Method for detecting assembly precision of backlight module based on machine vision

Abstract

The method for detecting the assembly precision of the backlight module based on machine vision comprises the following steps: placing the backlight module assembled on the cover plate on a detection platform, placing the backlight module on the top of the cover plate, arranging a machine vision camera above the backlight module, arranging a machine vision coaxial light source below the machine vision camera, and arranging a backlight light source below the backlight module; selecting two or more photographing points at the left edge of the backlight module or the right edge of the backlight module, and selecting two or more photographing points at the front end of the backlight module or the rear end of the backlight module. The purpose is to provide a method for detecting the assembly precision of the backlight module based on machine vision, which can accurately and efficiently complete the detection of the screen assembly precision without contacting the screen, thereby avoiding damaging the screen and making the detection result more stable and reliable.

Description

TECHNICAL FIELD

[0001] The present application relates to a method for detecting the assembly precision of a backlight module based on machine vision. BACKGROUND

[0002] In today's world, with the massive popularity of electronic products, the use of screens has appeared in various electronic products, from small electronic watches and mobile phones to large tablet computers and car head units, which may be equipped with a display screen. In the production process of such electronic products, there will inevitably be a lamination process of the backlight module (screen). For example, in the lamination process of the backlight module (screen), the backlight module needs to be directly combined with the rear protective shell, or the backlight module needs to be laminated with a glass cover plate. In the current process, the quality inspection of this lamination process is manually measured by a vernier caliper, and the distance between the two combined module outer edges is measured to determine whether the assembly precision meets the standard. However, the manual quality inspection has the disadvantages of low efficiency and unstable measurement results, i.e., different results may be measured by different personnel under different conditions. In addition, the vernier caliper may damage the screen during the measurement process.

[0003] In view of the problems existing in the prior art, the present application provides a method for detecting the assembly precision of a backlight module using machine vision. The method takes a photo of the backlight module (screen), then measures the distance between the laminated backlight module (screen) and the cover plate edge based on the obtained image, and finally completes the detection of the assembly precision of the backlight module (screen) and the cover plate without touching the screen of the backlight module. SUMMARY

[0004] The present application aims to provide a method for detecting the assembly precision of a backlight module and a cover plate without touching the screen of the backlight module, which can accurately and efficiently complete the detection of the assembly precision of the backlight module and the cover plate, thereby avoiding damage to the screen of the backlight module and making the detection results more stable and reliable.

[0005] The method for detecting the assembly precision of a backlight module based on machine vision of the present application comprises the following steps:

[0006] A. Place the backlight module assembled on the cover plate on the detection platform, with the backlight module on top of the cover plate. A machine vision camera is arranged above the backlight module, a machine vision coaxial light source is arranged below the machine vision camera, and a backlight light source is arranged below the backlight module.

[0007] B. Select two or more photographing points at the left edge of the backlight module or the right edge of the backlight module, and select two or more photographing points at the front end of the backlight module or the rear end of the backlight module.

[0008] C. Move the machine vision camera and machine vision coaxial light source to above a shooting point, turn on the machine vision coaxial light source, turn off the backlight light source, illuminate the shooting point, and use the machine vision camera to take a first shot of the shooting point to obtain a front view, then turn off the machine vision coaxial light source, turn on the backlight light source, and use the machine vision camera to take a second shot of the shooting point to obtain a backlight view;

[0009] Then move to above another shooting point, turn on the machine vision coaxial light source, turn off the backlight light source, illuminate the shooting point, and use the machine vision camera to take a first shot of the shooting point to obtain a front view, then turn off the machine vision coaxial light source, turn on the backlight light source, and use the machine vision camera to take a second shot of the shooting point to obtain a backlight view, and so on until all shooting points are completed twice;

[0010] D. Import all the pictures obtained by shooting into the machine vision algorithm;

[0011] E. In the window of the machine vision algorithm, use the mouse to perform orthogonal rectangular interception processing on each picture, retain the image part without dirt and holes in the picture, and remove the part with dirt pattern and hole pattern in the picture;

[0012] Then use the Halcon operator edges sub pix to extract the sub-pixel contour of the outer edge of the cover plate in the backlight view of all pictures, then fit a straight line region, and then obtain the inclination angle of the straight line region, and then use the rotation affine transformation and shape template matching to obtain a corrected image;

[0013] F. Preprocess the front view picture of the image processed in step E to increase the contrast of the pattern in the picture and highlight the backlight module edge and the cover plate edge;

[0014] G. Perform one-dimensional measurement on the left side or right side of each image processed in step F to obtain the distance between the left side edge of the backlight module and the left side edge of the cover plate along the left-right horizontal direction of each shooting point, or the distance between the right side edge of the backlight module and the right side edge of the cover plate along the left-right horizontal direction of each shooting point;

[0015] Perform one-dimensional measurement on the front end or rear end of each image processed in step F to obtain the distance between the front end of the backlight module and the front end of the cover plate along the front-back horizontal direction of each shooting point, or the distance between the rear end of the backlight module and the rear end of the cover plate along the front-back horizontal direction of each shooting point;

[0016] H. comparing the measured distance between the left edge of the backlight module of each photographing point and the left edge of the cover plate along the left-right horizontal direction with the set distance in the assembly drawing, or comparing the measured distance between the right edge of the backlight module of each photographing point and the right edge of the cover plate along the left-right horizontal direction with the set distance in the assembly drawing, while comparing the measured distance between the front end of the backlight module of each photographing point and the front end of the cover plate along the front-back horizontal direction with the set distance in the assembly drawing, or comparing the measured distance between the rear end of the backlight module of each photographing point and the rear end of the cover plate along the front-back horizontal direction with the set distance in the assembly drawing, if all the measured distances are within the design accuracy requirement range, the product is determined to be a qualified product;

[0017] if one or more of the measured distances are not within the design accuracy requirement range, the product is determined to be an unqualified product.

[0018] Preferably, in step B, two or three or four photographing points are selected at the left edge of the backlight module or at the right edge of the backlight module, and two or three or four photographing points are selected at the front end of the backlight module or at the rear end of the backlight module.

[0019] Preferably, in step F, the method for pre-processing the front view picture of the image processed in step E includes the following forms:

[0020] A1. The pre-processing effect of the image is achieved by calculating the gray standard deviation;

[0021] A2. The black and white difference of the image is made more distinct by increasing the contrast Illuminate, thereby highlighting the edges of the screen;

[0022] A3. The enhanced operator Emphasize is used to enhance the contrast of the image region by region using a mask region;

[0023] A4. The mean filter operator Mean Image is used to calculate the average value of the gray scale of the image region by region using a mask region, so that the overall gray scale of each small region is equal to the average value of the gray scale of the small region, thereby smoothing and filtering out some unwanted dust and dirt.

[0024] A5. The SobalAmp filter operator is used to perform Sobel operation on the image region by region using a mask region, thereby highlighting the edges of the image.

[0025] Preferably, in step H, the unqualified products are four cases of left-right offset out-of-tolerance, left-right tilt out-of-tolerance, up-down offset out-of-tolerance, and up-down tilt out-of-tolerance.

[0026] Preferably, the one-dimensional measurement on the left or right side of each image processed in step F is a multiple measurement method, obtaining the distance between the left edge of the backlight module and the left edge of the cover plate in the left-right horizontal direction at each shooting point, or obtaining the distance between the right edge of the backlight module and the right edge of the cover plate in the left-right horizontal direction at each shooting point.

[0027] The one-dimensional measurement on the front or back end of each image processed in step F is a multiple measurement method, obtaining the distance between the front end of the backlight module and the front end of the cover plate in the front-back horizontal direction at each shooting point, or obtaining the distance between the back end of the backlight module and the back end of the cover plate in the front-back horizontal direction at each shooting point.

[0028] The multiple measurement method is to generate 16-24 distance-uniform parallel normal lines around the edge of the image, and each normal line is used as a reference to perform a measurement, and the influence of a certain part of the image with a dirty pattern or a hole pattern on the true result is excluded through multiple measurements.

[0029] The beneficial effects of the present application are as follows:

[0030] The method for detecting the assembly precision of the backlight module based on machine vision provided by the present application is to use a machine vision camera to take a picture of the backlight module, and then measure the distance between the edge of the backlight module and the edge of the cover plate based on the obtained image, so as to accurately and efficiently detect the assembly precision of the backlight module without contacting the screen. Compared with the existing method, the machine vision camera has the advantages of high efficiency, high accuracy, stable working performance, etc. Therefore, the method for detecting the assembly precision of the backlight module based on machine vision has the characteristics of accurately and efficiently detecting the assembly precision of the backlight module and the cover plate without contacting the screen of the backlight module, so as to avoid damaging the screen of the backlight module and make the detection result more stable and reliable.

[0031] The method for detecting the assembly precision of the backlight module based on machine vision of the present application will be further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 The working principle diagram of the method for detecting the assembly precision of the backlight module based on machine vision of the present application.

[0033] Figure 2 The schematic diagram of selecting a shooting point in the method for detecting the assembly precision of the backlight module based on machine vision of the present application. DETAILED DESCRIPTION

[0034] Reference Figure 1 andFigure 2 The method for detecting the assembly precision of the backlight module based on machine vision of the present application comprises the following steps:

[0035] A, place the backlight module 2 assembled on the cover plate 1 on the detection platform, and place the backlight module 2 on the top of the cover plate 1, a machine vision camera is arranged above the backlight module 2, a machine vision coaxial light source is arranged below the machine vision camera, the machine vision coaxial light source is arranged above the backlight module 2, and a backlight light source is arranged below the backlight module 2;

[0036] B, select two or more photographing points 4 at the left side edge 3 of the backlight module 2 or the right side edge 13 of the backlight module, and select two or more photographing points 4 at the front end 5 of the backlight module 2 or the rear end 15 of the backlight module 2;

[0037] C, move the machine vision camera and the machine vision coaxial light source to above one photographing point 4, turn on the machine vision coaxial light source and turn off the backlight light source, illuminate the photographing point 4, take a first photograph of the photographing point 4 by using the machine vision camera to obtain a front view, then turn off the machine vision coaxial light source and turn on the backlight light source, take a second photograph of the photographing point 4 by using the machine vision camera to obtain a backlight view;

[0038] Then move to above another photographing point 4, turn on the machine vision coaxial light source and turn off the backlight light source, illuminate the photographing point 4, take a first photograph of the photographing point 4 by using the machine vision camera to obtain a front view, then turn off the machine vision coaxial light source and turn on the backlight light source, take a second photograph of the photographing point 4 by using the machine vision camera to obtain a backlight view, and the above steps are repeated until the second photographing is completed for all the photographing points 4;

[0039] D, import all the pictures obtained by photographing into a machine vision algorithm;

[0040] E, use a mouse to perform a rectangular orthogonal intercepting process on each picture in a window of the machine vision algorithm, retain the image part without dirt and holes in the picture, and remove the part with the dirt pattern and the hole pattern in the picture;

[0041] Then, the edges sub pix of the Halcon operator is used to extract the sub-pixel contour of the edge 6 of the cover plate 1 in the backlight view of all the pictures, a straight line region is fitted, the inclination angle of the straight line region is obtained, and then a rotation affine transformation and shape template matching are used to obtain a corrected image;

[0042] F. Preprocess the front view of the image processed in step E to increase the contrast of the pattern in the image and highlight the left edge 3, right edge 13, front end 5, rear end 15 and cover edge 6 of the backlight module 2.

[0043] G. Perform a one-dimensional measurement on the left or right edge of each image processed in step F to obtain the distance between the left edge 3 of the backlight module and the left edge 6 of the cover plate in the horizontal direction at each shooting point, or to obtain the distance between the right edge 13 of the backlight module and the right edge 6 of the cover plate in the horizontal direction at each shooting point.

[0044] Perform a one-dimensional measurement on the front or back end of each image processed in step F to obtain the distance between the front end of the backlight module and the front end of the cover plate in the front-back horizontal direction at each shooting point, or to obtain the distance between the back end 15 of the backlight module and the back end 6 of the cover plate in the front-back horizontal direction at each shooting point.

[0045] H. Compare the measured distance between the left edge 3 of the backlight module and the left edge 6 of the cover plate at each shooting point along the horizontal direction with the set distance in the assembly drawing; or compare the measured distance between the right edge 13 of the backlight module 2 at each shooting point 4 and the right edge 6 of the cover plate along the horizontal direction with the set distance in the assembly drawing; at the same time, compare the measured distance between the front end 5 of the backlight module 2 at each shooting point 4 and the front end 6 of the cover plate along the horizontal direction with the set distance in the assembly drawing; or compare the measured distance between the rear end 15 of the backlight module 2 at each shooting point 4 and the rear end 6 of the cover plate along the horizontal direction with the set distance in the assembly drawing. If all the measured distances are within the design accuracy requirements, the product is determined to be a qualified product.

[0046] If more than one of the measurement distances is outside the design accuracy requirements, the product is deemed unqualified.

[0047] As a further improvement of the present invention, in step B above, two, three or four shooting points 4 are selected at the left edge 3 or the right edge 13 of the backlight module 2, and two, three or four shooting points 4 are selected at the front end 5 or the rear end 15 of the backlight module 2.

[0048] As a further improvement of the present invention, the method for preprocessing the front view of the image processed in step F above includes the following form:

[0049] A1. Image preprocessing is achieved by calculating the grayscale standard deviation;

[0050] A2, Illuminate, which increases the contrast of the image, makes the black and white difference more distinct, and thus highlights the edges of the screen;

[0051] A3, Emphasize, which uses a mask area to enhance the contrast of the image region by region;

[0052] A4, Mean Image, which uses a mask area to calculate the average value of the gray scale of the image region by region, so that the overall gray scale of each small region is equal to the average value of the gray scale of the small region, thereby filtering out some unwanted dust and dirt.

[0053] A5, SobalAmp, which uses a mask area to perform Sobel operation on the image region by region, thereby highlighting the edges of the image.

[0054] As a further improvement of the present application, the unqualified products in step H are four cases of left-right offset out-of-tolerance, left-right tilt out-of-tolerance, up-down offset out-of-tolerance, and up-down tilt out-of-tolerance.

[0055] As a further improvement of the present application, the one-dimensional measurement of the left or right side of each image processed in step F is performed by multiple measurement method, to obtain the distance between the left edge 3 of the backlight module 2 of each shooting point 4 and the left edge 6 of the cover plate along the left-right horizontal direction, or the distance between the right edge 13 of the backlight module of each shooting point 4 and the right edge 6 of the cover plate along the left-right horizontal direction.

[0056] The one-dimensional measurement of the front or rear end of each image processed in step F is performed by multiple measurement method, to obtain the distance between the front end 5 of the backlight module 2 of each shooting point 4 and the front end 6 of the cover plate along the front-rear horizontal direction, or the distance between the rear end 15 of the backlight module 2 of each shooting point 4 and the rear end 6 of the cover plate along the front-rear horizontal direction.

[0057] The multiple measurement method is to generate 16-24 distance-uniform parallel normal lines along the edge of the image, and perform measurement based on each normal line, so as to exclude the influence of the part with dirt pattern or hole pattern in the picture on the true result.

Claims

1. A method for inspecting the assembly accuracy of a backlight module based on machine vision, characterized in that: Includes the following steps: A. Place the backlight module assembled on the cover plate on the testing platform. When placing it, make sure the backlight module is on top of the cover plate. A machine vision camera is located above the backlight module, a machine vision coaxial light source is located below the machine vision camera, and a backlight light source is located below the backlight module. B. Select two or more shooting points at the left edge or right edge of the backlight module, and select two or more shooting points at the front or rear of the backlight module. C. Move the machine vision camera and machine vision coaxial light source above a shooting point, turn on the machine vision coaxial light source, turn off the backlight source, illuminate the shooting point, and use the machine vision camera to take the first picture of the shooting point to obtain a front view. Then turn off the machine vision coaxial light source, turn on the backlight source, and use the machine vision camera to take the second picture of the shooting point to obtain a backlit image. Then move to the top of another shooting point, turn on the machine vision coaxial light source, turn off the backlight source, illuminate the shooting point, and take the first picture of the shooting point using the machine vision camera to obtain a front view. Then turn off the machine vision coaxial light source, turn on the backlight source, and take the second picture of the shooting point using the machine vision camera to obtain a backlit image. Repeat this process until all shooting points have been photographed twice. D. Import all the images obtained from taking photos into the machine vision algorithm; E. In the machine vision algorithm window, use the mouse to crop each image using orthogonal rectangles, retaining the image parts without dirt or holes, and removing the parts with dirt patterns or hole patterns. Then, the Halcon operator edges sub pix is ​​used to extract the subpixel contour of the outer edge of the cover plate in the backlight image of all images. Then, a straight line region is fitted, and the tilt angle of the straight line region is obtained. Then, the corrected image is obtained by using rotational affine transformation and shape template matching. F. Preprocess the front view of the image processed in step E to increase the contrast of the pattern in the image and highlight the edges of the backlight module and the cover plate. G. Perform a one-dimensional measurement on the left or right edge of each image processed in step F to obtain the distance between the left edge of the backlight module and the left edge of the cover plate in the horizontal direction at each shooting point, or obtain the distance between the right edge of the backlight module and the right edge of the cover plate in the horizontal direction at each shooting point. Perform a one-dimensional measurement on the front or back end of each image processed in step F to obtain the distance between the front end of the backlight module and the front end of the cover plate in the front-back horizontal direction at each shooting point, or the distance between the back end of the backlight module and the back end of the cover plate in the front-back horizontal direction at each shooting point. H. Compare the measured distance between the left edge of the backlight module and the left edge of the cover plate at each shooting point in the horizontal direction with the set distance in the assembly drawing; or compare the measured distance between the right edge of the backlight module and the right edge of the cover plate at each shooting point in the horizontal direction with the set distance in the assembly drawing; at the same time, compare the measured distance between the front end of the backlight module and the front end of the cover plate at each shooting point in the horizontal direction with the set distance in the assembly drawing; or compare the measured distance between the rear end of the backlight module and the rear end of the cover plate at each shooting point in the horizontal direction with the set distance in the assembly drawing. If all the measured distances are within the design accuracy requirements, the product is judged to be qualified. If more than one of the measured distances is outside the design accuracy requirements, the product is deemed unqualified.

2. The method for detecting the assembly accuracy of a backlight module based on machine vision according to claim 1, characterized in that: In step B, two, three, or four shooting points are selected at the left or right edge of the backlight module, and two, three, or four shooting points are selected at the front or rear end of the backlight module.

3. The method for detecting the assembly accuracy of a backlight module based on machine vision according to claim 2, characterized in that: The method for preprocessing the frontal image of the image processed in step E in step F includes the following forms: A1. Image preprocessing is achieved by calculating the grayscale standard deviation; A2. By increasing contrast, Illuminate makes the black and white differences in the image more distinct, thereby highlighting the edges of the screen; A3. Enhance the contrast of an image region by region using the Emphasize operator; A4. Mean Image, a mean filtering operator, uses a mask area to calculate the average gray level of each region of the image, so that the overall gray level of each small region is equal to the average gray level of that small region, thereby smoothly filtering out some unwanted dust and dirt. A5. The SobalAmp filter operator is used to perform Sobal operations on each region of the image using a mask region, thereby highlighting the image edges.

4. The method for detecting the assembly accuracy of a backlight module based on machine vision according to claim 3, characterized in that: The non-conforming products in step H fall into four categories: left-right offset exceeding tolerance, left-right tilt exceeding tolerance, up-down offset exceeding tolerance, and up-down tilt exceeding tolerance.

5. The method for detecting the assembly accuracy of a backlight module based on machine vision according to any one of claims 1 to 4, characterized in that: In step G, the one-dimensional measurement of the left or right edge of each image processed in step F is performed by multiple measurements to obtain the distance between the left edge of the backlight module and the left edge of the cover plate in the horizontal direction at each shooting point, or the distance between the right edge of the backlight module and the right edge of the cover plate in the horizontal direction at each shooting point. The one-dimensional measurement of the front or back end of each image processed in step F is performed by multiple measurements to obtain the distance between the front end of the backlight module and the front end of the cover plate in the front-back horizontal direction at each shooting point, or the distance between the back end of the backlight module and the back end of the cover plate in the front-back horizontal direction at each shooting point. The multiple measurement method involves generating 16-24 evenly spaced parallel normals around the edge of the image, and performing a measurement based on each normal. By performing multiple measurements, the influence of dirty patterns or hole patterns in a certain part of the image on the true result can be eliminated.

Images