Backboard straightness detection method and device

Abstract

The invention relates to the technical field of backboard straightness detection, in particular to a backboard straightness detection method and device, and the method comprises the steps: setting a theoretical value of a profiling block, placing the profiling block on a backboard positioning table, scanning a measured value of the profiling block, comparing the theoretical value with the measured value, and completing the detection and verification. The method comprises the following steps: putting a back plate into a back plate positioning table, acquiring three-dimensional coordinates of main and auxiliary positioning angles of the back plate, and fitting an assembly reference plane based on the three-dimensional coordinates; continuously scanning the edge of the backboard, identifying a functional area of the backboard, obtaining at least one piece of continuous contour point cloud data avoiding the edge of the functional area, projecting the continuous contour point cloud data to an assembly reference plane, obtaining a series of two-dimensional projection points, and fitting to obtain an assembly boundary line; and calculating the distances from all the two-dimensional projection points to the assembly boundary line, obtaining the maximum inward invasion amount and the maximum outward warping amount, and obtaining a straightness detection result according to the maximum inward invasion amount and the maximum outward warping amount. According to the invention, the stability and accuracy of backboard straightness detection are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of backboard straightness detection, and particularly relates to a backboard straightness detection method and device. BACKGROUND

[0002] The backboard refers to the rear cover shell of a television product, which is used for packaging and protecting internal circuits and components, and is a component of the product appearance. The backboard is a large thin-walled injection molding part or a metal stamping part, and is extremely prone to slight bending and warping that is difficult to detect due to internal stress release, handling or process fluctuations. For a television backboard, unqualified edge straightness will directly result in an unsightly gap with different widths or a step feeling of uneven edges after assembly with a front frame, which seriously affects the appearance level and quality feeling of the product. Straightness out of tolerance may also cause assembly stress and even abnormal noise.

[0003] In the manufacturing process of a television backboard, existing technologies for realizing straightness detection mainly include manual gauge comparison and two-dimensional visual image method. The manual gauge comparison method has the disadvantages of low measurement efficiency, human judgment error and difficulty in fully capturing the continuous appearance of the edge, which is prone to miss local defects. The two-dimensional visual image method is often based on a simple three-point fitting or overall fitting principle, and does not fully consider the particularity of the television backboard as an assembly. The assembly reference is a virtual plane composed of multiple scattered positioning angles, and the edge includes non-matching functional areas such as heat dissipation grids and interface areas, which makes the existing method unable to accurately simulate the real assembly relationship between the backboard and the front frame.

[0004] Therefore, there is a need for a backboard straightness detection method and device oriented to assembly functions. SUMMARY

[0005] In view of at least one of the above technical problems, the present application provides a backboard straightness detection method and device, which adopts a dynamic reference establishment combined with three-dimensional full-field scanning to realize high-precision straightness detection against interference and stable identification of three-dimensional warping deformation.

[0006] The present application provides a backboard straightness detection method, which comprises the following steps: S10: setting a profiling block theoretical value, placing a profiling block on a backboard positioning table, scanning a measured value of the profiling block, comparing the theoretical value with the measured value, and completing detection verification; S20: placing a backboard into the backboard positioning table, acquiring three-dimensional coordinates of main positioning angles and auxiliary positioning angles of the backboard, and fitting an assembly reference plane based on the three-dimensional coordinates; S30: Continuously scanning the edge of the backboard to identify the heat dissipation grid area and the interface area of the backboard, denoted as a functional area, and obtaining continuous contour point cloud data of at least one edge of the backboard avoiding the functional area; S40: Projecting the contour point cloud data to the assembly reference plane to obtain a series of two-dimensional projection points, and fitting an assembly boundary line according to all the two-dimensional projection points; S50: Calculating the distance from all the two-dimensional projection points to the assembly boundary line to obtain a maximum internal invasion amount and a maximum external deviation amount, and obtaining a straightness detection result according to the maximum internal invasion amount and the maximum external deviation amount.

[0007] Further, in step S10, at least one standard geometric feature is provided on the profiling block as the theoretical value, the standard geometric feature is configured to simulate the key assembly feature of the backboard, including at least one of a high-precision plane for verifying flatness measurement, a standard height step for verifying the interface area, or a standard angle bevel for verifying the assembly edge guide angle; the measured value includes at least one of flatness, height value or angle value, which is selected corresponding to the standard geometric feature.

[0008] Further, in step S20, an assembly reference plane is fitted, including: Selecting any three non-collinear angle points from the main positioning angle and the auxiliary positioning angle to form a first spatial triangle, calculating any two edge vectors of the first spatial triangle and obtaining a first vector product; Selecting three non-collinear main positioning angles and auxiliary positioning angles different from the first spatial triangle to form a second spatial triangle, calculating any two edge vectors of the second spatial triangle and obtaining a second vector product; Calculating a comprehensive normal vector according to the first vector product and the second vector product to obtain the geometric center coordinates of all the main positioning angles and the auxiliary positioning angles; Based on the comprehensive normal vector as the normal direction of the assembly reference plane, and combining the geometric center coordinates as the reference point on the assembly reference plane, the reference equation of the assembly reference plane is determined.

[0009] Further, in step S20, an assembly reference plane is fitted, further including: Obtaining a product design digital model of the backboard, the digital model of which is pre-defined with an ideal assembly plane cooperating with the front frame; Combining the reference equation with the digital model to calculate an initial spatial transformation matrix, and based on the initial spatial transformation matrix, performing iterative calculation to obtain a final spatial transformation matrix; mapping the ideal assembly plane to a coordinate system in which the backplane positioning table is located through the final spatial transformation matrix to obtain a final assembly reference plane; In the single iteration process of the iterative calculation, the three-dimensional coordinates of the main positioning angle and the auxiliary positioning angle are transformed into a coordinate system in which the digital model is located based on the initial spatial transformation matrix, and temporary point pairs are established with theoretical angle points in the digital model. Based on all the temporary point pairs, a temporary spatial transformation matrix with the smallest distance between the temporary point pairs is calculated. Based on the temporary spatial transformation matrix, the above steps are repeated until the distance between the temporary point pairs is less than a preset threshold, and the temporary spatial transformation matrix at this time is recorded as the final spatial transformation matrix.

[0010] Further, in step S30, the following steps are included: At least one optical three-dimensional profile measurement unit is used to continuously scan the edge of the backplane. The relative translational motion between the optical three-dimensional profile measurement unit and the edge of the backplane is controlled, and based on the identification of the heat dissipation grid area and the interface area, the optical three-dimensional profile measurement unit is controlled to cross or bypass the functional area. During the scanning process, the scanning direction of the optical three-dimensional profile measurement unit and the extension direction of the edge form a non-zero angle, and during the relative motion, continuous profile point cloud data is obtained.

[0011] Further, in step S40, obtaining a series of two-dimensional projection points further includes: Based on all the two-dimensional projection points, an initial assembly boundary line is obtained by an initial fitting; The perpendicular distances of all the two-dimensional projection points to the initial assembly boundary line are calculated to form a distance data set; The statistical distribution characteristics of the distance data set are calculated, and a dynamic anomaly discrimination boundary is defined according to the statistical distribution characteristics, which distinguishes normal data points representing backplane bending from local anomaly data points caused by scratches, dirt or measurement fly points; According to the dynamic anomaly discrimination boundary, abnormal projection points are identified and removed from all the two-dimensional projection points to form a series of valid two-dimensional projection points.

[0012] Further, in step S40, defining a dynamic anomaly discrimination boundary includes: The median, upper quartile and lower quartile of the distance data set are calculated; Based on the median, the median absolute deviation of each of the perpendicular distances in the distance data set is calculated; calculating a quartile range based on the upper quartile and the lower quartile; The median, the median absolute deviation and the quartile range are taken as the statistical distribution characteristics, and the dynamic anomaly discrimination boundary is defined by linear combination.

[0013] Further, in step S40, a fitting assembly boundary line is obtained, including: Based on all the effective two-dimensional projection points, the corresponding effective two-dimensional projection points are uniformly divided into a plurality of continuous strip intervals along the edge extension direction for each edge; The effective two-dimensional projection points in each strip interval are respectively subjected to local straight line fitting to obtain a plurality of local fitting line segments; The fitting goodness of each local fitting line segment is calculated, and all the fitting goodness are sorted from high to low in value, and the local fitting line segments corresponding to the fitting goodness ranked within a preset percentage are selected as residual fitting line segments; The slopes of all the residual fitting line segments are calculated, and clustering is performed according to the slopes, and the cluster containing the largest number of residual fitting line segments is selected as a dominant consensus cluster; Based on all the dominant consensus clusters, a fitting assembly boundary line is obtained.

[0014] The application also provides a backboard straightness detection device, comprising: A backboard positioning table for carrying a backboard and detecting straightness; A verification module sets a theoretical value of a profiling block, places a profiling block on the backboard positioning table, scans the measured value of the profiling block, compares the theoretical value with the measured value, and completes detection verification; A reference fitting module places a backboard into the backboard positioning table, obtains three-dimensional coordinates of four corner points of the backboard, and fits an assembly reference plane based on the three-dimensional coordinates of the four corner points; A point cloud scanning module continuously scans the edges of the backboard to obtain continuous contour point cloud data of at least one edge of the backboard; A straight line fitting module projects the contour point cloud data to the assembly reference plane to obtain a series of two-dimensional projection points, and fits an assembly boundary line based on all the two-dimensional projection points; A result detection module calculates the distance of all the two-dimensional projection points to the assembly boundary line, obtains the maximum inward intrusion and the maximum outward warping, and obtains a straightness detection result according to the maximum inward intrusion and the maximum outward warping.

[0015] Further, the point cloud scanning module comprises: An optical three-dimensional profile measurement unit is arranged to perform relative translational movement with the edge of the back plate, and the scanning direction of the optical three-dimensional profile measurement unit is at a non-zero angle with the extension direction of the edge, and the continuous profile point cloud data is obtained during the relative movement.

[0016] The technical scheme can realize the following technical effects: The present application creates a virtual assembly fitting plane as a detection reference, and adopts a scanning avoidance function area and an anti-interference data processing strategy, successfully realizes the prior prediction of the television back plate assembly quality; the maximum internal invasion amount and the maximum external amount of the final output can directly and reliably quantify the uniformity of the gap and the surface step risk after assembly, thereby eliminating the appearance quality defects visible to the user from the source, and improving the quality control from abstract geometric measurement to assembly prediction facing user experience.

[0017] The above description is only a summary of the technical scheme of the present application, in order to more clearly understand the technical means of the present application, and can be implemented according to the content of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical scheme in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description, obviously, the drawings in the following description are only some embodiments described in the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0019] Figure 1 The flowchart of the back plate straightness detection method in the embodiment of the present application is shown in the figure. Figure 2 The logic diagram of the detection and verification in the embodiment of the present application is shown in the figure. Figure 3 The flowchart of fitting assembly reference plane in the embodiment of the present application is shown in the figure. Figure 4 The optimization flowchart of fitting assembly reference plane in the embodiment of the present application is shown in the figure. Figure 5 The flowchart of obtaining effective two-dimensional projection points in the embodiment of the present application is shown in the figure. Figure 6 The flowchart of defining a dynamic abnormality discrimination boundary in the embodiment of the present application is shown in the figure. Figure 7 The flowchart of fitting an assembly boundary line in the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] This invention provides a method such as Figures 1 to 7 The backplate straightness testing method shown includes the following steps: S10: Set the theoretical value of the contour block. Place a contour block on the back plate positioning stage. This means that the contour block has known and precise geometric dimensions and shape. Store this information as theoretical values. Then, control an optical three-dimensional contour measurement device, such as a 3D line laser scanner, to scan the contour block, obtain the contour point cloud data of the contour block surface, and extract the measured values, such as the flatness of a plane or the height of a step. Finally, compare the theoretical values ​​with the measured values ​​to complete the detection and verification, ensuring the initial accuracy of the measurement device itself.

[0023] S20: Place the backplate into the backplate positioning stage and obtain the three-dimensional coordinates of the main positioning angle and the auxiliary positioning angle of the backplate. Based on the three-dimensional coordinates of these positioning angles, fit an assembly reference plane. The conventional implementation method uses the least squares method for plane fitting, that is, find a plane that minimizes the sum of the squares of the distances from the four corner points to the plane, thereby establishing a measurement reference associated with the actual position of the backplate, effectively reducing the error introduced by the imperfect placement of the backplate.

[0024] S30: Specifically, a linear motor slide can be controlled to drive a line laser scanner to perform continuous relative movement along the edge to be measured on the back plate. During the movement, the heat dissipation grid area and interface area of ​​the back plate are identified and recorded as functional areas. The scanner continuously scans the edge of the back plate to obtain continuous contour point cloud data of the three-dimensional shape of at least one edge of the back plate that avoids the functional area. The position information of the edge in space is completely recorded. Preferably, all four edges of the back plate can be scanned simultaneously, which reduces the time required for the detection process.

[0025] S40: Project the profile point cloud data to the assembly reference plane to obtain a series of two-dimensional projection points, and fit an assembly boundary line according to all the two-dimensional projection points; a commonly used and reliable implementation is to perform linear fitting by using the least square method, that is, to find a straight line such that the sum of the squares of the perpendicular distances of all the two-dimensional projection points to the straight line is the smallest.

[0026] S50: Calculate the distances of all the two-dimensional projection points to the assembly boundary line to obtain the maximum inward intrusion and the maximum outward protrusion, and obtain the straightness detection result according to the maximum inward intrusion and the maximum outward protrusion, which can take the sum of the absolute values of the two deviation values as a specific numerical value, and output as the straightness detection result, and further determine whether the corresponding backboard straightness is qualified.

[0027] In this embodiment, the instrument is tested by using a profiling block, so that the basis of measurement is accurate. In order to restore the real assembly state of the backboard in the whole television set, a virtual assembly reference plane is calculated by using the positioning corners on the backboard which are specially used for alignment with the front frame, so that all subsequent checks are based on the effect after the backboard is assembled with the front frame. Considering that the television backboard contains functional areas such as heat dissipation grids and interfaces, which are recessed, these edges will be recorded during laser scanning, showing a sudden and huge recess or break, so it is necessary to identify and bypass the functional areas such as heat dissipation holes and interfaces which do not participate in assembly, and the focus is on determining the assembly edge which may cause cracks during assembly. Finally, the measured three-dimensional profile point cloud data is mapped to the previously established virtual assembly plane, and an assembly boundary line is fitted. By calculating the distance of each point on the actual edge to the assembly boundary line, two intuitive indicators, namely the maximum inward intrusion and the maximum outward protrusion, are obtained. Thus, the abstract straightness geometric measurement is converted into a prediction of the assembly quality, and the appearance of the backboard and the front frame during assembly is intuitively displayed to the quality inspection personnel, so that the backboard straightness detection is more specific.

[0028] The maximum inward intrusion represents the change in the gap width after assembly. During assembly, the backboard position corresponding to the maximum inward intrusion will directly touch the internal structure of the front frame, causing the gap to become extremely narrow or even zero, which may cause assembly interference, generate assembly stress, or from the appearance, a thin gap here will suddenly disappear, which is very conspicuous. The maximum outward protrusion represents the change in the height of the step, and during assembly, the backboard position corresponding to the maximum outward protrusion is farthest from the front frame, which will cause the gap to be abnormally wide, and may form a step or a difference that the user can touch and see, which seriously affects the appearance and texture.

[0029] In some embodiments of the present application, as Figure 2As shown, in step S10, at least one standard geometric feature is provided on the profiling block as a theoretical value, the standard geometric feature is configured to simulate key assembly features of the backboard, including at least one of a high-precision plane for verifying flatness measurement, a standard height step for verifying the interface area, or a standard angle bevel for verifying the assembly edge angle; the measured value includes at least one of the flatness, the height value or the angle value, which is selected corresponding to the standard geometric feature.

[0030] The high-precision plane is a large reference surface for simulating the overall backboard to be attached to the front frame, mainly used to verify the flatness measurement accuracy and the linear accuracy in the Z-axis direction of the measurement device, and is the basis for realizing uniform gaps; the standard height step simulates the step between the mounting surface of the interface area and the backboard body, and is specially used to evaluate the height direction measurement accuracy and resolution, to prevent the interface from being too high or too low due to measurement inaccuracy, thereby causing local step difference; the standard angle bevel is a guide angle or a round corner profile simulating the assembly edge of the backboard, which affects the guidance during assembly and the light and shadow effect of the gap. In actual application, at least one of the features can be selected flexibly for verification according to different accuracy requirements and application scenarios, avoiding the increase in cost caused by overdesign, and ensuring the pertinence and effectiveness of the verification; the three can also be combined to form a complete measurement system performance evaluation system, and the plane, the step and the bevel on the profiling block are measured in turn, which is equivalent to a complete measurement drill simulating the real backboard. If these features on the profiling block can be accurately measured, it means that the above measurement method can ensure the measurement effect of the real backboard.

[0031] In some embodiments of the present application, as Figure 3 As shown, in step S20, a fitting assembly reference plane is fitted, including: The main positioning angle and the auxiliary positioning angle of the backboard can be listed as four corner points, the four corner points are denoted as point A, point B, point C and point D, any three non-collinear corner points are selected from the four corner points to form a first spatial triangle, for example, point A, point B and point C can be selected to form a first spatial triangle, any two edge vectors of the first spatial triangle (such as edge vector AB and edge vector AC, the edge vector AB is obtained by subtracting the coordinates of point A from the coordinates of point B, and the edge vector AC is obtained by subtracting the coordinates of point A from the coordinates of point C) are calculated and a first vector product is obtained. Three non-collinear corner points (such as point A, point B and point D) different from the first spatial triangle are selected from the four corner points of the main positioning angle and the auxiliary positioning angle to form a second spatial triangle, any two edge vectors of the second spatial triangle (such as edge vector AB and edge vector AD) are calculated and a second vector product is obtained. The comprehensive normal vector is calculated according to the first vector product and the second vector product, preferably, the first vector product and the second vector product can be subjected to vector addition operation, and then normalization processing is performed to obtain the comprehensive normal vector, so that the geometric center coordinates of all primary positioning angles and secondary positioning angles are obtained; Based on the comprehensive normal vector as the normal direction of the assembly reference plane, and in combination with the geometric center coordinates as the reference point on the assembly reference plane, a point method type plane equation can be used to determine the reference equation of the assembly reference plane.

[0032] In an actual measurement environment, when a certain angle point has a coordinate deviation due to a slight damage, an attached contaminant or a measurement noise, the traditional least square method will pull the assembly reference plane to the direction of the abnormal point due to the pursuit of the overall optimization of all points, resulting in systematic errors, and in the embodiment, the two independent spatial triangle normal vectors are subjected to cross verification and fusion, an inherent anti-interference mechanism is constructed, the reference deviation caused by the damage or the measurement noise of the individual angle point can be effectively resisted, and the accuracy of the reference direction is theoretically ensured; the consistency of the two normal vectors is compared to automatically identify the abnormal situation of the angle point data, and reliability evaluation is provided for detection; the position of the assembly reference plane is determined through the geometric center point, the spatial positioning accuracy of the plane is balanced, the assembly reference plane represents the overall spatial posture of the back plate, and a solid and reliable reference foundation is laid for subsequent straightness detection.

[0033] In some embodiments of the present application, as shown in FIG. Figure 4 In step S20, the assembly reference plane is fitted, and further includes: The digital model of the back plate in the modeling stage of the design stage is acquired, such as a CAD digital model, the digital model includes an ideal assembly plane of the back plate and the front frame, and ideal geometric shapes and sizes of the back plate, and an initial spatial transformation matrix is calculated in combination with the reference equation; the initial spatial transformation matrix establishes a preliminary correspondence relationship from the actually measured coordinate system to the coordinate system in the digital model, and the correspondence relationship can be realized by aligning the normal vector of the assembly reference plane with the normal vector of the corresponding plane in the digital model, and by least square matching of the angle point coordinates with the theoretical angle points in the digital model.

[0034] An iterative calculation is performed based on the initial spatial transformation matrix to obtain a final spatial transformation matrix; The ideal assembly plane is mapped to the coordinate system in which the back plate positioning table is located through the final spatial transformation matrix to obtain a final assembly reference plane; In a single iteration process of the iterative calculation, the three-dimensional coordinates of the primary positioning angle and the secondary positioning angle are transformed to the coordinate system in which the digital model is located based on the initial spatial transformation matrix, and temporary point pairs are established with the theoretical angle points in the digital model. Based on all the temporary point pairs, a temporary spatial transformation matrix with the minimum distance between the temporary point pairs is calculated so that the overall distance between all the temporary point pairs is minimized, which can be achieved by singular value decomposition (SVD) method, quaternion method or least square method.

[0035] The above steps are repeated based on the temporary spatial transformation matrix, and when the distance between the temporary point pairs is less than a preset threshold value, the temporary spatial transformation matrix at this time is recorded as the final spatial transformation matrix.

[0036] In practical applications, the specific implementation of the iterative algorithm can adopt ICP (Iterative Closest Point) algorithm and its variants, and for different types of backboard structures, the threshold parameter of iteration termination can be adjusted to balance the calculation efficiency while ensuring the accuracy. Through the intelligent iterative registration mechanism, the embodiment effectively compensates for the comprehensive influence of manufacturing errors, measurement errors and clamping errors, and in the iteration process, the optimal correspondence between the measured corner points and the theoretical corner points is found, and the spatial transformation parameters are continuously optimized by minimizing the overall distance. The measurement device error calibration and compensation can find the optimal registration scheme when the backboard has manufacturing tolerances or slight deformation within the allowable range, ensuring that the assembly reference plane can reflect the actual state of the backboard and will not be disturbed by random measurement noise.

[0037] In some embodiments of the present application, in step S30, the following steps are included: At least one optical three-dimensional profile measurement unit is used to continuously scan the edge of the backboard, including but not limited to 3D line laser scanner or structured light three-dimensional scanner; The relative translational motion between the optical three-dimensional profile measurement unit and the edge of the backboard is controlled, and based on the identification of the heat dissipation grid area and the interface area, the optical three-dimensional profile measurement unit is controlled to cross or bypass the functional area. The relative translational motion can be achieved by fixing the backboard and moving the measurement unit along the edge direction by using a linear motor sliding table; or the measurement unit can be fixed, and the backboard can be moved by using a conveyor belt or a robot, both of which need to ensure that the relative motion speed is stable and controllable to ensure the uniformity and consistency of the point cloud data. Through the stored digital model, the accurate coordinates, size and boundary of the heat dissipation grid area and the interface area in the three-dimensional space can be obtained, and the scanning path is planned with the parameters of the heat dissipation grid area and the interface area recorded synchronously. In the recording process, it is necessary to identify which coordinate point represents the beginning of the functional area, and then avoid the functional area from this coordinate point and end at which coordinate point to ensure that the chaotic profile of the functional area, such as the vertical wall of the grid and the deep hole of the interface, will not be recorded in the point cloud data.

[0038] During the scanning process, the scanning direction of the optical 3D contour measurement unit forms a non-zero angle with the extension direction of the edge. When the scanning direction is perpendicular to the extension direction of the edge, the single scan line of the scanner will completely cover the entire contour section of the edge, and can simultaneously capture the top surface, edge line and side surface information of the edge. When the scanning direction forms a 45-degree angle with the extension direction of the edge, the scan line will pass through the edge obliquely, which can provide a denser point cloud distribution. In practical applications, the specific value of this angle can be adjusted according to the geometric characteristics of the edge and the accuracy requirements. The range is usually selected between 45° and 90°.

[0039] Finally, during the relative motion, the measurement unit continuously scans at a fixed frequency. Each scanning cycle generates a contour line containing hundreds of three-dimensional coordinate points. As the relative motion continues, these continuous contour lines connect end to end in space, forming a contour point cloud data that completely covers the edge to be measured, providing an accurate raw data basis for subsequent straightness evaluation.

[0040] In some embodiments of the present invention, such as Figure 5 As shown, in step S40, obtaining a series of two-dimensional projection points also includes: An initial fit is performed based on all two-dimensional projection points. The classic least squares method can be used to solve an optimization problem that minimizes the sum of the squares of the distances from all points to the line, thus obtaining an initial assembly boundary line. The initial assembly boundary line is represented by a slope-intercept equation and serves as a reference benchmark for subsequent calculations.

[0041] The vertical distance from all two-dimensional projection points to the initial assembly boundary line is calculated using the point-to-line distance formula, forming a distance dataset that forms the basis for subsequent statistical analysis.

[0042] The statistical distribution characteristics of the distance dataset are calculated, and a dynamic anomaly discrimination boundary is defined based on these characteristics. This boundary distinguishes between normal data points representing backplate bending and localized anomaly data points caused by scratches, dirt, or measurement drift. Normal data points representing backplate bending are caused by mold wear, material internal stress, systematic process fluctuations, etc., typically exhibiting large-scale, regular contour changes. These are genuine manufacturing errors that directly affect assembly gaps and are what need to be captured and evaluated in subsequent processing. Localized anomaly data points caused by scratches, dirt, or measurement drift occur during production flow and handling. Or, if these abnormal points are generated accidentally during the measurement process, they may appear as isolated spikes or valleys. These are accidental damages that do not reflect the true manufacturing quality of the backplate. If they are included in the fitting process, they will severely distort the position of the assembly boundary line. Therefore, they need to be removed. After the above data classification and removal, the discrimination criteria can be automatically adjusted according to the actual distribution characteristics of each measurement data. This avoids the high false negative or false negative rates that may occur under different working conditions due to fixed thresholds. It also ensures the consistency and accuracy of abnormal point identification. It can flexibly cope with backplates of different specifications, different processing quality states, and changing environmental conditions, and has wide applicability and robustness.

[0043] Based on the dynamic anomaly discrimination boundary, abnormal projection points are identified and eliminated from all two-dimensional projection points to form a series of valid two-dimensional projection points. For each two-dimensional projection point, if its vertical distance to the initial assembly boundary line falls outside the anomaly discrimination boundary, it is identified as an abnormal projection point and eliminated; otherwise, it is retained as a valid two-dimensional projection point.

[0044] In some embodiments of the present invention, such as Figure 6 As shown, in step S40, a dynamic anomaly detection boundary is defined, including: Calculate the median, upper quartile, and lower quartile of the distance dataset. The median, as a robust centrality measure of the data distribution, effectively resists the influence of outliers. All vertical distance values ​​are sorted in ascending order, and the value in the middle position is taken as the median. When the data volume is even, the average of the two middle values ​​is taken. The upper quartile is the value at the 75th percentile of the dataset, indicating that 75% of the data are less than or equal to this value. The lower quartile is the data value at the 25th percentile. The calculation of these two quartiles is also based on the sorted data sequence and can be determined by linear interpolation.

[0045] The absolute deviation of the median from the median of each vertical distance in the distance dataset is calculated based on the median. First, the absolute deviation of each data point in the distance dataset from the median is calculated. Then, these absolute deviations are combined into a new dataset, and finally, the median of the new dataset is calculated.

[0046] The interquartile range is calculated based on the upper quartile and the lower quartile, and the interquartile range can be obtained by subtracting the lower quartile from the upper quartile, and the interquartile range reflects the dispersion degree of the middle 50% of the data and is not sensitive to extreme values.

[0047] The median, the median absolute deviation, and the interquartile range are taken as statistical distribution characteristics, and a dynamic anomaly discrimination boundary is defined by linear combination.

[0048] The median is taken as the boundary center reference, and a certain range is expanded to both sides, and the expansion range is determined by the median absolute deviation and the interquartile range. The two indicators are linearly combined according to preset weight coefficients to jointly constitute the floating interval of the boundary. The median absolute deviation mainly contributes to the consideration of the dispersion degree of the distance of the data center, and the interquartile range provides a reference for the overall distribution range of the data.

[0049] In practical applications, various boundary definition strategies can be used, for example, the median absolute deviation and the interquartile range are combined in a fixed proportion to determine the offset of the upper and lower limits of the boundary; or the combination weight is dynamically adjusted according to the distribution form of the data, for example, when the data distribution is relatively concentrated, the weight of the median absolute deviation is appropriately increased; when the data distribution range is wide, the weight proportion of the interquartile range is increased; when the data distribution presents skewness characteristics, the expansion range of the upper and lower limits of the boundary can be calculated respectively, and different weight coefficients are combined to adapt to the actual data distribution form.

[0050] In some embodiments of the present application, as shown in FIG. 4, Figure 7 In step S40, a fitting boundary line is obtained, including: Based on all valid two-dimensional projection points, the corresponding valid two-dimensional projection points are uniformly divided into a plurality of continuous strip intervals along the extension direction of each edge. The extension direction of the edge can be obtained by calculating the long side direction of the minimum circumscribed rectangle of the valid two-dimensional projection points on the edge, or by principal component analysis to determine the main direction of the distribution of the valid two-dimensional projection points. Along the extension direction of the edge, all the valid two-dimensional projection points are uniformly divided into a plurality of continuous strip intervals. For example, for an edge with a length of 1200 mm, it can be divided into 20 continuous strip intervals, and each interval has a width of 60 mm. The strip intervals can be adjacent to each other, or a partial overlap region can be provided to ensure continuity.

[0051] Local straight line fitting is performed on the valid two-dimensional projection points in each strip interval to obtain a plurality of local fitting line segments. The global fitting problem that may be affected by local defects is converted into a plurality of relatively pure local fitting problems. Each local fitting line segment generated by each strip interval contains information such as slope, intercept, and fitting quality.

[0052] The goodness of fit of each local fitting line segment is calculated, and all the goodness of fit is ranked from high to low in value, and the local fitting line segment corresponding to the goodness of fit ranked within a preset percentage is selected as a remaining fitting line segment; the calculation of the goodness of fit can be based on the explanation ability of the local fitting line segment to the data points in the interval, for example, by calculating the determination coefficient, or by calculating the dispersion degree of the distance of all points in the interval to the line segment; the lower the dispersion degree, the better the fitting effect of the line segment on the data points in the interval; then all the local fitting line segments are sorted according to their goodness of fit from high to low, and the line segments ranked within the preset percentage range are selected as the remaining fitting line segments; it can be set to keep only the line segments ranked in the top 70% in terms of goodness of fit, and automatically eliminate the last 30% of line segments with poor quality.

[0053] The slopes of all the remaining fitting line segments are calculated, and clustering is performed according to the slopes, and the cluster containing the most remaining fitting line segments is selected as the leading consensus cluster; the clustering algorithm can select K-means clustering, DBSCAN clustering and other unsupervised learning methods; through clustering, the line segments with similar slopes are classified into the same category, then the number of line segments contained in each category is counted, and the category containing the most line segments is selected as the leading consensus cluster, ensuring that the final result can represent the common trend of most high-quality intervals.

[0054] The arithmetic mean of the slopes and intercepts of all the remaining fitting line segments can be calculated based on all the leading consensus clusters, and an assembly boundary line is fitted.

[0055] The television backboard is large in size and relatively soft in material, and a problem is prone to occur in production, that is, there is no obvious defect in the overall profile, but there is obvious bulging or depression in a corner. If a straight line is directly fitted by using all points, the local bulging will deviate the whole line, resulting in that the backboard qualified as a whole is misjudged as unqualified due to a local defect; or the real risk of the local defect is hidden in the overall data and is difficult to be directly recognized, thereby causing the occurrence of defect misjudgment and omission. By equally dividing a complete edge of the television backboard into a plurality of strip intervals along the length of the edge, and independently performing straight line fitting operation on the data points in each strip interval, a section reference line representing the trend of the strip interval itself is obtained, and the fitting goodness of each section reference line is calculated, that is, a fitting goodness sorting screening mechanism is introduced to identify and eliminate those low-quality intervals greatly affected by local defects, so as to ensure that the subsequent analysis is based on reliable data. The slope-based clustering analysis further strengthens this advantage, effectively avoids the overall deviation caused by a small number of abnormal intervals by finding the consensus direction of most high-quality intervals, reflects the overall and real bending trend of the backboard, and can flexibly cope with various complex industrial detection scenes, whether it is to process the edge with burrs or to cope with the backboard with local deformation, the stable detection performance can be maintained.

[0056] The application also provides a backboard straightness detection device, comprising: A backboard positioning table for carrying the backboard and detecting the straightness; A verification module for setting a theoretical value of a profiling block, placing a profiling block on the backboard positioning table, scanning the measured value of the profiling block, comparing the theoretical value with the measured value, and completing the detection verification; A reference fitting module for placing the backboard into the backboard positioning table, obtaining the three-dimensional coordinates of the four corner points of the backboard, and fitting an assembly reference plane based on the three-dimensional coordinates of the four corner points; A point cloud scanning module for continuously scanning the edge of the backboard to obtain continuous contour point cloud data of at least one edge of the backboard; A straight line fitting module for projecting the contour point cloud data to the assembly reference plane to obtain a series of two-dimensional projection points, and fitting an assembly boundary line according to all the two-dimensional projection points; A result detection module for calculating the distance from all the two-dimensional projection points to the assembly boundary line, obtaining the maximum inward intrusion and the maximum outward protrusion, and obtaining the straightness detection result according to the maximum inward intrusion and the maximum outward protrusion.

[0057] The backboard straightness detection device in the application can effectively implement the backboard straightness detection method, and the technical effects are as described in the above embodiments, which will not be repeated here.

[0058] In some embodiments of the application, the point cloud scanning module comprises: The optical three-dimensional profile measurement unit is in relative translational motion with the edge of the backplane, and the scanning direction of the optical three-dimensional profile measurement unit is at a non-zero angle with the extension direction of the edge, and continuous profile point cloud data is obtained during the relative motion.

[0059] Similarly, the above optimization scheme of the backplane straightness detection device can also be correspondingly implemented to achieve the optimization effect of the backplane straightness detection method, which will not be described here again.

[0060] Although the present application has been described in connection with specific embodiments thereof, it will be evident for a person skilled in the art that various modifications and combinations can be made without departing from the spirit and scope of the application. Accordingly, the present description and drawings are merely illustrative of the present application defined in the appended claims and do not limit the scope of the present application in any manner. Obviously, various modifications and changes can be made to the present application by those skilled in the art without departing from the scope of the present application. Thus, it is intended that the present application cover any and all modifications, variations, combinations or equivalents that fall within the scope of the present application and its equivalents.

Claims

1. A backplane straightness detection method, characterized in that, The method comprises the following steps: S10: setting a theoretical value of a profiling block, placing a profiling block on a backboard positioning table, scanning a measured value of the profiling block, comparing the theoretical value with the measured value, and completing detection verification; S20: placing the backboard into the backboard positioning table, obtaining three-dimensional coordinates of the main positioning angle and the auxiliary positioning angle of the backboard, and fitting an assembly reference plane based on the three-dimensional coordinates; S30: continuously scanning the edge of the backboard, identifying the heat dissipation grid area and the interface area of the backboard, recording as a functional area, and obtaining at least one continuous contour point cloud data of the edge of the backboard avoiding the functional area; S40: projecting the contour point cloud data to the assembly reference plane to obtain a series of two-dimensional projection points, and fitting an assembly boundary line according to all the two-dimensional projection points; S50: calculating the distance from all the two-dimensional projection points to the assembly boundary line to obtain the maximum internal invasion and the maximum external warping, and obtaining the straightness detection result according to the maximum internal invasion and the maximum external warping.

2. The backboard straightness detection method according to claim 1, characterized by, In step S10, at least one standard geometric feature is arranged on the profiling block as the theoretical value, and the standard geometric feature is configured to simulate the key assembly features of the backboard, including at least one of a high-precision plane for verifying flatness measurement, a standard height step for verifying the interface area, or a standard angle bevel for verifying the assembly edge guide angle; the measured value includes at least one of flatness, height value or angle value, which is selected corresponding to the standard geometric feature.

3. The backboard straightness detection method according to claim 1, wherein In step S20, fitting an assembly reference plane comprises: selecting any three non-collinear angle points from the main positioning angle and the auxiliary positioning angle to form a first space triangle, calculating any two edge vectors of the first space triangle and obtaining a first vector product; selecting three non-collinear main positioning angles and auxiliary positioning angles different from the first space triangle to form a second space triangle, calculating any two edge vectors of the second space triangle and obtaining a second vector product; calculating a comprehensive normal vector according to the first vector product and the second vector product to obtain the geometric center coordinates of all the main positioning angles and the auxiliary positioning angles; determining the reference equation of the assembly reference plane based on the comprehensive normal vector as the normal direction of the assembly reference plane, and combining the geometric center coordinates as the reference point on the assembly reference plane.

4. The backboard straightness detection method according to claim 3, wherein In step S20, fitting an assembly reference plane further comprises: obtaining a product design digital model of the backboard, and the digital model is pre-defined with an ideal assembly plane cooperating with the front frame; combining the reference equation and the digital model to calculate an initial spatial transformation matrix, and performing iterative calculation based on the initial spatial transformation matrix to obtain a final spatial transformation matrix; mapping the ideal assembly plane to the coordinate system of the backboard positioning table through the final spatial transformation matrix to obtain a final assembly reference plane; The single iteration process of the iterative calculation includes: based on the initial spatial transformation matrix, transforming the three-dimensional coordinates of the main positioning angle and the auxiliary positioning angle into a coordinate system in which the digital model is located, and establishing temporary point pairs with theoretical angle points in the digital model; Based on all the temporary point pairs, a temporary spatial transformation matrix with the smallest distance between the temporary point pairs is calculated; Based on the temporary spatial transformation matrix, the above steps are repeated until the distance between the temporary point pairs is less than a preset threshold, and the temporary spatial transformation matrix at this time is recorded as the final spatial transformation matrix.

5. The backboard straightness detection method of claim 1, wherein Step S30 includes: Continuously scanning the edge of the back plate by using at least one optical three-dimensional profile measurement unit; Controlling the relative translation motion between the optical three-dimensional profile measurement unit and the edge of the back plate, and controlling the optical three-dimensional profile measurement unit to cross or bypass the functional area based on the identification of the heat dissipation grid area and the interface area; During the scanning process, the scanning direction of the optical three-dimensional profile measurement unit and the extension direction of the edge form a non-zero angle, and the continuous profile point cloud data is obtained during the relative motion.

6. The backboard straightness detection method of claim 1, wherein In step S40, obtaining a series of two-dimensional projection points further includes: Based on all the two-dimensional projection points, an initial fitting is performed to obtain an initial assembly boundary line; Calculate the perpendicular distance of all the two-dimensional projection points to the initial assembly boundary line to form a distance data set; Calculate the statistical distribution characteristics of the distance data set, and define a dynamic anomaly discrimination boundary according to the statistical distribution characteristics, which distinguishes normal data points representing back plate bending from local anomaly data points caused by scratches, dirt or measurement fly points; According to the dynamic anomaly discrimination boundary, identify and remove abnormal projection points from all the two-dimensional projection points to form a series of valid two-dimensional projection points.

7. The backboard straightness detection method according to claim 6, wherein In step S40, defining a dynamic anomaly discrimination boundary includes: Calculate the median, upper quartile and lower quartile of the distance data set; Based on the median, calculate the median absolute deviation of each of the perpendicular distances in the distance data set; Based on the upper quartile and the lower quartile, calculate the interquartile range; The median, median absolute deviation and interquartile range are linearly combined as the statistical distribution characteristics, and the dynamic anomaly discrimination boundary is defined.

8. The backboard straightness detection method according to claim 6, wherein In step S40, fitting to obtain an assembly boundary line includes: Based on all the valid two-dimensional projection points, the corresponding valid two-dimensional projection points are uniformly divided into a plurality of continuous strip intervals along the edge extension direction for each edge; Respectively, the valid two-dimensional projection points in each strip interval are locally linearly fitted to obtain a plurality of local fitting line segments; Calculate the fitting goodness of each local fitting line segment, and sort all the fitting goodnesses from high to low, and select the local fitting line segments corresponding to the fitting goodnesses ranked within a preset percentage from the local fitting line segments, and record them as remaining fitting line segments; Calculate the slope of all the remaining fitting line segments, and cluster according to the slope, select the cluster containing the most number of the remaining fitting line segments as the dominant consensus cluster; Fit a fitting boundary line based on all the dominant consensus clusters.

9. A backboard straightness detection apparatus using the backboard straightness detection method according to any one of claims 1 to 8, characterized by It comprises: A backboard positioning table for carrying a backboard and detecting straightness; A verification module sets a theoretical value of a profiling block, places a profiling block on the backboard positioning table, scans the measured value of the profiling block, compares the theoretical value with the measured value, and completes the detection verification; A reference fitting module places the backboard on the backboard positioning table, obtains the three-dimensional coordinates of the four corner points of the backboard, and fits a fitting reference plane based on the three-dimensional coordinates of the four corner points; A point cloud scanning module continuously scans the edge of the backboard to obtain continuous contour point cloud data of at least one edge of the backboard; A straight line fitting module projects the contour point cloud data onto the fitting reference plane to obtain a series of two-dimensional projection points, and fits a fitting boundary line based on all the two-dimensional projection points; A result detection module calculates the distance of all the two-dimensional projection points to the fitting boundary line, obtains the maximum inward intrusion and the maximum outward warping, and obtains the straightness detection result according to the maximum inward intrusion and the maximum outward warping.

10. The backboard straightness detection apparatus according to claim 9, wherein The point cloud scanning module comprises: An optical three-dimensional profile measurement unit, which moves relatively with the edge of the backboard, and the scanning direction of the optical three-dimensional profile measurement unit and the extension direction of the edge form a non-zero angle, and the continuous contour point cloud data is obtained during the relative movement.

Images