in a batch based on structured light projection, characterized in that, The method comprises: pre-disposing a structured light source and an image acquisition device in a detection area, wherein the image acquisition device has a preset calibration parameter; after a first part to be detected is conveyed to a detection position of the detection area, the structured light source projects a preset structured light sequence to the first part to be detected in time, while the image acquisition device acquires images of the first part to be detected, and structured light deformation information is extracted according to the image acquisition result; three-dimensional coordinate conversion is performed based on a triangulation method according to the structured light deformation information and the preset calibration parameter, and three-dimensional point cloud data of the first part to be detected is generated; dimension detection is performed based on the three-dimensional point cloud data, and first dimension information of the first part to be detected is acquired; before the first part to be detected is conveyed to the detection area, the first part to be detected passes through a part recognition area; in the part recognition area, first part to be detected basic information is locally called, wherein the first part to be detected basic information comprises first standard dimension information and tolerance constraint information; the structured light source and the image acquisition device are disposed and calibrated according to the first standard dimension information, and the preset calibration parameter is generated; the preset structured light sequence is four sinusoidal fringes programmed by a four-step phase shifting method according to the first standard dimension information, wherein a preset phase offset exists between each two sinusoidal fringes; before the first part to be detected is conveyed to the detection area, the structured light source projects the four sinusoidal fringes to the detection position in time according to a preset time interval, while the image acquisition device acquires images of the detection position, and an original fringe image sequence is obtained; after the first part to be detected is conveyed to the detection position of the detection area, the structured light source projects the four sinusoidal fringes to the first part to be detected in time according to a preset time interval, while the image acquisition device acquires images of the first part to be detected, and a first fringe image sequence is obtained.

2. The structured light projection based part batch dimension detection method of claim 1, wherein, the structured light deformation information is extracted according to the image acquisition result, which comprises: after the first fringe image sequence is subjected to Gaussian filtering, first fringe information sequence is obtained through gray value gradient analysis; phase recovery is performed on the first fringe information sequence, and a first phase diagram sequence is obtained; a phase restriction interval is obtained according to the preset phase offset, phase unwrapping is performed on the first phase diagram sequence according to the phase restriction interval, and first continuous phase is obtained; original continuous phase of the original fringe image sequence is obtained by analogy; a first phase difference is calculated by taking the difference between the first continuous phase and the original continuous phase, and the first phase difference is taken as the structured light deformation information.

3. The structured light projection based part batch dimension detection method of claim 2, wherein, the dimension detection based on the three-dimensional point cloud data and the acquisition of the first dimension information of the first part to be detected comprise: part geometric features are extracted according to the three-dimensional point cloud data, wherein the part geometric features comprise edges, hole diameters, diameters, curvatures and contour lines; dimension calculation is performed according to the part geometric features, and the first dimension information is obtained.

4. The structured light projection based part batch dimension detection method of claim 1, wherein, the method further comprises: a first dimension deviation is calculated by taking the difference between the first dimension information and the first standard dimension information based on the first standard dimension information. If the first size deviation conforms to the tolerance constraint information, output a first size detection result as a detection pass; If the first size deviation does not conform to the tolerance constraint information, output a first size detection result as a detection fail.

5. The structured light projection based part batch dimension detection method of claim 4, wherein, The method further comprises: Obtaining N size detection results of N parts to be detected in the same batch, and counting the number of detection fails according to the N size detection results, wherein N is the number of batch detection; Calculating the proportion of the number of detection fails in the number of batch detection to obtain a same-batch unqualified rate; If the same-batch unqualified rate is greater than or equal to a same-batch unqualified rate threshold, the N parts to be detected are transmitted to a batch rework area; If the same-batch unqualified rate is less than the same-batch unqualified rate threshold, the N parts to be detected are transmitted to a batch qualified area.

6. The structured light projection based part batch dimension detection method of claim 5, wherein, The method further comprises: Before transmitting the N parts to be detected to the batch rework area, extracting N size information of the N parts to be detected; Based on the N size information, performing size concentration evaluation and size change trend analysis; According to the size concentration evaluation result and the size change trend analysis result, generating batch rework feedback information, and performing batch rework management according to the batch rework feedback information.

7. A device for detecting dimensions of a part in a batch based on structured light projection, characterized in that, The device for implementing the part batch size detection method based on structured light projection according to any one of claims 1-6, the device comprising: An equipment arrangement module configured to arrange a structured light source and an image acquisition device in a detection area, wherein the image acquisition device has preset calibration parameters; An image acquisition module configured to, when a first part to be detected is transmitted to a detection position in the detection area, project a preset structured light sequence to the first part to be detected by the structured light source, and simultaneously acquire an image of the first part to be detected by the image acquisition device, and extract structured light deformation information according to the image acquisition result; A three-dimensional coordinate conversion module configured to, according to the structured light deformation information and the preset calibration parameters, perform three-dimensional coordinate conversion based on a triangulation method to generate three-dimensional point cloud data of the first part to be detected; A size detection module configured to perform size detection based on the three-dimensional point cloud data to obtain first size information of the first part to be detected.

Images