A 3C product computer closed-loop control method based on multi-dimensional data analysis

Abstract

The application discloses a 3C product computer closed-loop control method based on multi-dimensional data analysis, and relates to the technical field of industrial process computer control, which realizes intelligent control of automatic labeling of 3C products through five steps. In the first step, product instruction and production line information are acquired, a labeling surface is determined, and 3C size is matched in combination with the size; in the second step, product labeling points are scanned and recognized based on the instruction information, a visible laser frame is projected by driving a laser emitter, labeling points are calibrated in real time, and the position is ensured to be accurate; in the third step, necessary product residence time is obtained in combination with the labeling point position and the production line information, and operation completion is ensured; in the fourth step, labeling is performed within the residence time, single-time quality is detected in real time, success rate within an interval statistical period is calculated, and the success rate is compared with a threshold value; and in the fifth step, if the success rate is lower than the threshold value, the residence time is automatically optimized, and parameters are reconfigured; if the success rate is higher than the threshold value, current parameters are kept to run, adaptive adjustment is realized, and the demand for efficient and high-quality labeling is met.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial process computer control, and more particularly to a 3C product computer closed-loop control method based on multi-dimensional data analysis. BACKGROUND

[0002] The traditional labeling system is a trigger type labeling operation process based on a basic sensor. The pre-stage equipment debugging and parameter presetting need to be completed to ensure the accurate connection of the subsequent labeling action. First, the sensor is fixed at the preset detection position of the production line conveyor belt to ensure that the detection path is not blocked and the product signal can be stably captured. At the same time, the labeling machine is fixed to ensure that the labeling head is aligned with the specified labeling point of the conveyor belt, and the direction and position of the labeling meet the requirements. Then, the sensitivity and detection mode of the sensor are adjusted according to the physical characteristics of the product to avoid missed detection or false triggering due to product characteristic differences. At the same time, the core parameters such as specifications, delay time and labeling pressure are entered into the labeling machine control system. When the product flows through the sensor detection area with the conveyor belt, the sensor captures the product signal and immediately transmits the trigger signal to the labeling machine control system. The system starts the labeling action according to the preset delay time, and the labeling machine completes the peeling and labeling operation according to the preset parameters to accurately attach to the specified position of the product. However, it still has some shortcomings in actual use, such as 1. Rigid positioning, low flexibility and intelligence: the existing method highly depends on precise clamps, preset mechanical coordinates and other rigid hardware to achieve positioning. Once the product model is changed, the machine must be stopped and redesigned, which cannot adapt to the flexible production demand of small batch and multiple varieties; 2. Labeling quality cannot be guaranteed in real time: whether the labeling is skewed, blistering, missing or other problems can only be found by manual sampling inspection or separate offline detection equipment; 3. System parameters are static and lack self-optimization ability: all parameters of the existing system are pre-set and fixed, and the system cannot sense or automatically adjust when the production line beat fluctuates. It must be intervened and debugged afterwards. SUMMARY

[0003] In order to overcome the above-mentioned defects of the prior art, the present application provides a 3C product computer closed-loop control method based on multi-dimensional data analysis, which solves the problems raised in the background art by the following scheme, comprising:

[0004] S1, intelligent matching: the system obtains product specification information and production line information, determines the product surface according to the product specification information, and matches the 3C specifications and the size of the label wheel according to the size of the product labeling surface;

[0005] S2, face visual laser frame calibration: based on digital information processing, the collected product specification information is used to determine the labeling point, and the laser emitter is driven to project a visual laser frame to calibrate the point in real time;

[0006] S3, Stationary time calculation: According to the point and production line information, the necessary stationary time of the product in the station is calculated;

[0007] S4, Quality and success rate detection: The product is operated within the necessary stationary time, and the single labeling quality is detected in real time. The labeling success rate in the time period is calculated at intervals, and compared with the set success rate threshold;

[0008] S5, Dynamic production line adjustment: If the success rate is lower than the set threshold, the system automatically adjusts the station stay time and re-matches the labeling wheel. If the success rate remains above the threshold, the current parameters are maintained to continue running.

[0009] Preferably, among all the labeling wheels that meet the conditions of wheel surface length> widest distance and wheel surface width> narrowest distance, the smallest labeling wheel is selected.

[0010] Preferably, the labeling point position is calculated only once, and there is no difference in labeling point position for the same type of product in the same batch, which remains completely consistent.

[0011] Preferably, the labeling success rate is calculated at intervals of one hour, and compared with the set labeling success rate threshold. If the labeling success rate is lower than the set threshold, the system automatically adjusts the labeling station stay time and re-matches the labeling wheel. If the labeling success rate remains above the threshold, the current parameters are maintained to continue running.

[0012] Technical effects and advantages of the present application:

[0013] 1. High-precision positioning and guidance: The digital information processing automatically identifies the product labeling surface and labeling point position, and drives the laser emitter to project a visible frame line to real-time mark the labeling point position, improving the labeling position accuracy and operation visibility;

[0014] Intelligent judgment of labeling quality: The visual system detects the adhesion coverage rate of the visible laser frame selected area and the product surface in real time, accurately judges whether the single labeling is successful, and ensures the objectivity and reliability of quality detection;

[0015] Adaptive labeling rhythm optimization: The system can dynamically calculate the reasonable stationary time of the labeling station according to the production line running state and the labeling point position, and count the labeling success rate in the interval period. When the success rate is lower than the set threshold, the labeling process is automatically optimized. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The overall structure of the present application is shown in the schematic diagram.

[0017] Figure 2 The labeling surface selection process of the present application is shown in the schematic diagram.

[0018] Figure 3 This is a schematic diagram of the labeling quality inspection process of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] like Figure 1 The method shown is a closed-loop control method for 3C products based on multidimensional data analysis, including intelligent matching of labeling surfaces, calibration of labeling surfaces with visible laser frames, calculation of dwell time at labeling stations, detection of labeling quality and success rate, and dynamic adjustment of the production line.

[0021] like Figure 2 The method shown is a closed-loop control method for 3C products based on multidimensional data analysis. It extracts the surface features of the product, excludes functional surfaces, covered and obstructed surfaces and easily detachable surfaces. If there are no remaining surfaces, the labeling surface is manually selected. If there are remaining surfaces, the labeling surface is selected from the remaining surfaces.

[0022] like Figure 3 The method shown is a closed-loop control method for 3C products based on multidimensional data analysis. It projects a laser into the area within a visible laser frame, determines whether the area within the frame is covered based on the reflected signal, and determines that the pasting is successful if the coverage rate meets the threshold.

[0023] S1. Intelligent surface matching: The system obtains product instruction manual information and production line information, determines the product surface based on the product instruction manual information, and matches the 3C specifications and label roller size according to the size of the product labeling surface.

[0024] The labeling surface is determined based on the product instruction manual information. The product surface characteristics are obtained from the product instruction manual information, and functional product surfaces, covering and obstructing surfaces, and easily detachable surfaces are excluded in sequence. The remaining surfaces are considered valid surfaces, and the plane with the largest area among the valid surfaces is selected as the labeling surface for 3C products.

[0025] Read the product manual, the functional product surface is the physical surface that the product manual indicates the largest proportion of components that realize the product functionality operation; the covering surface is the surface that is permanently or periodically covered, blocked, or itself is a movable surface in the product complete assembly state or normal use period; the easy-to-fall surface is the surface that the product manual describes that the material properties, surface treatment process, or use environment will cause the adhesive adhesion to significantly decrease, and cannot guarantee long-term firm adhesion; all the remaining surfaces are identified as effective surfaces, the area of the effective surface is identified based on digital information processing, and the largest surface is selected as the labeling surface; if there is no selectable surface after excluding the functional product surface, the covering surface, and the easy-to-fall surface, then the labeling surface is manually selected.

[0026] 3C specification is determined based on the labeling surface area, and the 3C fitting coefficient is calculated, The 3C fitting coefficient is limited in a range, and the 3C specification that matches the 3C certification area is obtained; in this example, the 3C fitting coefficient of 20% to 50% is selected; if the 3C fitting coefficient is too small, the 3C information readability is poor, and in order to adapt to the small area, the font or icon is reduced, and the key content is difficult to identify; if the 3C fitting coefficient is too large, it causes cost waste, increases production cost and pasting working hours; the smallest 3C specification that meets the 3C fitting coefficient is selected as the 3C specification for this production.

[0027] The 3C size that matches the 3C specification area is determined, and a suitable label pressing wheel is matched; the wheel face length of the label pressing wheel is greater than the maximum transverse distance between any two points on the outer contour, to ensure that the wheel body can completely cover the transverse range when rolling, and there is no edge missing area; the wheel face width of the label pressing wheel needs to be greater than the maximum longitudinal distance between any two points on the outer contour, to avoid local longitudinal not being pressed due to insufficient wheel width, and to cause bubble or wrinkle problems; in all the label pressing wheels that meet the conditions of wheel face length > widest distance and wheel face width > narrowest distance, the smallest size label pressing wheel is selected.

[0028] S2, visible laser frame calibration: based on the product manual information calculated and collected by digital information processing, the labeling point is determined, and the laser emitter is driven to project a visible laser frame to real-time calibrate the point.

[0029] The labeling point is determined based on digital information processing, an industrial camera is used to shoot an image of the labeling surface, and digital information processing is used to exclude the area where the functional components or text or trademarks are located, non-flat areas, and low adhesion areas.

[0030] The functional components or the areas where the words or trademarks are located are represented as color mutations or clear geometric contours. Based on the gray contrast exclusion, the captured label surface image is converted into a pixel graph. The original color pixel graph is subjected to gray processing. The color information of each pixel point of the color pixel graph is subjected to gray processing to obtain the gray value I(x, y) of each pixel point, wherein (x, y) represents the pixel point in the xth row and the yth column. The Canny edge detection algorithm is used. The gray value I(x, y) is inputted and subjected to convolution operation with a two-dimensional Gaussian kernel G(x, y) to obtain the smoothed pixel point , wherein the calculation formula of the two-dimensional Gaussian function is: ; the gradients of the smoothed pixel points in the horizontal and vertical directions are calculated. The partial derivative is calculated using the Sobel operator. The directional gradient G x is used for detecting the edge in the x direction. x =S x ·I s (x, y), wherein the Sobel kernel , the directional gradient G y is used for detecting the edge in the horizontal y direction. y =S y ·I s (x, y), wherein the Sobel kernel , for each pixel point (x, y), the gradient amplitude M(x, y) and the gradient direction theta(x, y) can be calculated; M(x, y) = sqrt(G x 2 + G y 2 ) ， can be approximately considered , , the angle range returned by the arctan2 function is [-pi, pi]; the gradient direction theta(x, y) of the current point is classified into one of the four main directions, namely the horizontal direction, the diagonal direction, the vertical direction and the anti-diagonal direction. In this example, the diagonal direction is selected for calculation. Along the gradient direction, the gradient amplitude M(x, y) of the current point (x, y) is compared with the gradient amplitudes of its two adjacent points. If M(x, y) is greater than or equal to the amplitudes of the two adjacent points along the gradient direction, the amplitude of the point is retained. Otherwise, the amplitude is set to zero. The refined gradient amplitude is denoted as N(x, y). The high threshold value H and the low threshold value L are set. If N(x, y) > H, it is a strong edge pixel. If L < N(x, y) <= H, it is a weak edge pixel. If N(x, y) <= L, it is a non-edge pixel. All strong edge pixels are retained. Only the weak edge pixels connected with the strong edge pixels are retained. Finally, the edge pixel set is obtained. Based on this, the edge binary graph is defined as: , if the pixel point (x, y) belongs to the edge pixel set, E(x, y) = 1, otherwise, E(x, y) = 0.

[0031] For each pixel point (x, y) in the edge binary image E(x, y), the following calculation is performed to determine the n x n neighborhood centered at (x, y), count the number of all edge images in the neighborhood, and the total number of edge pixels in the neighborhood is n=3 in this example. The edge density D is calculated as follows: 9 is the number of pixels in the 3 x 3 neighborhood; a threshold W is set to determine whether the edge density satisfies D > W. If it satisfies, the pixel belongs to the edge dense area pixel; if it does not satisfy, the pixel belongs to the non-functional area pixel; the functional component or the area where the text or trademark is located is composed of the contour of the connected edge dense area pixels, and the contour and the space enclosed by the contour together form the contour.

[0032] The non-flat area has regular texture or irregular texture. The judgment method is to emit a laser line to the object surface, capture the reflected laser line shape through the camera, and calculate the 3D contour point at the position of the laser line by using the triangulation method; move the object to obtain the 3D contour point of the entire surface, and after obtaining the 3D contour point data, calculate the standard deviation of the height of all points. The smaller the standard deviation, the smoother the surface; the larger the standard deviation, the greater the surface fluctuation. A threshold is set for the standard deviation, and the surface equal to or greater than the threshold is judged as a non-flat area.

[0033] After the camera captures the reflected laser line image of the object surface, the pixel coordinates of the camera imaging plane are obtained through the image processing algorithm, denoted as (u i , v i) , i is the i-th contour point on a single laser line, where i=1, 2, …, N, and N is the number of contour points on a single laser line; first, the pixel coordinates are converted into physical coordinates of the camera imaging plane, with the camera principal point coordinates (u0, v0) as the origin. The camera principal point coordinates are the pixel coordinates of the intersection of the camera optical axis and the imaging plane, that is, the pixel plane origin offset unit, mm; then,

[0034] The u-axis direction physical coordinate X img =(u i −u0)×α,

[0035] The v-axis direction physical coordinate Y img =(v i −v0)×β,

[0036] Wherein, a and β are the actual physical length corresponding to 1 pixel on the camera pixel scale factor imaging plane, a is the u-axis direction camera pixel scale factor, and β is the v-axis direction camera pixel scale factor; a world coordinate system is established based on the production line, the product moving direction on the production line is Y axis, the vertical to the production line plane is Z axis, and the axis vertical to Y-Z plane is X axis; the product moves along X axis or Y axis, in the example, the product moves along Y axis, that is, the moving direction is perpendicular to the laser line length direction, and the physical coordinates are converted into world coordinates; the calculation method is as follows,

[0037] Y i =Y img , the Y coordinates of all points in a single scan are the same, which are given by the mobile platform positioning, Y img is calculated from the physical coordinates,

[0038] X i =k x ×X img i+bx, X img is calculated from the physical coordinates, i is the i-th contour point on a single laser line, k x , and bx is the X direction calibration coefficient, which is determined by the calibration process.

[0039] According to the angle θ between the laser plane and the camera optical axis, the camera focal length f and other parameters, the height Z i is derived by using the triangular geometric relationship; first, the light direction vector of the camera optical center to any point on the laser line is calculated, the camera optical center in the world coordinate system is (0, 0, H), wherein H is the vertical distance from the camera optical center to the product placement plane, then the projection of the direction vector of the laser line from the camera to any point on the laser line in the imaging plane satisfies: , wherein f is the camera focal length, L is the horizontal distance from the camera optical center to any point on the laser line in the X-Z plane , which is simplified as L≈H−Z i ; the laser plane satisfies the equation , θ is the angle between the laser plane emitted by the laser emitter and the camera optical axis, that is, the angle between the center axis from the camera to the object, substitute the expression of X i and arrange, and finally the calculation formula of the height Z i is obtained; ; the product is controlled to move uniformly along the Y axis of the world coordinate system, the moving step is ΔY, which is determined by the required measurement accuracy, every time the product moves a step, the calculation of Z i is repeated, and all 3D contour points (X ij , Y ij , Z ij), where j = 1, 2, …, M, M is the number of moving steps; and finally obtaining the 3D contour point cloud data of the entire object surface, and the height values of all points constitute a height data set {Z ij , i = 1, …, N, j = 1, …, M}, N is the number of contour points on a single laser line.

[0040] By calculating the standard deviation of the height data set {Z ij}, the surface fluctuation degree is quantified, the larger the standard deviation, the more uneven the surface, and the average value Z represents the reference height of the object surface, and the calculation formula is ; the standard deviation σZ of the height data is calculated, the standard deviation σZ reflects the dispersion degree of all height values relative to the average value Z, the larger the dispersion degree, the more obvious the surface fluctuation, and the flatness threshold σth is set, unit: mm, σth is determined by actual application requirements and needs to be calibrated through experiments or industry specifications; if the calculated surface height standard deviation σZ≥σth, it is determined that the surface is a non-flat area; if σZ<σth, it is determined to be a flat surface.

[0041] Based on the information in the specification, the low adhesion area is determined, and the area with a coating is determined as a low adhesion area.

[0042] The remaining area is a candidate area, if the shape of the candidate area is regular, the center point of the minimum circumscribed rectangle is used as the labeling point, if the shape of the area is irregular, the geometric centroid of the area is used as the labeling point, and when the labeling operation is performed, the labeling point is accurately aligned with the center; the visible laser frame is projected with the labeling point as the center, and the visible laser frame is the visualization of the 3C contour, that is, the shape and size of the visible laser frame completely match the contour of the selected 3C, and the determination of the labeling point is only calculated once, and there is no difference in the labeling point of the same type of product in the same batch, and it remains completely consistent.

[0043] S3, the station staying time is calculated: according to the point and the production line information, the necessary staying time of the product in the station is calculated; the calculation formula of the necessary staying time is , wherein t1 is the product flip alignment time, t2 is the bonding operation time, t3 is the labeling quality inspection time, and t4 is the production cycle redundancy time; the product is flipped and adjusted to determine that the labeling surface is directly opposite the labeling machine, and the time required for the labeling surface to be perpendicular to the production line is the product flip alignment time, and the calculation formula is , wherein θ is the required flip angle of the product, unit: degree, and ω is the rated angular velocity of the equipment flip machine, unit: degree / s; the total time of the stripping and conveying to the product surface is the bonding operation time, and the calculation formula is , wherein L is the length of a single sheet, in mm, v1 is the conveying speed of the peeling machine, in mm / s, d is the effective pressing stroke of the pressing machine, in mm, v2 is the rated pressing speed of the pressing machine, in mm / s, 0.2 is the time set in this example for conveying to the product surface, in s; the time consumed for detecting the labeling quality based on digital information processing is the labeling quality inspection time, the buffer time reserved for responding to production line fluctuations is the production rhythm redundancy time, the labeling quality inspection time and the production rhythm redundancy time are empirical values determined after statistical analysis and practical verification based on historical data accumulated from the operation of the production line; the production line information includes the product required turning angle, the rated angular velocity of the equipment turning machine, the conveying speed of the peeling machine, the effective pressing stroke of the pressing machine, and the rated pressing speed of the pressing machine, and is obtained by the intelligent matching step of the labeling surface.

[0044] S4, quality and success rate detection: the product is operated for a necessary stay time, and the single labeling quality is detected in real time, the labeling success rate in the time period is calculated after a period of time, and compared with the set success rate threshold.

[0045] The labeling quality detection method is that after completing the labeling operation, the visible laser frame selection area is divided into a plurality of grid areas and laser is projected, if the reflected signal intensity in the grid area is I0, the grid area is covered, if the reflected signal intensity in the grid area is I, the grid area is not covered, the covered area in the visible laser frame selection area is calculated, and the covered area is equal to the area occupied by the grid with reflected signal intensity I0; the coverage rate is calculated, When the coverage rate is greater than or equal to the set threshold, the pasting is successful, and if the coverage rate is less than the set threshold, it is judged that the pasting is unqualified.

[0046] Before detection, the parameters are calibrated by sample test to provide a reference for subsequent calculation, the laser is projected on n products under the same detection environment, the reflected signal is collected and averaged, and the average value is the standard reflected intensity I0; the laser is projected on the visible laser frame selection area of n unlabeled products under the same environment, the reflected signal is collected and averaged, and the average value is the product surface reflected intensity I.

[0047] The visual laser framed area is evenly divided into N grid areas, each grid area is equal, recorded as A, the grid area must be determined according to the maximum allowed missing area, the area of a single grid cannot be greater than the maximum allowed missing area, the maximum allowed missing area = the area of the visual laser framed area x (1-set qualified coverage rate), in this example, the set qualified coverage rate is 95%; each grid area is projected with laser, and the reflected signal intensity is measured, whether each grid is covered is judged according to the reflected signal intensity, if the reflected signal intensity is I0, the grid is a covered grid, if the reflected signal intensity is I, the grid is a non-covered grid, the number of grids with reflected signal intensity I0 is counted, recorded as M, the total area S of the visual laser framed area is equal to the sum of the areas of all grids, and the coverage rate is calculated: the coverage rate is calculated according to the covered area and the total area If the coverage rate is greater than or equal to the threshold value, it is judged that the pasting is successful, in this example, the threshold value is set to 97%, that is, the product with an area of 97% or more in the visual laser framed area is pasted successfully.

[0048] The labeling success rate is the proportion of pasted successfully products to the total pasted products in a period of time, in this example, the labeling success rate is detected every hour, and the proportion of the number of products pasted successfully in an hour to the total number of products pasted in an hour is calculated.

[0049] S5, dynamic production line adjustment: if the success rate is detected to be lower than the set threshold value, the system automatically adjusts the station residence time and re-matches the labeling wheel, if the success rate is maintained above the threshold value, the current parameters are kept to continue running.

[0050] Secondly, the drawings of the disclosed embodiments only involve the structures involved in the disclosed embodiments, other structures can refer to the usual design, and in the case of no conflict, the same embodiment and different embodiments of the present application can be combined with each other;

[0051] Finally, the above only describes the preferred embodiments of the present application, and is not used to limit the present application, any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A 3C product computer closed-loop control method based on multi-dimensional data analysis, characterized in that, The method comprises the following steps: S1, surface intelligent matching: the system obtains product specification information and production line information, determines the labeling surface according to the product specification information, and matches the 3C specification and the size of the labeling wheel according to the size of the product labeling surface; S2, surface visual laser frame calibration: based on digital information processing, the product specification information collected is determined to determine the labeling point, and the laser emitter is driven to project a visual laser frame to calibrate the labeling point in real time; the visual laser frame is a visual representation of the 3C contour, that is, the shape and size of the visual laser frame completely match the contour of the selected 3C specification; S3, calculation of the necessary residence time of the product in the station: according to the labeling point and the production line information, the necessary residence time of the product in the station is calculated; S4, quality and success rate detection: the product is operated within the necessary residence time, and the single labeling quality is detected in real time, and the labeling success rate within a period of time is calculated, and compared with the set success rate threshold; The necessary stay time calculation formula is Wherein, t1 is the product flip alignment time consumption, t2 is the fitting operation time consumption, t3 is the label quality inspection time consumption, and t4 is the production beat redundancy time consumption; the label quality detection method is that after completing the label operation, the visual laser frame selection area is divided into a plurality of grid areas and laser is projected, if the grid area has a reflected signal intensity I0, the grid area is covered, if the grid area has a reflected signal intensity I, the grid area is not covered, the covered area in the visual laser frame selection area is calculated, and the covered area is equal to the area occupied by the grid with the reflected signal intensity I0; the coverage rate is calculated, If the coverage rate is greater than or equal to a set threshold, the pasting is successful, and if the coverage rate is less than the set threshold, it is judged that the pasting is unqualified. S5, dynamic production line adjustment: if the success rate is lower than the set threshold, the system automatically adjusts the station residence time and re-matches the labeling wheel, and if the success rate is maintained above the threshold, the current parameters are kept to continue running.

2. The 3C product computer closed-loop control method based on multi-dimensional data analysis according to claim 1, characterized in that: The product labeling surface is determined based on the product specification information, the product surface features are obtained from the product specification information, the functional product surface, the covered and shielded surface, and the easily falling surface are excluded in turn, the remaining surface is considered as an effective surface, the largest plane in the effective surface is selected as the 3C labeling surface.

3. The 3C product computer closed-loop control method based on multi-dimensional data analysis according to claim 2, characterized in that: The 3C specification is determined based on the labeling surface area, and the 3C fitting coefficient is calculated, The 3C fitting coefficient is limited in a range, and the 3C specification of the matching 3C area is obtained.

4. The 3C product computer closed-loop control method based on multi-dimensional data analysis according to claim 1, characterized in that: The labeling point is determined based on digital information processing, an industrial camera is used to shoot an image of the labeling surface, digital information processing is used to exclude the area where functional components, text or trademarks are located, non-flat areas, low adhesion areas, the remaining area is a candidate area, if the candidate area shape is regular, the center point of the minimum enclosing rectangle is used as the labeling point, if the area shape is irregular, the geometric centroid of the area is used as the labeling point, and when the labeling operation is performed, the labeling point is accurately aligned with the center.

5. The 3C product computer closed-loop control method based on multi-dimensional data analysis according to claim 1, characterized in that: The labeling success rate is the proportion of the products successfully labeled in a period of time to the total labeling products.

Images