A data dynamic splicing method and system for multi-camera curved screen Demura compensation

Abstract

This invention provides a dynamic data stitching method and system for Demura compensation on multi-camera curved screens, relating to the field of display technology. The method includes performing flat-field correction, dark current correction, and dead pixel correction on multiple cameras to obtain uniformity-corrected camera data; generating a positioning map and acquiring data through the corrected cameras, calculating the sub-pixel coordinate mapping relationship for each camera; dynamically estimating the curvature change of the curved screen, constructing a curved surface calibration model, and generating curved surface correction coefficients; correcting the brightness range of the curved surface correction coefficients to obtain corrected dynamic curved surface correction coefficients; performing brightness correction on the grayscale images acquired by the multiple cameras, and achieving seamless data stitching through brightness calibration between cameras. This method effectively solves the data acquisition and stitching challenges of large-size curved screens, is applicable to various display devices, reduces hardware costs and process complexity, and has high accuracy, flexibility, and broad application prospects.

Description

Technical Field

[0001] This invention relates to the field of display technology, and in particular to a method and system for dynamic data stitching with Demura compensation for multi-camera curved screens. Background Technology

[0002] In the current wave of technological development, display technology has made significant progress in many fields, especially in automotive displays. With the booming development of the automotive industry and the ever-increasing demand for in-car entertainment experiences, the application scope of automotive displays continues to expand. From traditional dashboards to feature-rich central control screens and rear-seat in-car TVs, automotive displays are not only increasing in size but also becoming more diverse in form, gradually moving towards larger sizes and curved surfaces. While this trend enhances the visual experience, it also brings many challenges to the Mura compensation process in display technology.

[0003] 1. Camera Resolution Limitations Due to Large Screen Sizes: While camera resolution has advanced rapidly, it still falls short when faced with ever-increasing automotive screen sizes. Even the highest resolution cameras currently available struggle to capture complete data from large screens. For example, the ever-expanding size of automotive OLED curved screens has exceeded the resolution limits of existing cameras, making it impossible for a single camera to acquire complete and accurate screen data. This severely restricts the implementation of the Mura compensation process, hindering the precise analysis and correction of the screen's brightness and darkness characteristics.

[0004] 2. Optical System Problems Caused by Curved Screen Characteristics: On the one hand, the curvature of a curved screen often exceeds the depth of field range of the optical system. Because the distance from different positions on the curved screen to the camera's sensor varies, the distance from the camera gradually changes relative to the optical axis center, resulting in reduced brightness in areas outside the depth of field. This difference in brightness is not caused by the screen's own light-emitting characteristics, but rather by optical system errors. When these optical system errors are mixed with the screen's inherent brightness characteristics, it becomes impossible to accurately capture the screen's true brightness characteristics, greatly hindering Mura compensation. On the other hand, increased product curvature leads to complex and diverse surface shapes, and the height differences further exacerbate the optical system problems. The effective area that a single camera can capture is extremely limited, necessitating the use of multiple cameras to collect data.

[0005] 3. The Challenge of Stitching Data from Multiple Cameras: In exploring solutions to the aforementioned problems through multi-camera data acquisition, new challenges have emerged. Differences in the consistency of optical systems and the varying characteristics of different cameras make seamless data stitching difficult. Variations in camera focus, brightness response, and focus plane position can all cause problems during stitching, affecting the final stitching effect and data accuracy.

[0006] 4. Limitations of Existing Demura Technology: Currently, most traditional Demura technologies rely on a single camera to collect complete screen data to achieve accuracy. However, as screen sizes continue to increase and camera resolution becomes more difficult to improve, this approach is gradually becoming insufficient. For example, Chinese patent CN119364194A discloses a brightness splicing method and system for Demura technology applied to medium and large-sized screens. Although it proposes a multi-camera data splicing scheme based on correcting optical system errors, and the effect is acceptable for flat screen data splicing, when processing curved screen data splicing, the method of calculating the correction coefficient based on local curvature needs to be preset according to product characteristics, lacking flexibility and unable to adapt well to complex and diverse curved screen shapes.

[0007] With the continuous pursuit of high image quality, large size and curved surfaces in automotive display technology, existing technologies have many shortcomings in handling Demura compensation data splicing for multi-camera curved screens. There is an urgent need for an innovative technology to break through these bottlenecks, achieve more accurate and efficient data splicing and Mura compensation, and improve the quality of automotive displays. Summary of the Invention

[0008] To address these issues, this invention provides a method and system for dynamic data stitching with Demura compensation for multi-camera curved screens. This method solves the problems in the prior art, such as insufficient camera resolution due to the increased size of automotive OLED curved screens, difficulties in data acquisition by optical systems due to the curved surface characteristics, the difficulty in seamlessly stitching multi-camera data, and the poor flexibility of existing Demura technology in handling curved screen data stitching.

[0009] To address the aforementioned problems, embodiments of the present invention provide a method for dynamic data stitching for Demura compensation on multi-camera curved screens, comprising the following steps:

[0010] Step S1: Perform flat field correction, dark current correction and distortion correction on multiple cameras to obtain the camera after uniformity correction;

[0011] Step S2: Generate a positioning map and collect data using multiple calibrated cameras, then calculate the sub-pixel coordinate mapping relationship for each camera.

[0012] Step S3: Based on the sub-pixel coordinate mapping relationship, dynamically estimate the curvature change of the curved screen, construct a curved surface calibration model, and generate curved surface correction coefficients;

[0013] Step S4: Correct the brightness range of the surface correction coefficient to obtain the corrected dynamic surface correction coefficient;

[0014] Step S5: Based on the corrected dynamic surface correction coefficient, perform brightness correction on the grayscale images captured by multiple cameras, and achieve seamless data stitching through brightness calibration between cameras.

[0015] Preferably, the dynamic estimation of the curvature change of the curved screen, the construction of the curved surface calibration model, and the generation of curved surface correction coefficients specifically include:

[0016] Select the pixel coordinates of the central region of the sub-pixel coordinate mapping relationship, and calculate the interval of the sub-pixels in the horizontal and vertical directions;

[0017] Based on the aforementioned interval, standard equally spaced sub-pixel coordinates are constructed, and the deviation between the actual sub-pixel coordinates and the standard sub-pixel coordinates is calculated.

[0018] After filtering the deviation, a surface is fitted, and the surface correction coefficient is obtained by normalizing it according to the center of the surface.

[0019] Preferably, the step of calculating the deviation between the actual sub-pixel coordinates and the standard sub-pixel coordinates specifically involves: obtaining the positional deviation of the actual sub-pixel coordinates relative to the standard sub-pixel coordinates in the horizontal or vertical direction based on the bending direction of the curved surface.

[0020] Preferably, the step of correcting the brightness range of the surface correction coefficient specifically includes:

[0021] Acquire grayscale images and calculate the horizontal or vertical projections. After filtering, obtain the ratio of the minimum value to the center value as the correction value.

[0022] Based on the correction value, the surface correction coefficient is superimposed and corrected to obtain a dynamic surface correction coefficient that adapts to the actual trend of brightness and darkness difference on the surface.

[0023] Preferably, the seamless data stitching achieved through brightness calibration between cameras specifically includes:

[0024] Select the overlapping area of ​​adjacent cameras along the camera stitching direction and calculate the brightness ratio of the overlapping area;

[0025] Based on the brightness ratio, the brightness data of adjacent cameras are normalized and corrected, and then stitched together sequentially until the fusion of all camera data is completed.

[0026] Preferably, the generated positioning map is acquired by multiple calibrated cameras, and the sub-pixel coordinate mapping relationship corresponding to each camera is calculated. Specifically, based on the sub-pixel coordinate mapping technology of the coded positioning map, the sub-pixel coordinates Map1, Map2, ..., MapN of each camera partition are obtained, where N represents the number of cameras.

[0027] This invention also provides a dynamic data stitching system for Demura compensation on multi-camera curved screens. This system is used to implement the aforementioned dynamic data stitching method for Demura compensation on multi-camera curved screens, specifically including:

[0028] The camera correction module is used to perform flat field correction, dark current correction and distortion correction on multiple cameras to obtain a camera with uniformity correction.

[0029] The localization map processing module is used to generate localization maps and acquire data through multiple calibrated cameras, and calculate the sub-pixel coordinate mapping relationship for each camera.

[0030] The surface estimation module is used to dynamically estimate the surface curvature based on the sub-pixel coordinate mapping relationship, construct a surface calibration model, and generate surface correction coefficients.

[0031] A brightness correction module is used to correct the brightness range of the surface correction coefficient to obtain a dynamic surface correction coefficient.

[0032] The data stitching module is used to perform brightness correction on the grayscale image based on the dynamic surface correction coefficient, and to achieve seamless data stitching through brightness calibration between cameras.

[0033] Preferably, the surface estimation module includes:

[0034] The interval calculation unit is used to select the pixel coordinates of the central region of the sub-pixel coordinate mapping relationship and calculate the interval of the sub-pixels in the horizontal and vertical directions.

[0035] The deviation calculation unit is used to construct standard equally spaced sub-pixel coordinates based on the interval, and to calculate the positional deviation between the actual sub-pixel coordinates and the standard sub-pixel coordinates;

[0036] The surface fitting unit is used to filter the deviation and fit the surface to generate surface correction coefficients.

[0037] Preferably, the brightness correction module includes:

[0038] The correction value calculation unit is used to acquire grayscale images and calculate the horizontal or vertical projection, and obtain the ratio of the minimum value to the center value as the correction value.

[0039] The coefficient correction unit is used to superimpose and correct the surface correction coefficients based on the correction value to obtain the dynamic surface correction coefficients.

[0040] Preferably, the data splicing module includes:

[0041] The overlapping area processing unit is used to select the overlapping area of ​​adjacent cameras along the camera stitching direction and calculate the brightness ratio of the overlapping area.

[0042] The successive stitching unit is used to normalize and correct the brightness data of adjacent cameras based on the brightness ratio, and to complete the fusion of all camera data by successive stitching.

[0043] As can be seen from the above technical solutions, this invention application has the following beneficial effects:

[0044] (1) Improved splicing accuracy and flexibility: This application solves the rigidity problem of traditional preset curvature correction coefficients in complex curved screen applications by calibrating the brightness relationship between cameras in real time, dynamically estimating the curvature change of the curved screen, and constructing a curved screen calibration model. For example, when facing automotive OLED curved screens of different sizes and curvatures, it can adjust in real time according to the pixel coordinate offset to generate accurate curved screen correction coefficients, making the splicing accuracy of multi-camera data higher, the adaptability stronger, and able to meet diverse display needs.

[0045] (2) Reduced Costs and Complexity: High-precision data acquisition is achieved through multi-camera collaboration and optimization algorithms, eliminating the need for ultra-high-resolution cameras or customized optical systems. On the one hand, this avoids the use of high-cost hardware equipment, reducing hardware procurement costs; on the other hand, it reduces reliance on complex optical systems, lowering process complexity. Taking the inspection of automotive displays as an example, multiple relatively low-cost industrial cameras, combined with the algorithm of this application, can achieve data acquisition and splicing of large-size curved screens, improving production efficiency while reducing overall costs.

[0046] (3) Enhanced adaptability to application scenarios: This technical solution is applicable to display devices of different sizes and curvatures, such as 40-inch curved screens and 8-inch foldable screens for vehicles. Its dynamic model can automatically adjust parameters according to the screen shape, breaking through the limitation of a single scenario. It can be applied to both in-vehicle display systems and VR / AR display systems, greatly expanding the application scope of the technology. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the embodiments will be briefly described below. Referring to the accompanying drawings will provide a clearer understanding of the features and advantages of the present invention. The drawings are illustrative and should not be construed as limiting the present invention in any way. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort. Wherein:

[0048] Figure 1 A flowchart of a dynamic data stitching method for Demura compensation on a multi-camera curved screen provided by the present invention;

[0049] Figure 2 These are data graphs before and after flat-field FFC correction in this invention;

[0050] Figure 3 This is a positioning diagram of the present invention;

[0051] Figure 4 This is a positioning map captured by the camera of this invention;

[0052] Figure 5 This is a partial visualization of the pixel mapping Map1 and Map2 of the present invention;

[0053] Figure 6 A comparison diagram of the standard sub-pixel coordinate Map_std constructed for this invention and the actual sub-pixel coordinate Map;

[0054] Figure 7 This is a schematic diagram of the positional deviation of the present invention;

[0055] Figure 8 This is a schematic diagram of the deviation after filtering according to the present invention;

[0056] Figure 9 This is a schematic diagram of the projection data of the present invention;

[0057] Figure 10 This is a comparison chart of unprocessed and processed brightness data from this invention;

[0058] Figure 11 This invention provides a block diagram of a dynamic data stitching system for Demura compensation on a multi-camera curved screen. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] Example 1:

[0061] To address the issues in existing technologies such as insufficient camera resolution due to the increased size of automotive OLED curved screens, difficulties in data acquisition by optical systems due to the curved surface characteristics, challenges in seamless stitching of multi-camera data, and the poor flexibility of existing Demura technology in handling curved screen data stitching, such as... Figure 1 As shown, this invention proposes a dynamic data stitching method for Demura compensation on multi-camera curved screens, the method comprising:

[0062] Step S1: Perform flat field correction, dark current correction, and bad pixel correction on multiple cameras to obtain the camera after uniformity correction;

[0063] Step S2: Generate a positioning map and collect data using multiple calibrated cameras, then calculate the sub-pixel coordinate mapping relationship for each camera.

[0064] Step S3: Based on the sub-pixel coordinate mapping relationship, dynamically estimate the curvature change of the curved screen, construct a curved surface calibration model, and generate curved surface correction coefficients;

[0065] Step S4: Correct the brightness range of the surface correction coefficient to obtain the corrected dynamic surface correction coefficient;

[0066] Step S5: Based on the corrected dynamic surface correction coefficient, perform brightness correction on the grayscale images captured by multiple cameras, and achieve seamless data stitching through brightness calibration between cameras.

[0067] As can be seen from the above technical solution, this invention proposes a dynamic data stitching method for Demura compensation on multi-camera curved screens. By performing flat field, dark current, and distortion correction on multiple cameras, the accuracy and stability of the camera-acquired data are ensured. Sub-pixel coordinate mapping based on coded positioning maps is used to calculate sub-pixel coordinate mapping relationships, improving positioning accuracy. Dynamic estimation of the curvature changes of the curved screen is used to construct a calibration model to generate correction coefficients, and brightness range correction is performed, enhancing adaptability to different curved screens and the flexibility of correction. Finally, the corrected coefficients are used to correct the brightness of the grayscale image and achieve seamless data stitching under brightness calibration between cameras, eliminating stitching color difference and brightness discontinuity caused by optical system errors. This method effectively solves the data acquisition and stitching problem of large-size curved screens, is applicable to various display devices, reduces hardware costs and process complexity, and has high accuracy, flexibility, and broad application prospects.

[0068] In step S1, flat field correction, dark current correction, and bad pixel correction are performed on multiple cameras to obtain a camera with uniformity correction.

[0069] In practical applications, taking the inspection of automotive displays as an example, multiple industrial cameras are first selected, such as two 151MVeworks high-resolution industrial cameras (the appropriate model and number of cameras can be selected according to factors such as screen size and accuracy requirements in different scenarios). When performing planar calibration on these cameras, the cameras are used to capture images under a uniform light source, for example, using a high-precision white exposure calibration plate as the uniform light source. Image data is acquired through these images, and the planar response curve of each camera is calculated. Based on this curve, equalization processing is performed to ensure that the response of each pixel to the same illumination is consistent, effectively avoiding image brightness unevenness caused by the camera's inherent characteristics. A comparison of data before and after calibration can intuitively present the calibration effect, such as... Figure 2 As shown.

[0070] Next, dark current correction is performed by turning off the light source and acquiring a dark current image. Defects are marked in the dark current image; these defects may be caused by manufacturing flaws in the camera chip or aging due to prolonged use. A suitable denoising algorithm is then used to denoise the dark current image, removing noise interference and ensuring the accuracy of subsequent data acquisition.

[0071] Finally, distortion correction was performed using a checkerboard calibration method to calculate optical distortion. In a darkroom environment (background illuminance <1 lux, ensuring minimal impact of ambient light on test results), a checkerboard calibration board was used to capture checkerboard images from different angles and positions. By analyzing and calculating these images, the camera's intrinsic parameters (such as focal length and principal point position) and extrinsic parameters (such as camera rotation and translation parameters) were determined, thereby correcting the camera's optical distortion and making the captured images more closely resemble the real scene.

[0072] In this embodiment, in step S2, a positioning map is generated and acquired by multiple calibrated cameras, and the sub-pixel coordinate mapping relationship corresponding to each camera is calculated.

[0073] Specifically, this invention uses sub-pixel coordinate mapping technology based on encoded positioning maps (such as the positioning map design scheme proposed in Chinese Patent CN119364194A) to create positioning maps, such as... Figure 3 As shown. The design of this positioning map needs to consider the accuracy and convenience of subsequent calculations of sub-pixel coordinate mapping relationships. Multiple calibrated cameras, such as the two Veworks cameras mentioned above, are used to capture positioning maps, as shown. Figure 4 As shown.

[0074] After shooting, the location map data acquired by the cameras is processed to calculate the sub-pixel coordinate mapping relationship corresponding to each camera. Specifically, the algorithm analyzes the feature information in the location map to obtain the sub-pixel coordinates Map1 and Map2 for each camera partition (if there are multiple cameras, Map3, ..., MapN are obtained sequentially). Taking two cameras as an example, the processed pixel mappings Map1 and Map2 can be displayed through local visualization, such as... Figure 5 As shown, this facilitates observation and verification of the accuracy of the mapping relationship.

[0075] In step S3, based on the sub-pixel coordinate mapping relationship, the curvature change of the curved screen is dynamically estimated, a curved surface calibration model is constructed, and curved surface correction coefficients are generated.

[0076] Specifically, for each sub-pixel coordinate mapping relationship Map obtained in step S2, the pixel coordinates of the 5×5 area at the center of the Map are selected. These pixels correspond to the region pixel coordinates of the camera optical axis center or the focal plane. Based on the pixel coordinates of this region, the horizontal spacing stepH and the vertical spacing stepV of the sub-pixels are calculated.

[0077] Based on the calculated intervals stepH and stepV, and combined with the Map index, a standard equally spaced sub-pixel coordinate Map_std is constructed. The actual sub-pixel coordinate Map is then compared with the constructed standard sub-pixel coordinate Map_std. Figure 6 As shown. The deviation D is calculated based on the bending direction of the surface. If the surface bends in the X direction, the x-coordinate deviation is used; if it bends in the Y direction, the y-coordinate deviation is used. Figure 7 As shown.

[0078] To eliminate interference factors in the deviation data, the deviation D is filtered. Appropriate filtering algorithms, such as Gaussian filtering, can be used to make the deviation data smoother and more accurate. Figure 8 As shown, the filtered data is fitted to a surface using least squares and other fitting algorithms to form a surface model. Normalization is then performed along the surface center to obtain the surface correction coefficient coeff. This coefficient reflects the bending characteristics of the curved screen and provides a basis for subsequent data correction.

[0079] In step S4, the brightness range of the surface correction coefficient is corrected to obtain the corrected dynamic surface correction coefficient.

[0080] Specifically, grayscale images are captured, and in vehicle-mounted display tests, images of different grayscale levels can be displayed on the screen. The horizontal or vertical projection of the captured grayscale images is calculated to obtain projection data, such as... Figure 9 As shown, the projected data is filtered to remove noise and outliers, making the data more stable and reliable. The ratio of the minimum value to the center value is then obtained from the filtered data, and this ratio is used as the correction value A.

[0081] The correction value A is superimposed on the surface correction coefficient coeff obtained in step S3, and the surface correction coefficient is scaled to make the brightness trend range reach the actual brightness-darkness difference trend of the surface, thereby obtaining the corrected dynamic surface correction coefficient coeff', ensuring the accuracy of subsequent brightness correction.

[0082] In this embodiment, in step S5, the brightness of grayscale images captured by multiple cameras is corrected based on the modified dynamic surface correction coefficient, and seamless data stitching is achieved through brightness calibration between cameras.

[0083] Specifically, multiple cameras capture grayscale images of each zone, and the brightness data of each zone's grayscale image is denoted as Lum1, Lum2, Lum3, ..., LumN. First, the brightness data of each zone is normalized using its respective exposure time to eliminate brightness differences caused by different camera exposure times.

[0084] Then, the normalized luminance data is corrected using the modified dynamic surface correction coefficient coeff' obtained in step S4, resulting in corrected luminance data Lum1', Lum2', Lum3', ..., LumN'.

[0085] When performing brightness calibration between cameras, overlapping areas of adjacent cameras are selected based on the camera stitching direction (horizontal or vertical). The brightness ratio of the overlapping areas is calculated, and the brightness data of adjacent cameras are normalized and corrected based on this ratio. For example, Lum2' is first corrected based on the brightness ratio of the overlapping areas of Lum1' and Lum2', and then the corrected Lum2' is stitched with Lum1' to obtain Lum_combine; then Lum3' is corrected based on the brightness ratio of the overlapping areas of Lum_combine and Lum3', and then stitched with Lum_combine to obtain a new Lum_combine. Following this rule, stitching is performed one by one along the camera stitching direction until all partitions are stitched, achieving seamless stitching of multi-camera data, such as... Figure 10 As shown, the splicing effect can be intuitively displayed.

[0086] Example 2:

[0087] like Figure 11 As shown, this invention provides a dynamic data stitching system for Demura compensation on multi-camera curved screens. This system is used to implement the dynamic data stitching method for Demura compensation on multi-camera curved screens described in Embodiment 1 above, specifically including:

[0088] The camera correction module 100 is used to perform flat field correction, dark current correction and distortion correction on multiple cameras to obtain a camera after uniformity correction.

[0089] The positioning map processing module 200 is used to generate a positioning map and acquire data through multiple calibrated cameras, and to calculate the sub-pixel coordinate mapping relationship corresponding to each camera.

[0090] The surface estimation module 300 is used to dynamically estimate the surface curvature based on the sub-pixel coordinate mapping relationship, construct a surface calibration model, and generate surface correction coefficients.

[0091] The brightness correction module 400 is used to correct the brightness range of the surface correction coefficient to obtain the dynamic surface correction coefficient.

[0092] The data stitching module 500 is used to perform brightness correction on grayscale images based on dynamic surface correction coefficients and to achieve seamless data stitching through brightness calibration between cameras.

[0093] Furthermore, the camera correction module of this invention is primarily responsible for performing flat field correction, dark current correction, and distortion correction on multiple cameras. In terms of hardware implementation, this module can be integrated into a dedicated camera control unit, which connects to multiple cameras and is capable of sending correction commands to the cameras and receiving data acquired by the cameras.

[0094] In terms of software algorithms, the flat field correction algorithm calculates the flat field response curve based on the image data of the uniform light source, the dark current correction algorithm is responsible for acquiring dark current images and marking bad points and denoising, and the distortion correction algorithm realizes the calculation of camera intrinsic and extrinsic parameters and distortion correction based on the checkerboard calibration method. Through the collaborative work of software and hardware, the camera after uniformity correction is obtained.

[0095] Furthermore, the positioning map processing module of the present invention is used to generate positioning maps and process positioning map data acquired by the camera. When generating the positioning map, based on the sub-pixel coordinate mapping technology of the encoded positioning map, a positioning map pattern that meets the requirements is generated using graphics generation software, and then the positioning map is displayed through a display device for the camera to capture.

[0096] After the camera acquires the positioning map, the computing unit in this module uses the corresponding algorithm to analyze the positioning map data, calculate the sub-pixel coordinate mapping relationship corresponding to each camera, obtain the sub-pixel coordinates Map1, Map2, ..., MapN of each camera partition, and store these mapping relationship data in the system's data storage unit for use by subsequent modules.

[0097] Furthermore, the surface estimation module includes an interval calculation unit, a deviation calculation unit, and a surface fitting unit. The interval calculation unit obtains the sub-pixel coordinate mapping relationship Map from the positioning map processing module, selects the pixel coordinates of the central region of the Map, and calculates the intervals stepH and stepV of the sub-pixels in the horizontal and vertical directions.

[0098] The deviation calculation unit constructs a standard equally spaced sub-pixel coordinate Map_std based on the results of the interval calculation unit, compares it with the actual sub-pixel coordinate Map, calculates the positional deviation D between the actual sub-pixel coordinates and the standard sub-pixel coordinates, and determines whether to calculate the deviation in the horizontal or vertical direction based on the bending direction of the curved surface.

[0099] The surface fitting unit receives the deviation D output by the deviation calculation unit, filters it, uses a suitable fitting algorithm to fit the surface, generates the surface correction coefficient coeff, and passes the coefficient to the subsequent brightness correction module.

[0100] Furthermore, the brightness correction module of the present invention includes a correction value calculation unit and a coefficient correction unit. The correction value calculation unit acquires grayscale image data, calculates the projection in the horizontal or vertical direction, and obtains the ratio of the minimum value to the center value after filtering the projection data as the correction value A.

[0101] The coefficient correction unit receives the surface correction coefficient coeff generated by the surface estimation module and the correction value A obtained by the correction value calculation unit. It then superimposes the correction value A onto the surface correction coefficient coeff to obtain a dynamic surface correction coefficient coeff' that adapts to the actual trend of brightness and darkness difference on the surface. Finally, it transmits this dynamic surface correction coefficient to the data stitching module.

[0102] Furthermore, the data stitching module of the present invention consists of an overlapping area processing unit and a successive stitching unit. The overlapping area processing unit selects overlapping areas of data collected by adjacent cameras according to the camera stitching direction and calculates the brightness ratio of the overlapping areas.

[0103] The successive stitching unit normalizes and corrects the brightness data of adjacent cameras based on the brightness ratio obtained from the overlapping area processing unit. Following the camera stitching order, the brightness data of adjacent cameras are corrected and stitched one after another until the fusion of all camera data is completed, achieving seamless stitching of multi-camera data and finally outputting complete stitched data.

[0104] This embodiment provides a dynamic data stitching system for Demura compensation of multi-camera curved screens, used to implement the aforementioned dynamic data stitching method for Demura compensation of multi-camera curved screens. Therefore, the specific implementation of the dynamic data stitching system for Demura compensation of multi-camera curved screens can be found in the previous section on the implementation of the dynamic data stitching method for Demura compensation of multi-camera curved screens. For example, the camera correction module 100, the positioning map processing module 200, the surface estimation module 300, the brightness correction module 400, and the data stitching module 500 are respectively used to implement steps S1, S2, S3, S4, and S5 in the aforementioned dynamic data stitching method for Demura compensation of multi-camera curved screens. Therefore, its specific implementation can be referred to the descriptions of the corresponding embodiments. To avoid redundancy, further details are omitted here.

[0105] To further illustrate the advantages of the technical solution of this invention, specific experiments are described below.

[0106] (I) Experimental Objectives

[0107] To verify the effectiveness of the multi-camera curved screen Demura compensation data dynamic stitching method and system of the present invention, a large-size, high-curvature vehicle OLED screen was used as the test object to evaluate the stitching accuracy and brightness uniformity.

[0108] (II) Test Equipment and Environment

[0109] A 45-inch ultra-wide curved OLED vehicle display was selected, with a curvature radius R≈1000mm. This display is representative and can well simulate the large-size, high-curvature screen situation in actual vehicle application scenarios.

[0110] Two 151M Veworks high-resolution industrial cameras, each with an aperture of F8, were used in a darkroom environment (background illuminance <1 lux) to minimize the interference of ambient light on the test results. A high-precision white exposure calibration plate was also used as an optical calibration tool to ensure the accuracy of camera calibration.

[0111] (III) Implementation Steps

[0112] First, camera calibration is performed, followed by flat field correction, dark current correction, and distortion correction. The operation steps are the same as those in the camera calibration section of the above method.

[0113] Next, location maps were collected and created. After that, two Veworks cameras were used to take pictures of the location maps. Then, the sub-pixel coordinate mapping relationship was calculated to obtain pixel maps Map1 and Map2.

[0114] Next, surface brightness estimation is performed. A 5×5 pixel area is selected at the center of each map, the horizontal and vertical intervals are calculated, standard sub-pixel coordinates are constructed, pixel deviations are calculated, and a surface model is fitted to generate the correction coefficient coeff.

[0115] Finally, brightness correction and stitching are performed. Correction parameter A is calculated, and coeff' is obtained by compensating for brightness differences. Brightness is corrected for each region, and brightness stitching is performed between cameras. By comparing the brightness data before and after stitching, it can be clearly seen that the stitched image has no stitching gaps, and brightness uniformity is significantly improved, effectively verifying the technical effect of the invention. In practical applications, quantitative analysis of indicators such as stitching accuracy and brightness uniformity further demonstrates the advantages of this invention in multi-camera curved screen Demura data stitching.

[0116] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0117] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0118] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The functions specified in one or more boxes. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0119] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for dynamic data stitching with Demura compensation on a multi-camera curved screen, characterized in that, Includes the following steps: Step S1: Perform flat field correction, dark current correction and distortion correction on multiple cameras to obtain the camera after uniformity correction; Step S2: Generate a positioning map and collect data using multiple calibrated cameras, then calculate the sub-pixel coordinate mapping relationship for each camera. Step S3: Based on the sub-pixel coordinate mapping relationship, dynamically estimate the curvature change of the curved screen, construct a curved surface calibration model, and generate curved surface correction coefficients; The process of dynamically estimating the curvature change of the curved screen, constructing a curved surface calibration model, and generating curved surface correction coefficients specifically includes: Select the pixel coordinates of the central region of the sub-pixel coordinate mapping relationship, and calculate the interval of the sub-pixels in the horizontal and vertical directions; Based on the aforementioned interval, standard equally spaced sub-pixel coordinates are constructed, and the deviation between the actual sub-pixel coordinates and the standard sub-pixel coordinates is calculated. After filtering the deviation, a surface is fitted, and the surface correction coefficient is obtained by normalizing it according to the center of the surface. Step S4: Correct the brightness range of the surface correction coefficient to obtain the corrected dynamic surface correction coefficient; The process of correcting the brightness range of the surface correction coefficient specifically includes: Acquire grayscale images and calculate the horizontal or vertical projections. After filtering, obtain the ratio of the minimum value to the center value as the correction value. The surface correction coefficients are superimposed and corrected based on the correction values ​​to obtain dynamic surface correction coefficients that adapt to the actual trend of brightness and darkness difference on the surface. Step S5: Based on the corrected dynamic surface correction coefficient, perform brightness correction on the grayscale images captured by multiple cameras, and achieve seamless data stitching through brightness calibration between cameras; The method of achieving seamless data stitching through brightness calibration between cameras specifically includes: Select the overlapping area of ​​adjacent cameras along the camera stitching direction and calculate the brightness ratio of the overlapping area; Based on the brightness ratio, the brightness data of adjacent cameras are normalized and corrected, and then stitched together sequentially until the fusion of all camera data is completed.

2. The method for dynamic data stitching for Demura compensation on multi-camera curved screens according to claim 1, characterized in that, The calculation of the deviation between the actual sub-pixel coordinates and the standard sub-pixel coordinates specifically involves: obtaining the positional deviation of the actual sub-pixel coordinates relative to the standard sub-pixel coordinates in the horizontal or vertical direction based on the bending direction of the curved surface.

3. The method for dynamic data stitching for Demura compensation on multi-camera curved screens according to claim 1, characterized in that, The generated positioning map is acquired by multiple calibrated cameras, and the sub-pixel coordinate mapping relationship of each camera is calculated. Specifically, based on the sub-pixel coordinate mapping technology of the coded positioning map, the sub-pixel coordinates Map1, Map2, ..., MapN of each camera partition are obtained, where N represents the number of cameras.

4. A dynamic data stitching system with Demura compensation for multi-camera curved screens, characterized in that, The system is used to implement the dynamic data stitching method for Demura compensation of multi-camera curved screens as described in any one of claims 1 to 3, specifically including: The camera correction module is used to perform flat field correction, dark current correction and distortion correction on multiple cameras to obtain a camera with uniformity correction. The localization map processing module is used to generate localization maps and acquire data through multiple calibrated cameras, and calculate the sub-pixel coordinate mapping relationship for each camera. The surface estimation module is used to dynamically estimate the surface curvature based on the sub-pixel coordinate mapping relationship, construct a surface calibration model, and generate surface correction coefficients. A brightness correction module is used to correct the brightness range of the surface correction coefficient to obtain a dynamic surface correction coefficient. The data stitching module is used to perform brightness correction on the grayscale image based on the dynamic surface correction coefficient, and to achieve seamless data stitching through brightness calibration between cameras.

5. The data dynamic stitching system for Demura compensation of multi-camera curved screens according to claim 4, characterized in that, The surface estimation module includes: The interval calculation unit is used to select the pixel coordinates of the central region of the sub-pixel coordinate mapping relationship and calculate the interval of the sub-pixels in the horizontal and vertical directions. The deviation calculation unit is used to construct standard equally spaced sub-pixel coordinates based on the interval, and to calculate the positional deviation between the actual sub-pixel coordinates and the standard sub-pixel coordinates; The surface fitting unit is used to filter the deviation and fit the surface to generate surface correction coefficients.

6. The data dynamic stitching system for Demura compensation of multi-camera curved screens according to claim 4, characterized in that, The brightness correction module includes: The correction value calculation unit is used to acquire grayscale images and calculate the horizontal or vertical projection, and obtain the ratio of the minimum value to the center value as the correction value. The coefficient correction unit is used to superimpose and correct the surface correction coefficients based on the correction value to obtain the dynamic surface correction coefficients.

7. The data dynamic stitching system for Demura compensation of multi-camera curved screens according to claim 4, characterized in that, The data splicing module includes: The overlapping area processing unit is used to select the overlapping area of ​​adjacent cameras along the camera stitching direction and calculate the brightness ratio of the overlapping area. The successive stitching unit is used to normalize and correct the brightness data of adjacent cameras based on the brightness ratio, and to complete the fusion of all camera data by successive stitching.

Images