Rearview mirror flatness detection method and system

Abstract

The invention relates to the technical field of mirror surface detection, in particular to a rearview mirror flatness detection method and system, and the method comprises the steps: setting a detection region, and projecting line laser to the detection region; the standard rearview mirror is detected, and the center line of the standard rearview mirror is obtained through the center line detection model and recorded as a reference center line; the rearview mirror to be detected is detected, and the center line of the rearview mirror to be detected is obtained through the center line detection model and recorded as a detection center line Calculating the defect degree gamma; comparing the reference center line with the detection center line to obtain an inclination angle beta of the detection center line; calculating the fluctuation zeta of the detection center line; a flatness score S is calculated according to the inclination angle beta, the fluctuation zeta and the defect degree gamma, and if the flatness score S exceeds a set area, it is judged that the flatness of the rearview mirror to be detected is unqualified; the problems of low detection efficiency and low detection fineness in the prior art can be effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of mirror detection, and in particular to a method and system for detecting the flatness of a rearview mirror. Background Art

[0002] Rearview mirrors are used as an auxiliary tool for drivers to observe the conditions behind the vehicle. Their flatness is closely related to the quality of the reflected image. If the rearview mirror is deformed or skewed, the reflected image will be distorted, affecting the driver's normal observation. Therefore, the flatness of the rearview mirror needs to be tested after production. The existing testing method is usually manual measurement by using a ruler to measure the mirror surface. This method is not only inefficient, but also relies mainly on manual observation. The test results are prone to missing minor deformation and skew issues, resulting in low detection precision. Summary of the Invention

[0003] The present invention provides a rearview mirror flatness detection method and system, which can effectively solve the problems of low detection efficiency and low detection precision in the background technology.

[0004] The present invention provides a method for detecting the flatness of a rearview mirror, comprising the following steps:

[0005] Set the detection area and project the line laser into the detection area;

[0006] Test the standard rearview mirror and obtain its centerline through the centerline detection model as the reference centerline;

[0007] The rearview mirror to be tested is tested, and its center line is obtained by the center line detection model and recorded as the detection center line;

[0008] Calculate the defectivity γ; compare the reference center line and the detection center line to obtain the inclination angle β of the detection center line; calculate the fluctuation ζ of the detection center line;

[0009] The flatness score S is calculated based on the tilt angle β, the fluctuation ζ and the defect γ. If the flatness score S exceeds the set range, the flatness of the rearview mirror to be tested is judged to be unqualified;

[0010] The centerline detection model specifically includes:

[0011] Turn off the line laser, acquire an image of the detection area and calculate the background light intensity parameters;

[0012] Turn on the line laser and transport the rearview mirror horizontally through the detection area at a fixed speed, with the rearview mirror facing upward and the transport direction perpendicular to the width direction of the line laser;

[0013] The reflection angle direction of the line laser is collected at fixed intervals to obtain an image of the detection area, which is preprocessed using background light intensity parameters, and then the center of mass position of the light intensity in the image is calculated;

[0014] Eliminate all images whose light intensity centroid position cannot be calculated;

[0015] With the ordinal number as the independent variable a and the centroid position of the light intensity obtained in time sequence as the dependent variable b, starting from a=0, the center line b(a) is obtained.

[0016] Furthermore, in the centerline detection model, the detection area image is preprocessed using the background light intensity parameter, specifically including:

[0017] The direction of the rearview mirror is the x-axis and the width of the line laser is the y-axis;

[0018] When the linear laser is turned off, the light intensity of the pixel at row x and column y in the detection area image is Ibg(x,y), and the mean μbg and standard deviation σbg of all pixels in the image are calculated;

[0019] When the line laser is turned on, the light intensity of the pixel at the xth row and yth column in the detection area image is Iraw(x,y);

[0020] Then the light intensity of the pixel at row x and column y in the image obtained after image preprocessing is:

[0021] Inorm(x,y)=(Iraw(x,y)-μbg) / σbg.

[0022] Furthermore, in the centerline detection model, the center of mass position of the light intensity in the image is calculated as follows:

[0023] The light intensity of the pixel at row x and column y in the image obtained after image preprocessing is recorded as Inorm(x,y);

[0024] Calculate the segmentation threshold T = μnorm + 3σnorm;

[0025] Where μnorm is the average light intensity of all pixels in the image, and σnorm is the standard deviation of the light intensity of all pixels in the image;

[0026] Filter out the pixels in the image with light intensity greater than T and record them as laser pixels, and calculate the centroid position b:

[0027] b=(Σy·Inorm(x,y)) / (ΣInorm(x,y));

[0028] Wherein, Σy·Inorm(x,y) represents the sum of the light intensities of all laser pixels multiplied by their corresponding column number y; ΣInorm(x,y) represents the sum of the light intensities of all laser pixels.

[0029] Furthermore, the defectivity γ is calculated as follows:

[0030] The images left in the centerline detection model are numbered 0, 1, 2, ...a according to their generation time;

[0031] The laser pixel light intensity obtained from the ath image when testing the standard rearview mirror is recorded as Inorm_base(a)(x,y);

[0032] The light intensity of the laser pixel obtained in the ath image when testing the rearview mirror to be tested is recorded as Inorm_test(a)(x,y);

[0033] For all values of a, the light intensity difference of the laser pixel point is calculated at each value △Ia(x,y) = Inorm_base(a)(x,y) - Inorm_test(a)(x,y);

[0034] Set the allowed range, record the number of laser pixels whose △Ia(x,y) exceeds the allowed range as p, and the total number of laser pixels as P;

[0035] Defectivity γ = p / P.

[0036] Furthermore, the method further includes a continuity detection step, specifically:

[0037] After filtering out the laser pixel points of each image, the x-values of the laser pixel points are checked to see if they are continuous. If they are continuous, the subsequent calculations are continued; if they are not continuous, the gaps are filled in by interpolation to ensure that the x-values of the laser pixel points are continuous.

[0038] Furthermore, the inclination angle β of the detection center line is obtained as follows:

[0039] The reference center line is recorded as b0(a), and the detection center line is recorded as b1(a);

[0040] When calculating each value of a, the offset △b(a)=b1(a)-b0(a);

[0041] Tilt angle β = arctan(Δb(a)avg / 2D);

[0042] Where △b(a)avg is the average value of all △b(a); D is the distance between the laser emission point and the mirror.

[0043] Furthermore, the fluctuation degree ζ of the detection center line is calculated as follows:

[0044] The detection center line is denoted as b1(a);

[0045] Find the maximum and minimum points of the detection center line, and calculate the difference in b1 values between adjacent maximum and minimum points; set the acceptance range, and record the number of differences that exceed the acceptance range as n;

[0046] Calculate the midpoints of adjacent maximum and minimum points, then calculate the slope of the line connecting the two adjacent midpoints, and record the standard deviation of the difference between all slopes and tanβ as m;

[0047] The calculated fluctuation degree ζ = 0.19·n·m.

[0048] Furthermore, the calculation formula of the flatness score S is as follows:

[0049] S=k1·|γ| / [γ]+k2·|β| / [β]+k3·|ζ| / [ζ];

[0050] Among them, k1, k2, and k3 are all set adjustment coefficients; [γ] is the maximum allowable value of the defect degree γ; [β] is the maximum allowable value of the inclination angle β; and [ζ] is the maximum allowable value of the fluctuation degree ζ.

[0051] The present invention also provides a rearview mirror flatness detection system, comprising:

[0052] A detection rack is provided with a detection area;

[0053] The laser emitter and camera are fixedly installed above the detection area;

[0054] A conveying device for conveying a rearview mirror;

[0055] The processor is used to implement the above-mentioned rearview mirror flatness detection method.

[0056] Furthermore, in the detection area, except for the mirror surface of the rearview mirror, the top surfaces of the remaining parts are paved with a black frosted material layer.

[0057] The technical solution of the present invention can achieve the following technical effects:

[0058] The present invention uses line laser scanning and camera image acquisition to quickly obtain the condition of the rearview mirror surface, and judges and analyzes the image of the rearview mirror surface based on multiple indicators. It can significantly improve the detection efficiency and accuracy of the rearview mirror flatness, and greatly reduce the technical dependence of the operator, perfectly adapting to the needs of intelligent manufacturing. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0060] Figure 1 Flowchart of the rearview mirror flatness detection method of the present invention;

[0061] Figure 2 Schematic diagram of the structure of the rearview mirror flatness detection system of the present invention;

[0062] Figure numerals: 1. Detection frame; 2. Laser emitter; 3. Camera; 4. Conveying device. DETAILED DESCRIPTION

[0063] The basic principles and main features of the technical solution of the present invention will be described below in conjunction with the accompanying drawings in the embodiments of the present invention. The following will be described more intuitively through one or more embodiments, and the described embodiments are only part of the embodiments of the present invention, not all embodiments.

[0064] In the description of the present invention, words indicating directions or positional relationships (such as up, down, left, right, etc.) are based on the directions shown in the drawings or some conventional positional relationships. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the features referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they should not be understood as limiting the present invention.

[0065] The present invention relates to a method for detecting the flatness of a rearview mirror. Figure 1 As shown, the steps include:

[0066] S00: Set the detection area and project a line laser into the detection area. The detection area is the area that the camera can capture. When the rearview mirror enters the detection area, the line laser will shine on the rearview mirror and be reflected toward the camera. The camera can then capture an image of the shape of the line laser on the rearview mirror.

[0067] S10: Test the standard rearview mirror. The standard rearview mirror is the one with the highest pass rate. Therefore, the image generated when testing this standard rearview mirror serves as the standard image and the benchmark for subsequent tests. The standard image is tested using the centerline detection model, and its centerline is recorded as the reference centerline. In addition to providing a benchmark for subsequent tests, this step also allows the system to perform a self-test. In other words, if the standard image or reference centerline obtained after testing the standard rearview mirror has significant flaws, it indicates that there is a problem with the system itself, and the system error can be corrected in a timely manner.

[0068] S20: The rearview mirror to be tested is tested, and the image generated is the test image. The shape of the line laser in the test image can correspond to the defect on the rearview mirror to be tested. The center line is obtained through the center line detection model and recorded as the detection center line.

[0069] S30: The line laser image on the test image is compared with the line laser image on the standard image to calculate the defectivity γ. The defectivity γ reflects the proportion of pixels with significantly abnormal light intensity on the line laser image on the test image, indicating that the reflective ability of the mirror corresponding to the problem pixel is defective.

[0070] The reference centerline and the detection centerline both reflect the symmetric center of the light intensity of the line laser on the two rearview mirrors. By comparing the reference centerline and the detection centerline, the inclination angle β of the detection centerline can be obtained. The larger the value of the inclination angle β, the more likely the rearview mirror to be tested is to be skewed, which means that there is a major problem with the assembly accuracy of the rearview mirror to be tested.

[0071] Calculate the fluctuation ζ of the detection center line. The fluctuation ζ can reflect the flatness of the rearview mirror surface to be tested. The smaller the fluctuation ζ, the flatter the surface of the rearview mirror.

[0072] S40: The tilt angle β, fluctuation ζ and defect γ affect each other, so they cannot be separated and judged separately. Instead, the flatness score S needs to be calculated based on the tilt angle β, fluctuation ζ and defect γ to reflect the comprehensive flatness of the rearview mirror surface. If the flatness score S exceeds the set area, the flatness of the rearview mirror to be tested is judged to be unqualified.

[0073] The centerline detection model is mainly used to analyze the image captured by the camera to identify the laser area from the image and obtain the centerline of the laser. Its main steps include environmental calibration, mirror detection, and centerline acquisition. The specific contents of each step are as follows:

[0074] Environmental calibration: Turn off the line laser, acquire an image of the detection area, and calculate the background light intensity parameters. This allows for better differentiation between the laser area and the background area in the image captured by the camera. Because the light intensity may vary between detections, this step must be performed each time the central detection model is started.

[0075] Mirror detection: Turn on the line laser and transport the rearview mirror horizontally through the detection area at a fixed speed, with the mirror surface facing upwards and the conveying direction perpendicular to the width direction of the line laser;

[0076] The reflection angle direction of the line laser is collected at fixed intervals to obtain an image of the detection area. The interval time is determined according to the conveying speed and image size, so that the rearview mirror in the front and rear images can just move forward one row of pixels. After the rearview mirror completely passes through the detection area, multiple detection images will be obtained.

[0077] Centerline acquisition: All images are pre-processed using background light intensity parameters. After eliminating the background light effect, the laser area in each detection image can be found. The light intensity centroid position in each image is then calculated. These light intensity centroid positions are then connected to form a curve, which is the centerline.

[0078] In order to ensure that the center lines obtained each time using this center line detection model can be compared with each other, it is necessary to at least make the starting point of the center line fixed. By eliminating all images where the center point of the light intensity cannot be calculated, it can be ensured that the starting points of all center lines correspond to each other.

[0079] With the ordinal number as the independent variable a (i.e. a is: 0, 1, 2, 3...), and the position of the centroid of the light intensity obtained in chronological order as the dependent variable b (corresponding to the value of y when the width direction of the subsequent mid-shift line laser is the y-axis), starting from a=0, the center line is obtained, and the expression of the center line is b(a). The meaning of this expression is as follows: if b(1)=10, it means that when a=1 (i.e. in the second image left), the position of the centroid of the light intensity is 10.

[0080] In the centerline detection model, the specific steps of detecting the area image and preprocessing it with the background light intensity parameter include:

[0081] The direction of the rearview mirror is the x-axis and the width of the line laser is the y-axis;

[0082] When the line laser is turned off, the light intensity of the pixel at the xth row and yth column in the detection area image is Ibg(x,y). The meaning of this expression is as follows: if Ibg(50,30) = 5, it means that the light intensity Ibg(50,30) of the pixel at the 50th row and 30th column in the detection area image is 5; and the mean μbg and standard deviation σbg of all pixels in the image are calculated;

[0083] When the line laser is turned on, the light intensity of the pixel at the xth row and yth column in the detection area image is Iraw(x,y), which has the same meaning as above.

[0084] Then the light intensity Inorm(x,y) of the pixel in the xth row and yth column of the image obtained after image preprocessing is:

[0085] Inorm(x,y)=(Iraw(x,y)-μbg) / σbg;

[0086] In this way, the image Iraw(x,y) is converted into a new image Inorm(x,y). In the new image, the difference between the light intensity Inorm of the laser area and the non-laser area will become more obvious, which is convenient for subsequent distinction.

[0087] In the centerline detection model, the center of mass position of the light intensity in the image is calculated as follows:

[0088] In the above calculation, the light intensity of the pixel at row x and column y in the image obtained after image preprocessing is Inorm(x,y);

[0089] Calculate the segmentation threshold T = μnorm + 3σnorm;

[0090] Where μnorm is the average light intensity of all pixels in the image Inorm(x,y), and σnorm is the standard deviation of the light intensity of all pixels in the image Inorm(x,y). The segmentation threshold T is used as a benchmark to distinguish the laser area and the non-laser area in the image Inorm(x,y). This benchmark is calculated directly from the image Inorm(x,y). Therefore, the segmentation threshold T will be calculated to the value that best suits the image to avoid the image being affected by background light and ensure the accuracy of the distinction.

[0091] Filter out the pixels in the image with light intensity greater than T and record them as laser pixels, and calculate the centroid position b:

[0092] b=(Σy·Inorm(x,y)) / (ΣInorm(x,y));

[0093] Wherein, Σy·Inorm(x,y) represents the sum of the light intensities of all laser pixels multiplied by their corresponding column number y. For example, if the light intensity of the laser pixel is Inorm(x,y), and the range of laser pixels is x = 10, y = 1, 2, 3, 4, ... 20, then Σy·Inorm(x,y) = 1·Inorm(10,1) + 2·Inorm(10,2) + 3·Inorm(10,3) + ... + 20·Inorm(10,20);

[0094] ΣInorm(x,y) represents the sum of the light intensities of all laser pixels. The example is the same as above, that is, ΣInorm(x,y)=Inorm(10,1)+Inorm(10,2)+Inorm(10,3)+…+Inorm(10,20).

[0095] The calculation of defectivity γ is as follows:

[0096] The images left in the centerline detection model are numbered 0, 1, 2, ...a according to their generation time;

[0097] When the standard rearview mirror and the rearview mirror to be tested are tested using this method, many Inorm(x,y) images will be generated. In order to better distinguish them, the suffix is now added to them:

[0098] The laser pixel point light intensity obtained in the ath image when testing the standard rearview mirror is recorded as Inorm_base(a)(x,y). For example, Inorm_base(2)(10,20) = 5, which means that when testing the standard rearview mirror, the laser pixel point light intensity at the 10th row and 20th column in the third image is 5.

[0099] The laser pixel intensity obtained in the ath image when testing the rearview mirror to be tested is recorded as Inorm_test(a)(x,y). For example, Inorm_test(2)(10,20) = 5, which means that when testing the standard rearview mirror, the laser pixel intensity at the 10th row and 20th column in the third image is 5;

[0100] Under normal circumstances, since the rearview mirror is restricted by a fixture during transportation, the positions of the laser pixels in each image are consistent, that is, the x and y value ranges in each image are the same, and the difference is the light intensity corresponding to the laser pixel point;

[0101] For all values of a, the light intensity difference of the laser pixel point is calculated at each value: △Ia(x,y)=Inorm_base(a)(x,y)-Inorm_test(a)(x,y). For example, if 200 images are obtained when testing the standard rearview mirror, then 200 images will be obtained for the rearview mirror to be tested, that is, a=0, 1^199;

[0102] Since the laser irradiation position is the same, the area of the laser in each image is now x = 10, y = 1, 2, 3...20;

[0103] Then when a=0, we need to calculate △I0(x,y), where x=10, y=1, 2, 3, ... 20, which means we will get a total of 20 △I0(x,y); similarly, when a=1, we will get a total of 20 △I1(x,y) ... until a=199, when we will get a total of 20 △I199(x,y);

[0104] Set the allowed range, record the number of laser pixels whose △Ia(x,y) exceeds the allowed range as p, and the total number of laser pixels as P;

[0105] Defectivity γ = p / P;

[0106] This step calculates the proportion of areas with defective light intensity in all mirror areas. The larger the value, the more serious the damage to the mirror surface.

[0107] Of course, during the test, some parts of the mirror surface may not reflect light due to impurities, which will affect the correspondence during calculation and the position of the light intensity centroid. Therefore, a continuity test step is also required for detection, specifically:

[0108] After screening out the laser pixel points of each image, the x-values of the laser pixel points are checked for continuity. If they are continuous, subsequent calculations can proceed. If they are not continuous, it means that impurities in the discontinuous parts may be blocking the laser, and the gaps need to be filled by interpolation to ensure that the x-values of the laser pixel points are continuous. This will ensure a clear correspondence between the images and minimize the error in the position of the light intensity centroid.

[0109] The inclination angle β of the detection center line is obtained as follows:

[0110] When the standard rearview mirror and the rearview mirror to be tested are tested using this method, a center line b(a) will be generated. In order to better distinguish them, the corresponding suffix is added to them:

[0111] The reference center line is recorded as b0(a), and the detection center line is recorded as b1(a);

[0112] When calculating each value of a, the offset △b(a)=b1(a)-b0(a);

[0113] Tilt angle β = arctan(Δb(a)avg / 2D);

[0114] Where △b(a)avg is the average value of all △b(a); D is the distance between the laser emission point and the mirror;

[0115] The tilt angle β mainly reflects the degree of angular rotation of the test center line relative to the reference center line. This angular rotation is caused by the fact that the mirror surface of the rearview mirror to be tested and the mirror surface of the standard rearview mirror do not overlap on the plane, resulting in different light intensities of the laser reflected into the camera. Therefore, the size of the tilt angle β can be used to determine the degree of error in the assembly of the mirror surface of the rearview mirror to be tested.

[0116] The specific calculation of the fluctuation degree ζ of the detection center line is:

[0117] The detection center line is denoted as b1(a);

[0118] Find the maximum and minimum points of the detection center line, and calculate the difference in the b1 values of adjacent maximum and minimum points; set an acceptance range. If it exceeds the acceptance range, it means that the span of the detection center line at this point is large. The number of differences exceeding the acceptance range is recorded as n. The larger the number of n, the more segments of the detection center line with large spans, which can indicate that the fluctuation of the detection center line is large.

[0119] Of course, judging the dimension solely by the span size is too simplistic. For example, if the detection center line fluctuates along a diagonal line, then it is possible that although the span appears large in the coordinate system, the fluctuation of the detection center line around the diagonal line is actually very small. The large span in the coordinate system is the result of the fluctuation plus the inclination of the diagonal line.

[0120] Therefore, when calculating the fluctuation ζ, it is also necessary to consider the inclination of the detection center line, calculate the midpoint of the adjacent maximum and minimum points, and then calculate the slope of the straight line connecting the two adjacent midpoints. If this slope is close to tanβ (that is, the slope of the straight line fitted by the detection center line as a whole), the standard deviation of the difference between all slopes and tanβ is recorded as m. The smaller m is, the more slopes are close to tanβ, that is, the more likely the detection center line is to fluctuate near the straight line fitted by the detection center line as a whole, and then the fluctuation ζ can be reduced accordingly.

[0121] After combining the above two, we can calculate the volatility ζ = 0.19·n·m;

[0122] Among them, 0.19 is the adjustment coefficient, and the value can be set according to actual conditions.

[0123] Finally, the calculation formula of the flatness score S is as follows:

[0124] S=k1·|γ| / [γ]+k2·|β| / [β]+k3·|ζ| / [ζ];

[0125] Among them, k1, k2, and k3 are all set adjustment coefficients, and k1+k2+k3=1, and the numerical value is set according to the safety priority; [γ] is the maximum allowable value of the defectivity γ; [β] is the maximum allowable value of the inclination angle β; [ζ] is the maximum allowable value of the fluctuation ζ.

[0126] When making a judgment, S should be judged first. After determining that S exceeds the set range, γ, β and ζ should be judged separately to determine which one has a larger deviation, and then make corresponding adjustment suggestions.

[0127] The present invention also relates to a rearview mirror flatness detection system, such as Figure 2 As shown, including:

[0128] Detection frame 1, provided with a detection area;

[0129] The laser emitter 2 and the camera 3 are fixed on the detection frame and are arranged above the detection area;

[0130] A conveying device 4 is used to convey the rearview mirror; a clamp for fixing the rearview mirror is provided on the conveying device;

[0131] The processor is used to implement the above-mentioned rearview mirror flatness detection method.

[0132] In the inspection area, and on all parts that will enter the inspection area, except for the mirror surface of the rearview mirror, the top surfaces of the fixture, the inspection frame 1 table and other parts are all covered with a layer of black matte material. The black matte material can absorb the laser as much as possible and cause the laser that cannot be absorbed to be diffusely reflected, so that the laser will not be reflected into the camera lens, avoiding interference with the background environment on the inspection.

[0133] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for detecting the flatness of a rearview mirror, characterized in that the steps include: Set the detection area and project the line laser into the detection area; Test the standard rearview mirror and obtain its centerline through the centerline detection model as the reference centerline; The rearview mirror to be tested is tested, and its center line is obtained by the center line detection model and recorded as the detection center line; Calculate the defectivity γ; compare the reference center line and the detection center line to obtain the inclination angle β of the detection center line; calculate the fluctuation ζ of the detection center line; The flatness score S is calculated based on the tilt angle β, the fluctuation ζ and the defect γ. If the flatness score S exceeds the set range, the flatness of the rearview mirror to be tested is judged to be unqualified; The centerline detection model specifically includes: Turn off the line laser, acquire an image of the detection area and calculate the background light intensity parameters; Turn on the line laser and transport the rearview mirror horizontally through the detection area at a fixed speed, with the rearview mirror facing upward and the transport direction perpendicular to the width direction of the line laser; The reflection angle direction of the line laser is collected at fixed intervals to obtain an image of the detection area, which is preprocessed using background light intensity parameters, and then the center of mass position of the light intensity in the image is calculated; Eliminate all images whose light intensity centroid position cannot be calculated; With the ordinal number as the independent variable a and the position of the centroid of the light intensity obtained in time sequence as the dependent variable b, starting from a=0, the center line b(a) is obtained.

2. The rearview mirror flatness detection method according to claim 1, characterized in that: In the centerline detection model, the detection area image is preprocessed using the background light intensity parameter, specifically including: The direction of the rearview mirror is the x-axis and the width of the line laser is the y-axis; When the linear laser is turned off, the light intensity of the pixel at row x and column y in the detection area image is Ibg(x,y), and the mean μbg and standard deviation σbg of all pixels in the image are calculated; When the line laser is turned on, the light intensity of the pixel at the xth row and yth column in the detection area image is Iraw(x,y); Then the light intensity of the pixel at row x and column y in the image obtained after image preprocessing is: Inorm(x,y)=(Iraw(x,y)-μbg) / σbg.

3. The rearview mirror flatness detection method according to claim 1, characterized in that: In the centerline detection model, the center of mass position of the light intensity in the image is calculated as follows: The light intensity of the pixel at row x and column y in the image obtained after image preprocessing is recorded as Inorm(x,y); Calculate the segmentation threshold T = μnorm + 3σnorm; Where μnorm is the average light intensity of all pixels in the image, and σnorm is the standard deviation of the light intensity of all pixels in the image; Filter out the pixels in the image with light intensity greater than T and record them as laser pixels, and calculate the centroid position b: b=(Σy·Inorm(x,y)) / (ΣInorm(x,y)); Wherein, Σy·Inorm(x,y) represents the sum of the light intensities of all laser pixels multiplied by their corresponding column number y; ΣInorm(x,y) represents the sum of the light intensities of all laser pixels.

4. The rearview mirror flatness detection method according to claim 3, characterized in that: The calculation of defectivity γ is as follows: The images left in the centerline detection model are numbered 0, 1, 2, ...a according to their generation time; The laser pixel light intensity obtained from the ath image when testing the standard rearview mirror is recorded as Inorm_base(a)(x,y); The light intensity of the laser pixel obtained in the ath image when testing the rearview mirror to be tested is recorded as Inorm_test(a)(x,y); For all values of a, the light intensity difference of the laser pixel point is calculated at each value △Ia(x,y) = Inorm_base(a)(x,y) - Inorm_test(a)(x,y); Set the allowed range, record the number of laser pixels whose △Ia(x,y) exceeds the allowed range as p, and the total number of laser pixels as P; Defectivity γ = p / P.

5. The rearview mirror flatness detection method according to claim 3, characterized in that: It also includes a continuity detection step, specifically: After filtering out the laser pixel points of each image, the x-values of the laser pixel points are checked to see if they are continuous. If they are continuous, the subsequent calculations are continued; if they are not continuous, the gaps are filled in by interpolation to ensure that the x-values of the laser pixel points are continuous.

6. The rearview mirror flatness detection method according to claim 1, characterized in that: The inclination angle β of the detection center line is obtained as follows: The reference center line is recorded as b0(a), and the detection center line is recorded as b1(a); When calculating each value of a, the offset △b(a)=b1(a)-b0(a); Tilt angle β = arctan(Δb(a)avg / 2D); Where △b(a)avg is the average value of all △b(a); D is the distance between the laser emission point and the mirror.

7. The rearview mirror flatness detection method according to claim 5, characterized in that: The specific calculation of the fluctuation degree ζ of the detection center line is: The detection center line is denoted as b1(a); Find the maximum and minimum points of the detection center line, and calculate the difference in b1 values between adjacent maximum and minimum points; set the acceptance range, and record the number of differences that exceed the acceptance range as n; Calculate the midpoints of adjacent maximum and minimum points, then calculate the slope of the line connecting the two adjacent midpoints, and record the standard deviation of the difference between all slopes and tanβ as m; The calculated fluctuation degree ζ = 0.19·n·m.

8. The rearview mirror flatness detection method according to claim 1, characterized in that: The calculation formula of the flatness score S is as follows: S=k1·|γ| / [γ]+k2·|β| / [β]+k3·|ζ| / [ζ]; Among them, k1, k2, and k3 are all set adjustment coefficients; [γ] is the maximum allowable value of the defect degree γ; [β] is the maximum allowable value of the inclination angle β; and [ζ] is the maximum allowable value of the fluctuation degree ζ.

9. A rearview mirror flatness detection system, characterized in that: include: A detection rack is provided with a detection area; A laser emitter and a camera are fixedly arranged above the detection area; A conveying device for conveying a rearview mirror; A processor, configured to implement the rearview mirror flatness detection method according to any one of claims 1 to 8.

10. The rearview mirror flatness detection method according to claim 9, characterized in that: In the detection area, except for the mirror surface of the rearview mirror, the top surface of the rest of the parts is covered with a layer of black matte material.

Images