Multi-view-angle-based LED display screen brightness and chrominance correction method and storage medium

Abstract

The invention discloses a multi-view-angle-based LED display screen brightness and chrominance correction method and a storage medium. The method comprises the steps that corresponding brightness and chrominance correction coefficients are obtained through measurement under preset view angles; according to a preset period, shooting through a camera to obtain a crowd image in front of the display screen to be corrected; performing face recognition on the crowd image, and determining a crowd geometric center according to a face recognition result; determining a visual angle corresponding to the geometric center of the crowd according to the parameters of the camera; and determining a corresponding brightness and chrominance correction coefficient according to the visual angle corresponding to the geometric center of the crowd, and performing brightness and chrominance correction on a display screen to be corrected according to the corresponding brightness and chrominance correction coefficient. According to the invention, the optimal display effect of the LED display screen under different visual angles can be ensured.

Description

Technical Field

[0001] The present invention relates to the field of optoelectronic display technology, and in particular to a method and storage medium for calibrating brightness and chromaticity of an LED display screen based on multiple viewing angles. Background Art

[0002] LED displays surpass other display technologies in terms of brightness, contrast, color gamut, response time, scalability and other performance. However, due to the massive number of pixels (especially Mini / Micro LED displays, which not only have a large number of LED chips, but also have tiny chip sizes), it is impossible to screen the LED chips one by one, resulting in inconsistent brightness and chromaticity of each chip. The display modules and display screens manufactured with such chips have poor consistency in brightness and chromaticity (hereinafter referred to as brightness and chromaticity), and brightness and chromaticity correction must be performed before use.

[0003] In the process of displaying images on the LED display, the control system and the driver chip are used to realize point-by-point control of the display. The driver chip controls the luminous intensity of the three primary colors of red, green and blue in the pixels of the LED display to achieve different proportions of brightness mixing and display different colors. The different display of all pixels on the display constitutes a fresh and vivid image. The pixel-level acquisition and correction of the LED display is to collect the brightness and chromaticity of the three primary colors of red, green and blue in each display pixel, obtain the brightness and chromaticity information of the three primary colors of red, green and blue, obtain the relative proportion of the brightness and chromaticity of the three primary colors of each pixel by calculation, construct the correction coefficient, and input the correction coefficient into the control system to realize the brightness and chromaticity correction of the three primary colors of each pixel.

[0004] At present, the brightness and color correction of LED screens are all performed in the normal direction of the screen (i.e., 0° viewing angle), that is, the screen control system only stores the correction coefficients measured in the normal direction, and there are no correction coefficients for other viewing angles, nor is there a function that can sense the viewing angle of the viewer in front of the screen. After correction, the screen can achieve a high brightness and color uniformity in the normal direction, but when the viewing angle deviates from the normal direction, due to the different light distribution of the RGB three-color chips and the crosstalk between adjacent pixels, the LED screen will still have serious brightness and color non-uniformity, which greatly affects the display quality of the screen, such as Figure 1 As shown. Moreover, as the viewing angle increases, the brightness and color non-uniformity will become more serious, greatly reducing the display quality of the display and seriously affecting the audience's viewing experience. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a brightness and chromaticity correction method and a storage medium for an LED display screen based on multiple viewing angles, which can ensure the best display effect of the LED display screen at different viewing angles.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: a method for luminance and chromaticity correction of an LED display screen based on multiple viewing angles, comprising: Measure and obtain corresponding brightness and chromaticity correction coefficients at each preset viewing angle; According to a preset cycle, an image of the crowd in front of the display screen to be calibrated is obtained by taking pictures with a camera; Performing face recognition on the crowd image, and determining the geometric center of the crowd according to the face recognition result; Determine the viewing angle corresponding to the geometric center of the crowd according to the parameters of the camera; According to the viewing angle corresponding to the geometric center of the crowd, a corresponding brightness and chromaticity correction coefficient is determined, and according to the corresponding brightness and chromaticity correction coefficient, brightness and chromaticity correction is performed on the display screen to be corrected.

[0007] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method described above is implemented.

[0008] The beneficial effects of the present invention are as follows: by performing facial recognition and positioning on the crowd in front of the display screen, the position of each viewer in front of the display screen can be located, and the optimal viewing angle of all the viewers can be determined; by measuring the corresponding brightness and chromaticity correction coefficients at multiple viewing angles, the corresponding correction coefficients can be called according to the audience's optimal viewing angle, and the brightness and chromaticity of the display screen can be corrected, so that the LED display screen can achieve the best image display effect at different viewing angles, greatly improving the display quality of the LED display screen and the viewing effect of the audience. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 This is a schematic diagram of brightness and chromaticity correction of LED display screens in the prior art; Figure 2 It is a flow chart of a method for calibrating brightness and chromaticity of an LED display screen based on multiple viewing angles of the present invention; Figure 3 This is a flow chart of a method according to Embodiment 1 of the present invention; Figure 4 A schematic diagram of a world coordinate system, a camera coordinate system, and an image coordinate system according to a first embodiment of the present invention; Figure 5 Schematic diagram of the viewing angle range of embodiment 1 of the present invention. DETAILED DESCRIPTION

[0010] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in combination with the implementation modes and the accompanying drawings.

[0011] Please refer to Figure 2 , a method for correcting brightness and chromaticity of an LED display screen based on multiple viewing angles, comprising: Measure and obtain corresponding brightness and chromaticity correction coefficients at each preset viewing angle; According to a preset cycle, an image of the crowd in front of the display screen to be calibrated is obtained by taking pictures with a camera; Performing face recognition on the crowd image, and determining the geometric center of the crowd according to the face recognition result; Determine the viewing angle corresponding to the geometric center of the crowd according to the parameters of the camera; According to the viewing angle corresponding to the geometric center of the crowd, a corresponding brightness and chromaticity correction coefficient is determined, and according to the corresponding brightness and chromaticity correction coefficient, brightness and chromaticity correction is performed on the display screen to be corrected.

[0012] From the above description, it can be seen that the beneficial effect of the present invention is that the best display effect of the LED display screen at different viewing angles can be guaranteed through viewing angle perception and multi-viewing angle correction coefficients.

[0013] Furthermore, before determining the viewing angle corresponding to the geometric center of the crowd according to the parameters of the camera, the method further includes: According to preset image distortion correction parameters, distortion correction is performed on the geometric center of the crowd.

[0014] From the above description, it can be seen that through distortion correction, the geometric center of the crowd is closer to its actual position in the real world, which can greatly improve the accuracy of subsequent perspective calculation; and, performing distortion correction after calculating the geometric center of the crowd, that is, only performing distortion correction on the geometric center of the crowd, can improve calculation efficiency.

[0015] Furthermore, the corresponding brightness and chromaticity correction coefficients are obtained by measuring at each preset viewing angle, including: Traverse a set of preset perspectives, and sequentially obtain a preset perspective as the current perspective; The display screen to be corrected is rotated to the current viewing angle and displays the uncorrected image, and the brightness and chromaticity of each pixel when the display screen to be corrected displays the uncorrected image is used as the initial brightness and chromaticity of each pixel; The average brightness and chromaticity of each pixel in the preset area of ​​the display screen to be corrected is used as the correction target brightness and chromaticity; The correction coefficient of each pixel is calculated according to the correction target brightness and chromaticity and the initial brightness and chromaticity of each pixel, and the brightness and chromaticity correction coefficient matrix corresponding to the current viewing angle is obtained according to the correction coefficient of each pixel.

[0016] Furthermore, performing face recognition on the crowd image and determining the geometric center of the crowd according to the face recognition result includes: Through a preset deep learning model, face recognition is performed on the crowd image to obtain the face positioning position, and the arithmetic mean of each face positioning position is calculated to obtain the geometric center of the crowd.

[0017] In an optional embodiment, the deep learning model may adopt P2P-Net.

[0018] Furthermore, determining the viewing angle corresponding to the geometric center of the crowd according to the parameters of the camera includes: According to the parameters of the camera, the geometric center of the crowd is mapped from the image coordinate system to the world coordinate system to obtain the world coordinates of the geometric center of the crowd in the world coordinate system; Calculating the viewing angle corresponding to the geometric center of the crowd according to the world coordinates of the geometric center of the crowd; The camera is arranged above the display screen to be calibrated, the optical axis of the camera lens and the center normal of the display screen to be calibrated are located in the same plane, and the same plane is perpendicular to the ground, and the parameters of the camera include installation height, horizontal field of view angle, vertical field of view angle, installation pitch angle and image resolution.

[0019] Furthermore, mapping the geometric center of the crowd from the image coordinate system to the world coordinate system according to the parameters of the camera to obtain the world coordinates of the geometric center of the crowd in the world coordinate system includes: Calculate the horizontal focal length, vertical focal length and image center coordinates according to the horizontal field of view angle, vertical field of view angle and image resolution; According to the horizontal focal length, the vertical focal length and the image center coordinates, the geometric center of the crowd is mapped from the image coordinate system to the camera coordinate system to obtain the coordinates of the geometric center of the crowd in the camera coordinate system; According to the installation height and the installation pitch angle, the geometric center of the crowd is mapped from the camera coordinate system to the world coordinate system to obtain the world coordinates of the geometric center of the crowd in the world coordinate system; The origin of the world coordinate system is the projection point of the optical center of the camera on the ground, the positive direction of the X-axis is the horizontal right direction, the positive direction of the Y-axis is the vertical upward direction, and the positive direction of the Z-axis is the direction of the projection of the shooting direction on the horizontal plane; The origin of the camera coordinate system is the optical center of the camera, the positive direction of the X-axis is the horizontal right direction, the positive direction of the Y-axis is the vertical upward direction, and the positive direction of the Z-axis is the direction of the projection of the shooting direction on the horizontal plane; The origin of the image coordinate system is the upper left corner of the image, the positive direction of the u-axis is the horizontal rightward direction, and the positive direction of the v-axis is the vertical downward direction.

[0020] Further, the geometric center of the crowd is mapped from the image coordinate system to the camera coordinate system through a first mapping formula; and the geometric center of the crowd is mapped from the camera coordinate system to the world coordinate system through a second mapping formula; The first mapping formula is:

[0021] The second mapping formula is:

[0022]

[0023]

[0024] Among them, (u 0 , v 0 , 1) represents the coordinates of the geometric center of the crowd in the image coordinate system; (x c0 ,y c0 , z c0 ) represents the coordinates of the geometric center of the crowd in the camera coordinate system; (x w0 ,y w0 , z w0 ) represents the world coordinates of the geometric center of the crowd in the world coordinate system; f x represents the horizontal focal length, f y Indicates the vertical focal length; (c w , c h ) represents the image center coordinates; θ represents the installation pitch angle of the camera; H represents the installation height of the camera.

[0025] From the above description, it can be seen that by mapping the geometric center of the crowd from the image coordinate system to the camera coordinate system and then to the world coordinate system, the position of the geometric center of the crowd in the actual scene can be obtained.

[0026] Furthermore, the FOV corresponding to the geometric center of the crowd is arctan (x w0 / z w0 ).

[0027] Furthermore, the determining of the corresponding brightness and chromaticity correction coefficient according to the viewing angle corresponding to the geometric center of the crowd includes: If the viewing angle corresponding to the geometric center of the crowd is within a range corresponding to a preset viewing angle, a brightness and chromaticity correction coefficient corresponding to the viewing angle is obtained, and brightness and chromaticity correction is performed on the display screen to be corrected according to the corresponding brightness and chromaticity correction coefficient.

[0028] It can be seen from the above description that since it is impossible and unnecessary to measure the corresponding correction coefficient for each viewing angle, when the viewing angle is within a certain viewing angle range, the correction coefficient corresponding to the viewing angle range is selected.

[0029] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method described above is implemented.

[0030] Embodiment 1 Please refer to Figure 3-5 , Embodiment 1 of the present invention is: a method for calibrating brightness and chromaticity of an LED display screen based on multiple viewing angles, such as Figure 3 As shown, the following steps are included: S1: Measure and obtain corresponding brightness and chromaticity correction coefficients at each preset viewing angle.

[0031] A set of different viewing angles is set, and the LED display screen is rotated to different viewing angles. At each viewing angle, the display screen first displays an image without brightness and chromaticity correction, and then the display screen is measured at the viewing angle to generate corresponding brightness and chromaticity correction coefficients. Specifically, the following steps are included: S101: Traverse a set of preset viewing angles and sequentially obtain a viewing angle as the current viewing angle.

[0032] S102: The display screen to be calibrated is rotated to the current viewing angle and displays the uncalibrated image, and the brightness of each pixel when the display screen to be calibrated displays the uncalibrated image is used as the initial brightness and chromaticity of each pixel.

[0033] S103: taking the average brightness and chromaticity of each pixel in a preset area of ​​the display screen to be corrected as the correction target brightness and chromaticity.

[0034] In this embodiment, the average brightness and chromaticity of each pixel in the central area of ​​the display screen is selected as the correction target brightness and chromaticity. In other embodiments, the average brightness and chromaticity of other areas can also be selected as the correction target brightness and chromaticity.

[0035] S104: Calculate the correction coefficient of each pixel according to the correction target brightness and chromaticity and the initial brightness and chromaticity of each pixel, and obtain the brightness and chromaticity correction coefficient matrix corresponding to the current viewing angle according to the correction coefficient of each pixel.

[0036] In this embodiment, the correction coefficient of pixel i is calculated based on the correction target brightness and chromaticity and the initial brightness and chromaticity of pixel i, and the correction coefficients of all pixels in the display screen form a correction coefficient matrix.

[0037] Through the above steps, the brightness and chromaticity correction coefficient matrix corresponding to the current viewing angle can be obtained, and then the display screen to be corrected is rotated to the next viewing angle, and the brightness and chromaticity correction coefficient matrix corresponding to the next viewing angle is continuously measured, that is, a set of preset viewing angles is continuously traversed, and the next viewing angle is sequentially obtained as the current viewing angle, and then steps S102-S104 are executed. This is repeated until a set of preset viewing angles is traversed, and the brightness and chromaticity correction coefficient matrix corresponding to each preset viewing angle is obtained, and stored in the LED display control system to form a brightness and chromaticity correction coefficient library.

[0038] S2: According to a preset cycle, an image of a crowd in front of the display screen to be calibrated is obtained by photographing with a camera.

[0039] In this embodiment, the camera is installed directly above the LED display screen, and the optical axis of the lens and the center normal of the LED display screen are located in the same plane, and the plane is perpendicular to the ground.

[0040] Specifically, the crowd images in front of the display screen can be captured at a preset time interval, and the captured image files can be saved in a designated folder. Whether there are new image files in the folder can be detected in real time. When there are new image files, face recognition and positioning are performed on the new image files, that is, step S3 is executed.

[0041] S3: Performing face recognition on the crowd image, and determining the geometric center of the crowd according to the face recognition result.

[0042] Face recognition and positioning can be performed through a deep learning model. In this embodiment, a P2P-Net deep learning model is used to perform face recognition and positioning on crowd images collected by a camera, which mainly includes the following steps: 1. Image input If it is in the training stage, the sample image is input into the P2P-Net deep learning model; if it is in the application stage, the image captured by the camera is input into the P2P-Net deep learning model.

[0043] 2. Feature extraction P2P-Net first uses the pre-trained VGG16 network as the backbone network to extract the features of the input image. VGG16 is a convolutional neural network that can extract hierarchical features of images through a series of convolutional layers and pooling layers. These features contain visual information such as color, texture, and shape of the image.

[0044] 3. Candidate point generation Based on feature extraction, P2P-Net slides a small convolution kernel (such as 3×3) on the feature map through the Point Proposal Network (PPN) to generate a candidate point at each sliding position, which represents the possible individual position.

[0045] 4. Feature Upsampling In order to improve positioning accuracy, P2P-Net adopts an upsampling strategy to convert low-resolution feature maps into high-resolution feature maps through deconvolution or upsampling operations. This can retain more detailed information and help accurately locate individual positions.

[0046] 5. Prediction point coordinates and confidence It is divided into two branches: coordinate regression and confidence. The coordinate regression branch predicts the offset of each candidate point relative to the original position by applying a series of convolutional layers on the feature map; the confidence branch predicts the probability that each candidate point is a real individual by applying a series of convolutional layers on the feature map.

[0047] 6. Matching algorithm P2P-Net uses the Hungarian algorithm to find the lowest-cost match between predicted points and true points, ensuring that each true point matches at most one predicted point and vice versa.

[0048] 7. Positioning output Candidate points whose confidence scores exceed the preset threshold are valid individual positions. P2P-Net outputs the precise position of each individual, i.e., the face location, based on the confidence score and coordinate offset. The confidence score threshold is defined by the confidence_threshold parameter, with a default value of 0.5.

[0049] Furthermore, during the training process, P2P-Net uses a loss function to measure the prediction performance of the model and guide the optimization of the model. In this embodiment, the loss includes classification loss (used to determine whether the candidate point is a face) and coordinate regression loss (used for positioning). The calculation formula of the classification loss is:

[0050] Among them, y i represents the true label of the i-th prediction point (0 represents non-face, 1 represents face); represents the probability that the i-th predicted point is a face, and N represents the number of predicted points.

[0051] The coordinate regression loss calculation formula is:

[0052] Among them, x i represents the true coordinate value of the i-th prediction point, represents the predicted coordinate value of the i-th prediction point, and N represents the number of prediction points.

[0053] In this embodiment, the total loss L total is the coordinate regression loss L reg and classification loss L cls The weighted sum of , that is, the loss function is expressed as:

[0054] Among them, ω 1 and ω 2 Represents the weight coefficient, which is used to balance the importance of coordinate regression loss and classification loss.

[0055] During training, the model parameters are optimized by minimizing the total loss. The choice of weights usually depends on the needs of the specific task. If the positioning accuracy is more important than the classification accuracy, the weight of the coordinate regression loss can be increased. At the beginning of training, all weights can be set to the same value, such as ω 1 =ω 2 =1.

[0056] Through the trained deep learning model, the face recognition and positioning of the crowd images captured by the camera in the current cycle can be performed to obtain a set of face positioning positions, that is, the coordinates of each individual in the crowd (the coordinates in the image). By calculating the arithmetic mean of these coordinates, the geometric center of the crowd (u 0 , v 0 ), the calculation formula is:

[0057]

[0058] Among them, (u i , v i ) represents the coordinates of the i-th face location, and n represents the number of viewers in front of the display screen.

[0059] S4: Performing distortion correction on the geometric center of the crowd according to preset image distortion correction parameters.

[0060] Since the distortion of the camera lens itself may seriously affect the measurement accuracy, resulting in a deviation between the calculated viewing angle and the actual situation, distortion correction is required.

[0061] The distortion of camera lenses mainly includes radial distortion and tangential distortion. Radial distortion is the result of light bending due to the shape of the lens, making objects at the edge of the image larger or smaller than they actually are; tangential distortion is caused by the lens and imaging sensor not being completely parallel, resulting in pixel offset during imaging.

[0062] In this embodiment, distortion correction is performed by using the cv2.undistortPoints function, which is a tool for distortion correction in the OpenCV library. The cv2.undistortPoints function corrects the distorted pixel coordinates based on the camera's intrinsic parameters and distortion coefficients to restore their true positions in the undistorted image.

[0063] Specifically, first, through the camera calibration process, the camera's intrinsic matrix and distortion coefficient are obtained, where the intrinsic matrix contains the focal length and optical center coordinates of the camera, and the distortion coefficient describes the degree of distortion of the camera lens. Then, using the cv2.undistortPoints function, input the distorted pixel coordinates (i.e. the coordinates of the geometric center of the crowd), the intrinsic matrix and the distortion coefficient, and output the corrected pixel coordinates. This function can accurately correct the pixel coordinate offset caused by distortion through complex mathematical model calculations.

[0064] By performing distortion correction, the pixel coordinates of the geometric center of the crowd are closer to their real positions in the real world, which greatly improves the accuracy of perspective calculation. Especially in edge areas, uncorrected pixel coordinates will cause large errors due to severe distortion, but these errors are effectively reduced after distortion correction, thus ensuring the overall performance of the algorithm.

[0065] S5: Determine the viewing angle corresponding to the geometric center of the crowd after distortion correction according to the parameters of the camera.

[0066] First, determine the parameters of the camera. In this embodiment, the parameters of the camera include the installation height H (the height of the optical center of the camera from the ground), the horizontal field of view angle α (the viewing angle range of the camera in the horizontal direction), the vertical field of view angle β (the viewing angle range of the camera in the vertical direction), the installation pitch angle θ (that is, the angle between the optical axis of the camera and the horizontal plane. When the optical axis and the horizontal plane are coplanar, θ=0°; when the camera is tilted downward, θ>0°; when the camera is tilted upward, θ<0°), and the image resolution w×h (w pixels in the width direction and h pixels in the height direction).

[0067] Then, if Figure 4 As shown, the world coordinate system, camera coordinate system and image coordinate system are defined as follows: World Coordinate System O w X w Y w Z w : Taking the ground as the reference, the projection point of the camera optical center on the ground is the origin O w , the horizontal shooting direction is Z w The positive direction of the axis, horizontal right direction is X w Positive direction of axis, perpendicular to X w Z w The upward direction of the plane is Y w Positive axis direction.

[0068] Camera coordinate system O c X c Y c Z c :The camera optical center is the origin Oc , the horizontal shooting direction is Z c The positive direction of the axis, horizontal right direction is X c Positive direction of axis, perpendicular to X c Z c The upward direction of the plane is Y c Positive axis direction.

[0069] Image coordinate system ouv: The upper left corner of the image is taken as the origin o, the horizontal right direction is the positive direction of the u axis, and the vertical downward direction is the positive direction of the v axis.

[0070] Next, the horizontal focal length f is calculated based on the horizontal field of view angle α, the vertical field of view angle β and the image resolution. x , vertical focal length f y and the image center coordinates (c w , c h ), the specific calculation formula is as follows:

[0071]

[0072]

[0073]

[0074] Then, according to the horizontal focal length f x , vertical focal length f y and the image center coordinates (c w , c h ), the geometric center of the crowd after distortion correction (u 0 ',v 0 ') Map from the image coordinate system to the camera coordinate system to obtain the coordinates of the geometric center of the crowd in the camera coordinate system (x c0 ,y c0 , z c0 ), the specific mapping formula is as follows:

[0075] Then, according to the installation height H and the installation pitch angle θ, the geometric center of the crowd is mapped from the camera coordinate system to the world coordinate system to obtain the world coordinates (x w0 ,y w0 , z w0 ), the specific mapping formula is as follows:

[0076] Among them, R represents the rotation matrix and T represents the translation vector. Since the yaw angle (Yaw) and roll angle (Roll) of the camera are 0°, only the pitch angle (Pitch) is used to adjust the rotation angle of the camera around the X axis, so the rotation matrix R is:

[0077] The translation vector T is:

[0078] Finally, according to the world coordinates (x w0 ,y w0 , z w0 ), calculate the FOV corresponding to the geometric center of the crowd. The specific calculation formula is as follows:

[0079] S6: Determine a corresponding brightness and chromaticity correction coefficient according to the viewing angle corresponding to the geometric center of the crowd, and perform brightness and chromaticity correction on the display screen to be corrected according to the corresponding brightness and chromaticity correction coefficient. Specifically, the viewing angle obtained in the above steps is stored in a designated folder. Whether there is a new viewing angle in the folder is monitored in real time, and when there is a new viewing angle, the viewing angle is read. In this embodiment, when the viewing angle is within a certain viewing angle range, the correction coefficient corresponding to the range is selected, that is, the area where the viewing angle is located is determined, thereby determining the brightness and chromaticity correction coefficient that needs to be called.

[0080] For example, Figure 5 As shown, it is assumed that the display control system has stored 0°, ±A 1 ±A 2 ±A 3 The brightness and chromaticity correction coefficients corresponding to these 7 viewing angles, -α 3 , -α 2 , -α 1 , α 1 , α 2 , α 3 is the viewing angle boundary value, and the brightness and chromaticity correction coefficient calling rules are as follows: When FOV<-α 3 When calling -A 3 Brightness and chromaticity correction coefficients corresponding to viewing angle; When the FOV is in [-α 3 , -α 2 ), call -A 2 Brightness and chromaticity correction coefficients corresponding to viewing angle; When the FOV is in [-α 2 , -α 1 ), call -A 1Brightness and chromaticity correction coefficients corresponding to viewing angle; When the FOV is in [-α 1 , α 1 ], call the brightness and chromaticity correction coefficient corresponding to the 0° viewing angle; When the FOV is (α 1 , α 2 ], call A 1 Brightness and chromaticity correction coefficients corresponding to viewing angle; When the FOV is (α 2 , α 3 ], call A 2 Brightness and chromaticity correction coefficients corresponding to viewing angle; When FOV>α 3 When calling A 3 Brightness and chromaticity correction coefficients corresponding to viewing angle.

[0081] Finally, according to the determined brightness and chromaticity correction coefficients, the LED display control system performs brightness and chromaticity correction on the LED display.

[0082] In actual application scenarios, the viewing angle perception error is within ±2°, the brightness and chromaticity correction time is adjustable, and the fastest brightness and chromaticity correction time can be <1s, which can fully realize real-time correction of the LED display.

[0083] In this embodiment, a correction coefficient library for the LED display screen at different viewing angles is first established, and each viewer in front of the display screen is positioned through a deep learning model to obtain the optimal viewing angle for all viewers. According to the optimal viewing angle, the corresponding brightness and chromaticity correction coefficients are called to correct the display screen, and finally the optimal display effect of the LED display screen at different viewing angles is achieved.

[0084] This embodiment adopts the viewing angle perception technology and multi-viewing angle correction technology, which can locate the position of each audience in front of the display screen, thereby obtaining the best viewing angle for all audiences, and realizing real-time correction of the LED display screen at different viewing angles, ensuring that the LED display screen can have the best display effect at different viewing angles, greatly improving the display quality of the LED display screen and the viewing effect of the audience.

[0085] Embodiment 2 This embodiment is a computer-readable storage medium corresponding to the above embodiment, on which a computer program is stored. When the program is executed by the processor, the various steps of the brightness and chromaticity correction method of the LED display screen based on multi-viewing angles in the above embodiment are implemented, and the same technical effect can be achieved, which will not be repeated here.

[0086] In summary, the present invention provides a method and storage medium for correcting brightness and chromaticity of an LED display screen based on multiple viewing angles. By performing face recognition and positioning on the image in front of the display screen, the position of each viewer in front of the display screen can be located, and the optimal viewing angle for all viewers can be determined. By measuring the corresponding brightness and chromaticity correction coefficients at multiple viewing angles, the corresponding correction coefficients can be called according to the audience's optimal viewing angle to correct the brightness and chromaticity of the display screen, so that the LED display screen can achieve the best image display effect at different viewing angles, greatly improving the display quality of the LED display screen and the viewing effect of the audience.

[0087] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for luminance and chromaticity correction of LED display screen based on multi-viewing angles, characterized in that: include: Measure and obtain corresponding brightness and chromaticity correction coefficients at each preset viewing angle; According to a preset cycle, an image of the crowd in front of the display screen to be calibrated is obtained by taking pictures with a camera; Performing face recognition on the crowd image, and determining the geometric center of the crowd according to the face recognition result; Determine the viewing angle corresponding to the geometric center of the crowd according to the parameters of the camera; According to the viewing angle corresponding to the geometric center of the crowd, a corresponding brightness and chromaticity correction coefficient is determined, and according to the corresponding brightness and chromaticity correction coefficient, brightness and chromaticity correction is performed on the display screen to be corrected.

2. The calibration method according to claim 1, characterized in that: Before determining the viewing angle corresponding to the geometric center of the crowd according to the parameters of the camera, the method further includes: According to preset image distortion correction parameters, distortion correction is performed on the geometric center of the crowd.

3. The calibration method according to claim 1, characterized in that: The corresponding brightness and chromaticity correction coefficients are obtained by measuring at each preset viewing angle, including: Traverse a set of preset perspectives, and sequentially obtain a preset perspective as the current perspective; The display screen to be corrected is rotated to the current viewing angle and displays the uncorrected image, and the brightness and chromaticity of each pixel when the display screen to be corrected displays the uncorrected image is used as the initial brightness and chromaticity of each pixel; The average value of the brightness and chromaticity of each pixel in the preset area of ​​the display screen to be corrected is used as the correction target brightness and chromaticity; The correction coefficient of each pixel is calculated according to the correction target brightness and chromaticity and the initial brightness and chromaticity of each pixel, and the brightness and chromaticity correction coefficient matrix corresponding to the current viewing angle is obtained according to the correction coefficient of each pixel.

4. The calibration method according to claim 1, characterized in that: The performing face recognition on the crowd image and determining the geometric center of the crowd according to the face recognition result includes: Through a preset deep learning model, face recognition is performed on the crowd image to obtain the face positioning position, and the arithmetic mean of each face positioning position is calculated to obtain the geometric center of the crowd.

5. The calibration method according to claim 1, characterized in that: Determining the viewing angle corresponding to the geometric center of the crowd according to the parameters of the camera includes: According to the parameters of the camera, the geometric center of the crowd is mapped from the image coordinate system to the world coordinate system to obtain the world coordinates of the geometric center of the crowd in the world coordinate system; Calculating the viewing angle corresponding to the geometric center of the crowd according to the world coordinates of the geometric center of the crowd; The camera is arranged above the display screen to be calibrated, the optical axis of the camera lens and the center normal of the display screen to be calibrated are located in the same plane, and the same plane is perpendicular to the ground, and the parameters of the camera include installation height, horizontal field of view angle, vertical field of view angle, installation pitch angle and image resolution.

6. The calibration method according to claim 5, characterized in that: Mapping the geometric center of the crowd from the image coordinate system to the world coordinate system according to the parameters of the camera to obtain the world coordinates of the geometric center of the crowd in the world coordinate system includes: Calculate the horizontal focal length, vertical focal length and image center coordinates according to the horizontal field of view angle, vertical field of view angle and image resolution; According to the horizontal focal length, the vertical focal length and the image center coordinates, the geometric center of the crowd is mapped from the image coordinate system to the camera coordinate system to obtain the coordinates of the geometric center of the crowd in the camera coordinate system; According to the installation height and the installation pitch angle, the geometric center of the crowd is mapped from the camera coordinate system to the world coordinate system to obtain the world coordinates of the geometric center of the crowd in the world coordinate system; The origin of the world coordinate system is the projection point of the optical center of the camera on the ground, the positive direction of the X-axis is the horizontal right direction, the positive direction of the Y-axis is the vertical upward direction, and the positive direction of the Z-axis is the direction of the projection of the shooting direction on the horizontal plane; The origin of the camera coordinate system is the optical center of the camera, the positive direction of the X-axis is the horizontal right direction, the positive direction of the Y-axis is the vertical upward direction, and the positive direction of the Z-axis is the direction of the projection of the shooting direction on the horizontal plane; The origin of the image coordinate system is the upper left corner of the image, the positive direction of the u-axis is the horizontal rightward direction, and the positive direction of the v-axis is the vertical downward direction.

7. The calibration method according to claim 6, characterized in that: The geometric center of the crowd is mapped from the image coordinate system to the camera coordinate system by a first mapping formula; the geometric center of the crowd is mapped from the camera coordinate system to the world coordinate system by a second mapping formula; The first mapping formula is: The second mapping formula is: Wherein, (u0, v0, 1) represents the coordinates of the geometric center of the crowd in the image coordinate system; (x c0 ,y c0 , z c0 ) represents the coordinates of the geometric center of the crowd in the camera coordinate system; (x w0 ,y w0 , z w0 ) represents the world coordinates of the geometric center of the crowd in the world coordinate system; f x represents the horizontal focal length, f y Indicates the vertical focal length; (c w , c h ) represents the image center coordinates; θ represents the installation pitch angle of the camera; H represents the installation height of the camera.

8. The calibration method according to claim 7, characterized in that: The FOV corresponding to the geometric center of the crowd = arctan (x w0 / z w0 ).

9. The calibration method according to claim 1, characterized in that: The step of determining the corresponding brightness and chromaticity correction coefficient according to the viewing angle corresponding to the geometric center of the crowd includes: If the viewing angle corresponding to the geometric center of the crowd is within a range corresponding to a preset viewing angle, a brightness and chromaticity correction coefficient corresponding to the viewing angle is obtained, and brightness and chromaticity correction is performed on the display screen to be corrected according to the corresponding brightness and chromaticity correction coefficient.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

Images