Device and method for detecting surface defects of flexible OLED panel based on photometric stereo method
Through photometric stereoscopic method and multi-angle dome light source combined with YOLOv8-seg detection model and DenseFuse fusion network, the accuracy and speed problems of complex defect detection of flexible OLED panels are solved, and efficient defect detection is achieved.
Patent Information
- Application Number
- CN202410976393.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-20
- Publication Date
- 2024-11-08
AI Technical Summary
Flexible OLED panels have many complex defects in the manufacturing process. Traditional detection methods rely on artificial vision, with low accuracy and stability, and the fixed light source angle cannot be adjusted, making it difficult to meet the detection needs.
Using a detection method based on photometric stereoscopic method, multi-angle images are obtained through the camera and programmable multi-angle dome light source, combined with the YOLOv8-seg detection model and DenseFuse fusion network, a suitable lighting scheme is calculated to achieve comprehensive and comprehensive surface of flexible OLED panels. Accurate detection.
It improves the accuracy and speed of detection, reduces labor costs, and can identify small defects and complex defect types, enhancing the comprehensiveness and accuracy of detection.
Smart Images

Figure CN118913632A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a method for detecting surface defects of a flexible OLED panel, and in particular to a device and method for detecting surface defects of a flexible OLED panel based on a photometric stereo method. Background Art
[0002] As a new generation of display technology, flexible screens have broad application prospects in the field of electronic products and are widely used in smart phones, wearable devices, foldable devices and other fields. In the usual sense, flexible screens refer to flexible OLED (Flexible OLED, FOLED), which are display devices prepared on flexible substrates such as plastic, metal sheets, and glass sheets. Compared with traditional screens, flexible screens have obvious advantages. Not only are they thinner and lighter in size, but they also consume less power than the original devices, which helps to improve the battery life of the device. At the same time, based on their bendability and good flexibility, their durability is much higher than previous screens, which can effectively reduce the probability of accidental damage to the device.
[0003] Although flexible OLED has many advantages, it has many difficulties in actual mass production. The biggest problem is that the yield is not high. This is due to the particularity of flexible OLED panels. The defects generated during its manufacturing process are complex and diverse, such as dots, lines, scratches, cracks, foreign matter, different colors, dust, bumps, dirt, bubbles, broken pieces, etc. These defects not only affect the appearance quality of the product, but may also cause the function of the equipment to be unstable or even unable to work properly. There are the following problems in the defect detection of flexible OLED surfaces. First, flexible OLED panels have the characteristics of flexible product warping, complex and diverse defect types, scratches and cracks that require specific angles of light to highlight, some defects are similar in two-dimensional morphology, the defect size is small, and the accuracy of defect detection is high. This makes defect detection for flexible OLED panels a major difficulty; second, traditional defect detection methods are mainly based on artificial vision, which requires a lot of manpower and time costs, and are easily affected by human factors, resulting in low accuracy and stability of detection; third, the current defect detection device based on photometric stereo method has mostly fixed light source angle information and cannot be adjusted according to the actual detection situation. Summary of the invention
[0004] The present invention aims at the problems existing in the prior art and provides a flexible OLED panel surface defect detection device and detection method based on photometric stereo. Through preliminary detection, the defect direction and position information are obtained to calculate a suitable lighting scheme. Then, according to the lighting scheme, a programmable multi-angle dome light source is used in conjunction with a camera to obtain an image, and the photometric stereo method is used for subsequent processing. Finally, the detection model is used to realize the detection of surface defects of the flexible OLED panel.
[0005] Specifically, the method obtains a color first sample image of the surface of a flexible OLED panel in a fully illuminated state; optimizes the first sample image by preprocessing means to obtain a second sample image; performs preliminary detection on the second sample image by a YOLOv8-seg detection model to obtain defect information, thereby calculating a suitable lighting scheme; obtains three color third sample images according to the lighting scheme; forms images of 9 angles by channel separation of the three color third sample images, obtains a gradient map by using a photometric stereo method, and obtains a fourth sample image by curvature calculation; fuses the second sample image and the fourth sample image by a DenseFuse fusion network to obtain a fifth sample image; and performs defect detection on the fifth sample image by a YOLOv8-seg detection model, thereby achieving the purpose of performing comprehensive and accurate detection of the surface of the flexible OLED panel at a faster detection speed.
[0006] The present invention is achieved through the following technical solutions:
[0007] The present invention discloses a flexible OLED panel surface defect detection device based on photometric stereo method, comprising:
[0008] Camera, telecentric lens, programmable multi-angle dome light source, controller, host computer, stage;
[0009] The lower end of the camera is connected to a telecentric lens, and a programmable multi-angle dome light source and a stage are arranged in sequence below the telecentric lens. The host computer is connected to a controller, and the controller is respectively connected to the camera and the programmable multi-angle dome light source;
[0010] The camera and telecentric lens are used to capture images of the flexible OLED panel surface;
[0011] The programmable multi-angle dome light source is used for lighting, and is constructed from the outside to the inside as a shell, a curved first thin plate, and a transparent curved second thin plate. The surface of the curved first thin plate facing the curved second thin plate is provided with an LED light-emitting point, and the curved second thin plate is provided with a convex lens corresponding to the LED light-emitting point;
[0012] The controller is used to receive the control command of the host computer to control the adjustment of the camera and the programmable multi-angle dome light source, and perform lighting according to the lighting plan to cooperate with the camera to collect color images;
[0013] The host computer is used to send commands to the controller to control the camera and the programmable multi-angle dome light source to synchronously capture images on the surface of the flexible OLED panel, and can set the camera's exposure time, white balance, gain, and the luminous point, luminous point brightness, luminous point color, and luminous order of the programmable multi-angle dome light source;
[0014] The stage is used to place flexible OLED panel samples.
[0015] As a further improvement, the programmable multi-angle dome light source of the present invention has n layers of n LED light points distributed on the curved first thin plate from 25° to 70°, and the angle between each layer of light points and the horizontal plane is Each LED light point is evenly distributed.
[0016] As a further improvement, each LED light-emitting point of the programmable multi-angle dome light source described in the present invention includes lamp beads encapsulating three colors of red, green and blue to control the light color and brightness, and the brightness is adjusted individually in levels 0 to 255.
[0017] As a further improvement, corresponding convex lenses are evenly distributed on the two thin plates of the present invention according to the positions of the corresponding LED light-emitting points on the curved first thin plate, so as to control the direction of the light source and make the light convergent and uniform.
[0018] The present invention also discloses a method for detecting surface defects of a flexible OLED panel based on a photometric stereo method, comprising:
[0019] Sending an acquisition instruction to the controller, the controller controls the full light source to be turned on, and controls the camera to collect the first color sample image and return it to the host computer, thereby obtaining a first color sample image of the surface of the flexible OLED panel in the full light source turned on state;
[0020] Optimizing the first sample image by means of gamma correction and filtering preprocessing to obtain a second sample image;
[0021] Through the YOLOv8-seg detection model, the second sample image is preliminarily detected to obtain the defect direction and location information, so as to calculate the appropriate lighting solution;
[0022] According to the lighting scheme, the controller controls the programmable multi-angle dome light source to sequentially light up the selected LED light points of n layers of the dome light source, and each layer lights up the selected red, green and blue light points at the same time, and cooperates with the camera to obtain a color image, and a total of n color third sample images are collected;
[0023] Separate the channels of the n third color sample images to form 3*n images with 3*n angles, use the photometric stereo method, combine the angle information of the light-emitting point, obtain the gradient map according to the 3*n images with 3*n angles, and calculate the curvature to obtain the fourth sample image;
[0024] The second sample image and the fourth sample image are fused by a DenseFuse fusion network to obtain a fifth sample image;
[0025] The fifth sample image is subjected to defect detection through the YOLOv8-seg detection model, and the detection result is displayed, that is, whether there is a defect. If there is a defect, the first sample image and the fifth sample image including the defect range and defect type are displayed.
[0026] As a further improvement, the lighting scheme described in the present invention includes the required light-emitting points, light-emitting point brightness, light-emitting point color, and light-emitting order. The lighting scheme uses the YOLOv8-seg detection model to perform preliminary detection on the second sample image to obtain defect direction and position information, so as to calculate the appropriate lighting scheme. Specifically, the defect is detected and instance segmented by the YOLOv8-seg model of YOLOv8, and the position information of the defect is output. According to each pixel point of the defect, the leastsq function for nonlinear least squares fitting in the optimization module scipy.optimize module in Python is used to perform least squares fitting with the defect direction, which is described by the following formula 1):
[0027]
[0028] The lighting plan is set by combining the defect direction with the defect location and defect direction.
[0029] As a further improvement, the present invention obtains a gradient map through 3*n images at 3*n angles, specifically:
[0030] In the Lambert ideal diffuse reflection model, the intensity of diffuse reflected light can be described by the following formula: I = ρLN 2)
[0031] Where I is the brightness of the image sensor, ρ is the surface reflectivity of the object to be measured; L = (Lx, Ly, Lz) is the unit direction vector of the light source; N = (Nx, Ny, Nz); N is the unit normal vector of a point on the surface of the object to be measured;
[0032] According to formula 2), the brightness of a pixel in 3*n images at different angles obtained under different light sources, I1, I2, I3, I4, I5, I6, I7, I8, I9 are:
[0033] I1=ρL1N 3)
[0034] I2=ρL2N 4)
[0035] I3=ρL3N 5)
[0036] I4=ρL4N 6)
[0037] I5=ρL5N 7)
[0038] I6=ρL6N 8)
[0039] I7=ρL7N 9)
[0040] I8=ρL8N 10)
[0041] I9=ρL9N 11)
[0042] Since N and L are unit vectors, we have:
[0043] ρ=||L -1 I|| 12)
[0044]
[0045] Thus, the surface gradient of the object to be measured is obtained, that is, the height gradients p and q in the x and y directions.
[0046]
[0047] L=[L1,L2,L3,......L9] T 16)
[0048] I i =[I i1 , I i2 , I i3 ,......I i9 ]=ρ i LN i 17)
[0049] So
[0050] ρ i =|(LTL)-1LTI i |18)
[0051] N i =1ρ i (LTL)-1LI i 19);
[0052] And through the obtained gradient map, use the derivate_vector_field image processing operator and Gaussian curvature calculation to obtain the gradient curvature field, and highlight the defect information through the curvature map.
[0053] The beneficial effects of the present invention are as follows:
[0054] The present invention uses the YOLOv8-seg detection model to perform preliminary detection on the second sample image, obtain defect direction and position information, calculate the appropriate lighting scheme, and use a programmable multi-angle dome light source for lighting. This can target the flexible product warping and complex and diverse defect types of flexible OLED panels, so that defects such as scratches and cracks that require specific angles to highlight defect information and tiny defects are more obvious in imaging, thereby increasing the comprehensiveness and accuracy of subsequent defect detection.
[0055] The present invention uses photometric stereo method, combined with the angle information of the light-emitting point, obtains the gradient map according to the images at 9 angles, and obtains the fourth sample image by curvature calculation, which can effectively solve the problem that some defects are difficult to classify and identify because of their similar performance in two-dimensional morphology.
[0056] The present invention uses a DenseFuse fusion network to fuse the second sample image and the fourth sample image to obtain a fifth sample image, so that the fifth sample image retains two-dimensional color information while obtaining three-dimensional information, so that the image contains more information and the defect information is more obvious, thereby increasing the accuracy of subsequent defect detection and solving the problem of high requirements for defect detection accuracy.
[0057] The final detection method of the present invention performs defect detection on the fifth sample image through the YOLOv8-seg detection model, utilizes the deep learning detection network, reduces the labor cost of manual visual inspection, improves the recognition accuracy, and has a strong generalization ability. BRIEF DESCRIPTION OF THE DRAWINGS
[0058] Figure 1 is a data flow diagram of the present invention;
[0059] Figure 2 is a diagram of a defect detection device;
[0060] Figure 3 It is a schematic diagram of the structure of a programmable multi-angle dome light source;
[0061] Figure 4 is a schematic diagram of the distribution of LED light-emitting points of the curved first thin plate;
[0062] In the figure: 1 is a camera, 2 is a telecentric lens, 3 is a programmable multi-angle dome light source, 4 is a controller, 5 is a host computer, 6 is a stage, 7 is a housing, 8 is a first curved thin plate, 9 is an LED light-emitting point, 10 is a second curved thin plate, and 11 is a convex lens. DETAILED DESCRIPTION
[0063] The technical solution of the present invention is further described below by combining the accompanying drawings of the specification through specific embodiments:
[0064] The purpose of the present invention is to address the characteristics of flexible OLED panels, such as warping of flexible products, complex and diverse defect types, scratches and cracks that require specific angles of illumination to be prominent, similar two-dimensional appearance of some defects, small defect sizes, and high requirements for defect detection accuracy. A flexible OLED panel surface defect detection device and detection method based on photometric stereo method are proposed. The device for implementing the method of the present invention includes a camera 1, a telecentric lens 2, a programmable multi-angle dome light source 3, a controller 4, a host computer 5, a stage 6, Figure 1 is a data flow diagram of the present invention, Figure 2 is a diagram of a defect detection device of the present invention, Figure 3 It is a schematic diagram of the structure of a programmable multi-angle dome light source;
[0065] The specific implementation method of the present invention is as follows:
[0066] The present invention discloses a flexible OLED panel surface defect detection device based on photometric stereo method, comprising: a camera 1, a telecentric lens 2, a programmable multi-angle dome light source 3, a controller 4, a host computer 5, and a stage 6; the lower end of the camera 1 is connected to the telecentric lens 2, and the programmable multi-angle dome light source 3 and the stage 6 are arranged in sequence below the telecentric lens 2, the host computer 5 is connected to the controller 4, and the controller 4 is respectively connected to the camera 1 and the programmable multi-angle dome light source 3; the camera 1 and the telecentric lens 2 are used to collect an image of the surface of the flexible OLED panel located above the stage 6.
[0067] The programmable multi-angle dome light source 3 is used for lighting, and is constructed from the outside to the inside as a housing 7, a curved first thin plate 8, and a transparent curved second thin plate 10. The surface of the curved first thin plate 8 facing the curved second thin plate 10 is provided with an LED light-emitting point 9, and the curved second thin plate 10 is provided with a convex lens 11 corresponding to the LED light-emitting point 9;
[0068] The controller 4 is used to receive control commands from the host computer 5 to control the adjustment of the camera 1 and the programmable multi-angle dome light source 3, and perform lighting according to the lighting scheme to cooperate with the camera 1 to collect color images;
[0069] The host computer 5 is used to send commands to the controller 4 to control the camera 1 and the programmable multi-angle dome light source 3 to synchronously capture the image of the object to be measured, and can set the exposure time, white balance, gain of the camera 1 and the light-emitting point, light-emitting point brightness, light-emitting point color, and light-emitting sequence of the programmable multi-angle dome light source 3; the stage 6 is used to place the flexible OLED panel sample.
[0070] Figure 4This is a schematic diagram of the distribution of LED light points on the curved first thin plate. The programmable multi-angle dome light source is distributed on the curved first thin plate in 10 layers from 25° to 70°, with 24 LED light points on each layer, a total of 240 LED light points. The angle between the light points on each layer and the horizontal plane is 5°, and each LED light point on the same layer is evenly distributed, with each light point 15° apart. Each LED light point of the programmable multi-angle dome light source contains lamp beads encapsulated in red, green, and blue colors to control the light color and brightness, and the brightness is adjusted individually in 0 to 255 levels.
[0071] The corresponding convex lenses are evenly distributed on the second curved thin plate according to the positions of the corresponding LED light-emitting points on each layer on the first curved thin plate, so as to control the direction of the light source and make the light converge evenly in the central area of the stage.
[0072] The present invention also discloses a method for detecting surface defects of a flexible OLED panel based on a photometric stereo method, comprising:
[0073] Sending an acquisition instruction to the controller, the controller controls the full light source of the programmable multi-angle dome light source to be illuminated, and controls the camera to acquire the first color sample image and return it to the host computer, thereby obtaining a first color sample image of the surface of the flexible OLED panel in the full light source illuminated state;
[0074] Optimizing the first sample image by means of gamma correction and filtering preprocessing to obtain a second sample image;
[0075] Through the YOLOv8-seg detection model, the second sample image is preliminarily detected to obtain the defect direction and position information, so as to calculate the appropriate lighting scheme; the lighting scheme includes the required light-emitting point, light-emitting point brightness, light-emitting point color, and light-emitting order. The lighting scheme is preliminarily detected through the YOLOv8-seg detection model on the second sample image to obtain the defect direction and position information, so as to calculate the appropriate lighting scheme as follows:
[0076] The defects are detected and instance segmented through the YOLOv8-seg model of YOLOv8, and the location information of the defects is output. Based on the pixels of the defects, the leastsq function for nonlinear least squares fitting in the optimization module scipy.optimize in Python is used to perform least squares fitting based on the defect direction, which is described by the following formula 1):
[0077]
[0078] The lighting scheme is set by combining the defect direction with the defect location and defect direction;
[0079] According to the lighting scheme, the controller controls the programmable multi-angle dome light source to light up the selected LED light points of the three layers of the dome light source in turn, and each layer lights up the selected red, green and blue light points at the same time, and cooperates with the camera to obtain a color image, and a total of three color third sample images are collected;
[0080] The three color third sample images are separated according to the R, G, and B channels to form 3*3 images with 9 angles. The photometric stereo method is used to obtain the gradient map based on the images at 9 angles in combination with the angle information of the light-emitting point, and the fourth sample image is obtained by curvature calculation. The gradient map is obtained based on the images at 9 angles, specifically:
[0081] In the Lambert ideal diffuse reflection model, the intensity of diffuse reflected light can be described by the following formula:
[0082] I=ρLN 2)
[0083] Where I is the brightness of the image sensor, ρ is the surface reflectivity of the object to be measured; L = (Lx, Ly, Lz) is the unit direction vector of the light source; N = (Nx, Ny, Nz); N is the unit normal vector of a point on the surface of the object to be measured;
[0084] According to formula 2), the brightness of a pixel in 3*n images at different angles obtained under different light sources, I1, I2, I3, I4, I5, I6, I7, I8, I9 are:
[0085] I1=ρL1N 3)
[0086] I2=ρL2N 4)
[0087] I3=ρL3N 5)
[0088] I4=ρL4N 6)
[0089] I5=ρL5N 7)
[0090] I6=ρL6N 8)
[0091] I7=ρL7N 9)
[0092] I8=ρL8N 10)
[0093] I9=ρL9N 11)
[0094] Since N and L are unit vectors, we have:
[0095] ρ=||L -1 I|| 12)
[0096]
[0097] Thus, the surface gradient of the object to be measured is obtained, that is, the height gradients p and q in the x and y directions.
[0098]
[0099] L=[L1,L2,L3,......L9] T 16)
[0100] I i =[I i1 , I i2 , I i3 ,......I i9 ]=ρ i LN i 17)
[0101] So
[0102] ρ i =|(LTL)-1LTI i |18)
[0103] N i =1ρ i (LTL)-1LI i 19);
[0104] And through the obtained gradient map, use the derivate_vector_field image processing operator, through Gaussian curvature calculation, obtain the curvature field of the gradient, and highlight the defect information through the curvature map;
[0105] The second sample image and the fourth sample image are fused by a DenseFuse fusion network to obtain a fifth sample image;
[0106] The fifth sample image is subjected to defect detection through the YOLOv8-seg detection model, and the detection result is displayed, that is, whether there is a defect. If there is a defect, the first sample image and the fifth sample image including the defect range and defect type are displayed.
[0107] The above description is not a limitation of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several changes, modifications, additions or substitutions can be made without departing from the essential scope of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A flexible OLED panel surface defect detection device based on photometric stereo method, characterized in that: include: Camera, telecentric lens, programmable multi-angle dome light source, controller, host computer, stage; The lower end of the camera is connected to a telecentric lens, a programmable multi-angle dome light source and a stage are sequentially arranged below the telecentric lens, the host computer is connected to a controller, and the controller is respectively connected to the camera and the programmable multi-angle dome light source; The camera and telecentric lens are used to capture images of the surface of the flexible OLED panel; The programmable multi-angle dome light source is used for lighting, and is constructed from the outside to the inside as a shell, a curved first thin plate, and a transparent curved second thin plate. The surface of the curved first thin plate facing the curved second thin plate is provided with an LED light-emitting point, and the curved second thin plate is provided with a convex lens corresponding to the LED light-emitting point; The controller is used to receive the control command of the host computer to control the adjustment of the camera and the programmable multi-angle dome light source, and perform lighting according to the lighting plan to cooperate with the camera to collect color images; The host computer is used to send commands to the controller to control the camera and the programmable multi-angle dome light source to synchronously collect images on the surface of the flexible OLED panel, and can set the camera's exposure time, white balance, gain, and the luminous point, luminous point brightness, luminous point color, and luminous order of the programmable multi-angle dome light source; The stage is used to place flexible OLED panel samples.
2. The flexible OLED panel surface defect detection device based on photometric stereo method according to claim 1, characterized in that: The programmable multi-angle dome light source has n layers of n LED light points distributed on the curved first thin plate from 25° to 70°, and the angle between each layer of light points and the horizontal plane is Each LED light point is evenly distributed.
3. The flexible OLED panel surface defect detection device based on photometric stereo method according to claim 1 or 2, characterized in that: Each LED light-emitting point of the programmable multi-angle dome light source includes lamp beads encapsulating three colors of red, green and blue to control the light color and brightness, and the brightness is adjusted individually in levels 0 to 255.
4. The flexible OLED panel surface defect detection device based on photometric stereo method according to claim 3 is characterized in that: The corresponding convex lenses are evenly distributed on the second curved thin plate according to the positions of the corresponding LED light-emitting points on the first curved thin plate, so as to control the direction of the light source and make the light converge and uniform.
5. A method for detecting surface defects of a flexible OLED panel based on photometric stereo method, characterized in that: include: Sending an acquisition instruction to the controller, the controller controls the full light source to be turned on, and controls the camera to collect the first color sample image and return it to the host computer, thereby obtaining a first color sample image of the surface of the flexible OLED panel in the full light source turned on state; Optimizing the first sample image by means of gamma correction and filtering preprocessing to obtain a second sample image; Through the YOLOv8-seg detection model, the second sample image is preliminarily detected to obtain the defect direction and location information, so as to calculate the appropriate lighting solution; According to the lighting scheme, the controller controls the programmable multi-angle dome light source to sequentially light up the selected LED light points of n layers of the dome light source, and each layer lights up the selected red, green and blue light points at the same time, and cooperates with the camera to obtain a color image, and a total of n color third sample images are collected; Separate the channels of the n third color sample images to form 3*n images with 3*n angles, use the photometric stereo method, combine the angle information of the light-emitting point, obtain the gradient map according to the 3*n images with 3*n angles, and calculate the curvature to obtain the fourth sample image; The second sample image and the fourth sample image are fused by a DenseFuse fusion network to obtain a fifth sample image; The fifth sample image is subjected to defect detection through the YOLOv8-seg detection model, and the detection result is displayed, that is, whether there is a defect. If there is a defect, the first sample image and the fifth sample image including the defect range and defect type are displayed.
6. The method for detecting surface defects of flexible OLED panels based on photometric stereo method according to claim 5, characterized in that: The lighting scheme includes the required light-emitting points, light-emitting point brightness, light-emitting point color, and light-emitting order. The lighting scheme uses the YOLOv8-seg detection model to perform preliminary detection on the second sample image to obtain defect direction and position information, so as to calculate the appropriate lighting scheme as follows: The defects are detected and instance segmented through the YOLOv8-seg model of YOLOv8, and the location information of the defects is output. Based on the pixels of the defects, the leastsq function for nonlinear least squares fitting in the optimization module scipy.optimize in Python is used to perform least squares fitting based on the defect direction, which is described by the following formula 1): The lighting plan is set by combining the defect direction with the defect location and defect direction.
7. The method for detecting surface defects of flexible OLED panels based on photometric stereo method according to claim 5, characterized in that: The gradient map is obtained by using 3*n images with 3*n angles, specifically: In the Lambert ideal diffuse reflection model, the intensity of diffuse reflected light can be described by the following formula: I=ρLN 2) Where I is the brightness of the image sensor, ρ is the surface reflectivity of the object to be measured; L = (Lx, Ly, Lz) is the unit direction vector of the light source; N = (Nx, Ny, Nz); N is the unit normal vector of a point on the surface of the object to be measured; According to formula 2), the brightness of a pixel in 3*n images at different angles obtained under different light sources, I1, I2, I3, I4, I5, I6, I7, I8, I9 are: I1=ρL1N 3) I2=ρL2N 4) I3=ρL3N 5) I4=ρL4N 6) I5=ρL5N 7) I6=ρL6N 8) I7=ρL7N 9) I8=ρL8N 10) I9=ρL9N 11) Since N and L are unit vectors, we have: p=||L -1 I|| 12) Thus, the surface gradient of the object to be measured is obtained, that is, the height gradients p and q in the x and y directions. <h2 style=";text-align:left;direction:ltr">L=[L1, L2, L3,......L9]<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> 16) I i =[I i1 ,I i2 ,I i3 ,……I i9 ]=ρ i LN i 17) So And through the obtained gradient map, use the derivate_vector_field image processing operator and Gaussian curvature calculation to obtain the gradient curvature field, and highlight the defect information through the curvature map.
Citation Information
Cited By
A method for detecting OLED screen defects based on optical-magnetic multimodal fusion
CN119780107A
A method for detecting OLED screen defects based on optical-magnetic multimodal fusion
CN119780107B
Arched detection light source detection method
CN120703097A
Defect detection method and system for backlight assembly
CN121010550A
