A positioning method, system and related device for splicing sub-pixel groups of a display screen

Abstract

The application discloses a positioning method and system for a sub-pixel group of a spliced display screen and related devices, and is used for improving the accuracy of positioning of the sub-pixel group of the spliced display screen. The positioning method comprises the following steps: acquiring a target image in the spliced display screen; constructing a redundant positioning mechanism according to the sub-pixel group, and generating dense template points according to a preset multiple; preprocessing each sub-pixel group according to a Gaussian filtering algorithm; matching the preprocessed sub-pixel group with the dense template points to generate an initial coordinate set; screening the initial coordinate set and sorting the initial coordinate set according to column coordinates; detecting whether the distance between adjacent column coordinates in the sorted initial coordinate set is within a preset distance range; if not, removing the corresponding initial coordinate, inserting a dark point mark, detecting the next column coordinate, and finally generating a coordinate matrix; acquiring a standard row in the coordinate matrix, dynamically correcting the coordinate matrix through the standard row, and obtaining the positioning of the sub-pixel group of the spliced display screen according to the corrected coordinate matrix.

Description

Technical Field

[0001] This application relates to the field of display screen technology, and in particular to a method, system and related apparatus for positioning sub-pixel clusters of a splicing display screen. Background Technology

[0002] With the rapid development of display technology, video wall displays are increasingly widely used in large public spaces, commercial exhibitions, and professional fields. By splicing multiple single-screen units together, video wall displays can break the limitations of single-screen size, achieving ultra-large size and high-definition visual effects, meeting the diverse needs of scenarios such as stage performances, monitoring and command, and digital exhibition halls.

[0003] In existing technologies, positioning methods based on the regular arrangement characteristics of a single screen are commonly used. These methods primarily rely on the linear arrangement of pixels on a single screen and employ normalized cross-correlation (NCC) or gradient matching algorithms to achieve coordinate mapping. Specifically, applying this positioning method to a video wall display first requires calibrating the display to determine the linear arrangement pattern of its sub-pixel clusters. Then, based on the "single-screen linear expansion" logic within the positioning method, normalized cross-correlation (NCC) or gradient matching algorithms are used to map the coordinates of the sub-pixel clusters, thereby completing the positioning of the sub-pixel clusters.

[0004] However, positioning methods based on the regular arrangement characteristics of a single screen cannot handle the non-linear offsets in the row and column directions caused by pose deviations between spliced ​​displays. When there are pose deviations or gaps in the spliced ​​displays, the offset of sub-pixel clusters in the row and column directions is not linear. Positioning methods based on the regular arrangement characteristics of a single screen still use the linear expansion logic of a single screen for coordinate mapping. This misjudges the non-linear offset as a normal linear offset and continues to perform subsequent coordinate mapping according to linear rules, ultimately leading to an overall misalignment of the coordinate matrix. This severely affects the accuracy of sub-pixel cluster positioning on the spliced ​​display and cannot meet the application requirements of high-precision display scenarios. Summary of the Invention

[0005] To address the aforementioned technical problems, this application provides a method, system, and related apparatus for locating sub-pixel clusters in a splicing display screen.

[0006] The technical solution provided in this application is described below:

[0007] The first aspect of this application provides a method for locating sub-pixel clusters in a spliced ​​display screen, the method comprising:

[0008] Acquire a target image from a splicing display screen, the target image comprising several sub-pixel clusters;

[0009] A redundant positioning mechanism is constructed based on the total amount of the sub-pixel clusters, and dense template points are generated according to the redundant positioning mechanism. The generation ratio of the redundant positioning mechanism is a preset multiple of the total amount.

[0010] Each sub-pixel cluster is preprocessed according to a Gaussian filtering algorithm, the preprocessing including convolutional smoothing and noise reduction;

[0011] The preprocessed sub-pixel clusters are matched with the dense template points to generate an initial coordinate set;

[0012] The initial coordinate set is filtered by a preset shake coefficient and a preset shooting magnification, and the filtered initial coordinate set is sorted according to column coordinates;

[0013] Check whether the spacing between adjacent column coordinates in the sorted initial coordinate set is within a preset spacing range;

[0014] If not, remove the initial coordinates corresponding to the column coordinates, insert dark spot markers at the empty coordinate positions after removal, and perform the detection step of the next adjacent column coordinates until all column coordinates are detected, and generate a coordinate matrix for row clustering;

[0015] Obtain the standard rows in the coordinate matrix, and dynamically correct the coordinate matrix using the standard rows and a preset column spacing threshold. Based on the corrected coordinate matrix, obtain the positioning of the sub-pixel clusters of the splicing display screen.

[0016] Optionally, the step of filtering the initial coordinate set using a preset shake coefficient and a preset shooting magnification, and sorting the filtered initial coordinate set according to column coordinates, includes:

[0017] Sort the initial coordinates in the initial coordinate set in ascending order of row coordinates, and take the row coordinate of the first initial coordinate after sorting as the reference value;

[0018] The effective line range is calculated based on the aforementioned benchmark value and in combination with the preset shake coefficient and preset shooting magnification.

[0019] The row coordinates in the initial coordinate set are filtered by the effective row range, and the filtered initial coordinate set is sorted in ascending order of column coordinates;

[0020] The effective row range is represented by the following formula:

[0021] ;

[0022] in, This represents the reference value. Indicates the range of valid rows. This indicates the preset shooting magnification. This represents the preset jitter coefficient.

[0023] Optionally, obtaining the standard rows in the coordinate matrix includes:

[0024] Starting from the first row of the coordinate matrix, traverse row by row. If all coordinates in a row of the coordinate matrix are unmarked and the number of coordinates in that row is consistent with the preset theoretical number, then that row is taken as the standard row.

[0025] If each row of coordinates has a dark spot mark or no dark spot mark but the number of coordinates is inconsistent with the preset theoretical number, then the coordinate matrix of the row clustering is modified to the coordinate matrix of the column clustering, or the coordinate pattern of the column coordinates in the coordinate matrix is ​​used to generate the standard row.

[0026] Optionally, the step of matching the preprocessed sub-pixel clusters with the dense template points to generate an initial coordinate set includes:

[0027] The dense template points are overlaid on the preprocessed sub-pixel clusters for localization and matching to obtain the coordinates of each sub-pixel cluster;

[0028] The coordinates are cleaned up, and an initial coordinate set is generated based on the cleaned coordinates;

[0029] The process of cleaning up the coordinates includes:

[0030] When multiple dense template points are located and matched to the same sub-pixel cluster, resulting in multiple duplicate coordinates, the duplicate coordinates are cleared.

[0031] When the pixel value within the range matched by the dense template point is less than the first pixel threshold, the coordinates of the corresponding dense template point are cleared.

[0032] When the dense template point is located at the center of the sub-pixel cluster, the pixel value of the sub-pixel cluster is obtained, and the coordinates of the sub-pixel cluster are cleaned or retained based on the pixel value.

[0033] Optionally, after preprocessing each sub-pixel cluster according to the Gaussian filtering algorithm, the method further includes:

[0034] Obtain the pixel value of each sub-pixel cluster after preprocessing, and determine whether each pixel value is greater than the second pixel threshold in the preset triggering mechanism;

[0035] If so, then the step of matching the sub-pixel cluster with the dense template point is performed.

[0036] Optionally, after dynamically correcting the coordinate matrix using the standard rows and a preset column spacing threshold, the method further includes:

[0037] The first row in the coordinate matrix that has undergone dynamic correction is taken as the new standard row;

[0038] The new standard row is used in conjunction with a preset column spacing threshold to dynamically correct the next row in the coordinate matrix that needs to be dynamically corrected. When the row is dynamically corrected, it is used as the new standard row. This process is repeated until all rows in the coordinate matrix have been dynamically corrected.

[0039] Optionally, the preset multiple is 1.5 to 2 times the total amount.

[0040] A second aspect of this application provides a positioning system for sub-pixel clusters of a splicing display screen, the positioning system comprising:

[0041] An acquisition unit is used to acquire a target image in a splicing display screen, the target image comprising a plurality of sub-pixel clusters;

[0042] The first generation unit is used to construct a redundant positioning mechanism based on the total amount of the sub-pixel clusters, and generate dense template points according to the redundant positioning mechanism, wherein the generation ratio of the redundant positioning mechanism is a preset multiple of the total amount.

[0043] The processing unit is configured to preprocess each sub-pixel cluster according to a Gaussian filtering algorithm, the preprocessing including convolutional smoothing and noise reduction;

[0044] The second generation unit is used to match the preprocessed sub-pixel clusters with the dense template points to generate an initial coordinate set;

[0045] The sorting unit is used to filter the initial coordinate set by a preset shake coefficient and a preset shooting magnification, and sort the filtered initial coordinate set according to column coordinates;

[0046] The detection unit is used to detect whether the spacing between adjacent column coordinates in the sorted initial coordinate set is within a preset spacing range;

[0047] The third generation unit is used to remove the initial coordinates corresponding to the column coordinates if no, insert dark spot markers at the empty coordinate positions after removal, and perform the detection step of the next adjacent column coordinates until all column coordinates are detected, and generate a coordinate matrix for row clustering.

[0048] The correction unit is used to obtain the standard row in the coordinate matrix, and dynamically correct the coordinate matrix by using the standard row and a preset column spacing threshold, and obtain the positioning of the sub-pixel cluster of the splicing display screen according to the corrected coordinate matrix.

[0049] Optionally, the sorting unit is specifically used for:

[0050] Sort the initial coordinates in the initial coordinate set in ascending order of row coordinates, and take the row coordinate of the first initial coordinate after sorting as the reference value;

[0051] The effective line range is calculated based on the aforementioned benchmark value and in combination with the preset shake coefficient and preset shooting magnification.

[0052] The row coordinates in the initial coordinate set are filtered by the effective row range, and the filtered initial coordinate set is sorted in ascending order of column coordinates;

[0053] The effective row range is represented by the following formula:

[0054] ;

[0055] in, This represents the reference value. Indicates the range of valid rows. This indicates the preset shooting magnification. This represents the preset jitter coefficient.

[0056] Optionally, the correction unit is specifically used for:

[0057] Starting from the first row of the coordinate matrix, traverse row by row. If all coordinates in a row of the coordinate matrix are unmarked and the number of coordinates in that row is consistent with the preset theoretical number, then that row is taken as the standard row.

[0058] If each row of coordinates has a dark spot mark or no dark spot mark but the number of coordinates is inconsistent with the preset theoretical number, then the coordinate matrix of the row clustering is modified to the coordinate matrix of the column clustering, or the coordinate pattern of the column coordinates in the coordinate matrix is ​​used to generate the standard row.

[0059] Optionally, the second generating unit is specifically used for:

[0060] The dense template points are overlaid on the preprocessed sub-pixel clusters for localization and matching to obtain the coordinates of each sub-pixel cluster;

[0061] The coordinates are cleaned up, and an initial coordinate set is generated based on the cleaned coordinates;

[0062] The process of cleaning up the coordinates includes:

[0063] When multiple dense template points are located and matched to the same sub-pixel cluster, resulting in multiple duplicate coordinates, the duplicate coordinates are cleared.

[0064] When the pixel value within the range matched by the dense template point is less than the first pixel threshold, the coordinates of the corresponding dense template point are cleared.

[0065] When the dense template point is located at the center of the sub-pixel cluster, the pixel value of the sub-pixel cluster is obtained, and the coordinates of the sub-pixel cluster are cleaned or retained based on the pixel value.

[0066] Optionally, it also includes a judgment unit, specifically used for

[0067] Obtain the pixel value of each sub-pixel cluster after preprocessing, and determine whether each pixel value is greater than the second pixel threshold in the preset triggering mechanism;

[0068] If so, then the step of matching the sub-pixel cluster with the dense template point is performed.

[0069] Optionally, an update unit is also included, specifically for:

[0070] The first row in the coordinate matrix that has undergone dynamic correction is taken as the new standard row;

[0071] The new standard row is used in conjunction with a preset column spacing threshold to dynamically correct the next row in the coordinate matrix that needs to be dynamically corrected. When the row is dynamically corrected, it is used as the new standard row. This process is repeated until all rows in the coordinate matrix have been dynamically corrected.

[0072] Optionally, the preset multiple is 1.5 to 2 times the total amount.

[0073] A third aspect of this application provides a positioning device for sub-pixel clusters of a splicing display screen, the positioning device comprising:

[0074] Processor, memory, input / output units, and bus;

[0075] The processor is connected to the memory, the input / output unit, and the bus;

[0076] The memory stores a program, which the processor invokes to execute the first aspect and any one of the optional positioning methods in the first aspect.

[0077] A fourth aspect of this application provides a computer-readable storage medium storing a program that, when executed on a computer, performs the positioning method of the first aspect and any one of the first aspects.

[0078] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0079] This application achieves high-precision positioning of sub-pixel clusters in a spliced ​​display screen by constructing a redundant positioning mechanism and combining it with a dynamic correction strategy. First, dense template points are generated according to a preset multiple in the redundant positioning mechanism based on the total number of sub-pixel clusters. The excess of dense template points expands the search range and frequency for sub-pixel clusters, providing a more solid positioning foundation. Then, a Gaussian filtering algorithm is used to perform convolutional smoothing and noise reduction preprocessing on each sub-pixel cluster, which enhances the central features of the sub-pixel clusters, suppresses noise interference, and improves the clarity and reliability of the sub-pixel cluster features. Next, the preprocessed sub-pixel clusters are matched with the dense template points to generate an initial coordinate set. Then, the initial coordinate set is filtered by a preset jitter coefficient and a preset shooting magnification, and sorted by column coordinates. After sorting, the distance between adjacent column coordinates in the initial coordinate set is checked to see if it is within a preset distance range. If not, the corresponding initial coordinates are removed and dark spot markers are inserted, thereby generating a row-oriented clustering coordinate matrix. This accurately identifies and eliminates coordinates with nonlinear offsets caused by pose deviations or splicing gaps, restoring the physical arrangement of the sub-pixel clusters. Finally, the standard rows in the coordinate matrix are obtained, and the coordinate matrix is ​​dynamically corrected based on a preset column spacing threshold to obtain accurate sub-pixel cluster positioning. The entire positioning process does not rely on the linear arrangement of a single screen, avoiding the overall misalignment of the coordinate matrix caused by pose deviations or splicing gaps when positioning methods based on the regular arrangement characteristics of a single screen are applied to a spliced ​​display screen. This improves the positioning accuracy and reliability of the spliced ​​display screen in complex scenarios and meets the application requirements of high-precision display scenarios. Attached Figure Description

[0080] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0081] Figure 1 A schematic flowchart of an embodiment of the method for locating sub-pixel clusters of a splicing display screen provided in this application;

[0082] Figure 2 A schematic flowchart of another embodiment of the method for locating sub-pixel clusters of a splicing display screen provided in this application;

[0083] Figure 3 A schematic diagram of an embodiment of the positioning system for sub-pixel clusters of a splicing display screen provided in this application;

[0084] Figure 4 A schematic diagram of an embodiment of the positioning device for sub-pixel clusters of a splicing display screen provided in this application;

[0085] Figure 5 A schematic diagram of dense template points generated in the positioning of sub-pixel clusters of the splicing display screen provided in this application;

[0086] Figure 6 A schematic diagram illustrating the matching of sub-pixel clusters with dense template points in the positioning of sub-pixel clusters of the splicing display screen provided in this application;

[0087] Figure 7 A schematic diagram illustrating the dark dot markings in the positioning of sub-pixel clusters of the splicing display screen provided in this application;

[0088] Figure 8 A schematic diagram illustrating a case where one of the sub-pixel clusters of the splicing display provided in this application is abnormally lit.

[0089] Figure 9 A schematic diagram illustrating another case of abnormal lighting in the positioning of sub-pixel clusters of the splicing display screen provided in this application;

[0090] Figure 10 This is a schematic diagram illustrating the dynamic correction of the coordinate matrix using standard rows in the positioning of sub-pixel clusters of the splicing display screen provided in this application. Detailed Implementation

[0091] This application provides a method for locating sub-pixel clusters in a video wall display, which can improve the accuracy of sub-pixel cluster positioning. It should be noted that the sub-pixel cluster positioning method of this application is applied to a terminal device.

[0092] It should be noted that the positioning method, apparatus, and related devices for sub-pixel clusters of a splicing display screen provided in this application can be applied to terminals, systems, and servers. For example, a terminal can be a smartphone, computer, tablet, smart TV, smartwatch, portable computer terminal, or a desktop computer or other fixed terminal. For ease of explanation, this application uses a terminal as the implementing entity for illustrative purposes.

[0093] Please see Figure 1 and combined Figure 5 , Figure 6 , Figure 7 as well as Figure 10 This application first provides an embodiment of a method for locating sub-pixel clusters in a spliced ​​display screen, the embodiment including:

[0094] S101. Obtain the target image in the splicing display screen. The target image includes several sub-pixel clusters.

[0095] In this embodiment, acquiring the target image of the splicing display screen involves using image acquisition devices such as industrial cameras or high-definition cameras to capture images of the splicing display screen. Specifically, optical imaging technology is used to convert the physical pixels of the splicing display screen into a digital image (i.e., the target image) comprising several sub-pixel clusters, where each sub-pixel cluster is represented in the digital image as a pixel area with specific brightness and shape. Due to the non-uniform seams and pose deviations of the splicing display screen, the target image needs to cover the entire splicing area to preserve the spatial positional relationship of the sub-pixel clusters. Acquiring a target image containing sub-pixel clusters is the foundation for the positioning method. A high-quality target image can clearly present the boundary and center features of the sub-pixel clusters, avoiding positioning deviations caused by image blurring or noise.

[0096] It is important to note that when using industrial cameras, high-definition cameras, or other image acquisition devices to photograph the video wall display, it is crucial to ensure that the image resolution matches the actual pixel pitch of the display to fully capture the distribution of sub-pixel clusters. Furthermore, due to the non-uniform seams and pose deviations of the video wall display, the acquired target image must cover the entire splicing area to preserve the spatial relationships of sub-pixel clusters, providing a data foundation for subsequent positioning. If the target image fails to cover the gaps or edges of the video wall display, key sub-pixel clusters will be missed during dense template point matching, leading to accumulated errors.

[0097] S102. Construct a redundant positioning mechanism based on the total amount of sub-pixel clusters, and generate dense template points according to the redundant positioning mechanism. The generation ratio of the redundant positioning mechanism is a preset multiple of the total amount.

[0098] In this embodiment, the total number of sub-pixel clusters in the target image is first counted. Then, dense template points are generated according to the generation ratio in the redundancy positioning mechanism, making the number of dense template points significantly greater than the actual number of sub-pixel clusters. The generation ratio is a preset multiple of the total number of sub-pixel clusters. For example, if the total number of sub-pixel clusters is 1, then the number of dense template points generated according to the generation ratio in the redundancy positioning mechanism is 1*N, where N is a preset multiple. The redundancy positioning mechanism overcomes the limitations of traditional 1:1 templates, transforming the positioning method from "relying on a single match" to "based on redundancy filtering," significantly improving the positioning reliability in complex spliced ​​display scenarios.

[0099] Furthermore, generating dense template points that are a preset multiple of the total number of sub-pixel clusters is to create redundant positioning capabilities during subsequent matching. Even if some dense template points cannot be matched due to splicing errors, other dense template points can still accurately match, avoiding missed detections. Moreover, different dense template points can perform overlapping matching on the same sub-pixel cluster. For example, when there is irregular offset in the spliced ​​display screen, dense template point 1 and dense template point 2 can simultaneously lock the same sub-pixel cluster, thus preserving the most accurate matching result.

[0100] S103. Preprocess each sub-pixel cluster according to the Gaussian filtering algorithm. The preprocessing includes convolutional smoothing and noise reduction.

[0101] In this embodiment, a Gaussian filtering algorithm is used to preprocess each sub-pixel cluster. The preprocessing includes convolutional smoothing and noise reduction. The Gaussian filtering algorithm mainly constructs a two-dimensional Gaussian kernel function and utilizes the bell-shaped distribution characteristic of the two-dimensional Gaussian kernel function to perform a weighted average on each sub-pixel cluster and its neighborhood, attenuating high-frequency noise components while preserving the center brightness characteristics of the sub-pixel cluster. For example, for noise points with abrupt brightness changes, the Gaussian filtering algorithm assigns a smaller weight based on the distance between the noise point and its neighboring pixels, thereby reducing the impact of noise and making the edges of the sub-pixel cluster smoother and the center more prominent.

[0102] It is important to note that the standard deviation of the 2D Gaussian kernel function can be adjusted according to the expected size of the sub-pixel clusters to balance smoothing effects with detail preservation.

[0103] Preprocessing using the Gaussian filtering algorithm can effectively remove salt-and-pepper noise and Gaussian noise generated by light interference or sensor errors during the shooting process of sub-pixel clusters, preventing noise points from being misjudged as the center of the sub-pixel clusters. It can also weaken the irregular fluctuations at the edges of the sub-pixel clusters and highlight the brightness peaks of the sub-pixel clusters, making it easier to capture the true center position of the sub-pixel clusters during subsequent dense template point matching.

[0104] S104. Match the preprocessed sub-pixel clusters with dense template points to generate an initial coordinate set;

[0105] In this embodiment, during the matching process, it is necessary to first calculate the spatial distance between each dense template point and the surrounding sub-pixel clusters, and then match the dense template points to the nearest or most relevant sub-pixel clusters based on the spatial distance, and record the coordinates of each sub-pixel cluster after matching.

[0106] This ensures that each sub-pixel cluster is matched by at least one dense template point, avoiding coordinate loss due to stitching offset and positioning interruption due to single matching failure in traditional methods, thus providing a basis for the final generation of the coordinate matrix.

[0107] S105. Filter the initial coordinate set by preset shake coefficient and preset shooting magnification, and sort the filtered initial coordinate set according to column coordinates;

[0108] In this embodiment, a preset jitter coefficient and a preset shooting magnification are determined based on the physical characteristics of the splicing display screen. The preset jitter coefficient is a coordinate jitter compensation coefficient, the value of which is related to the precision of the splicing display screen, with a typical range of 0.05 to 0.15. The preset shooting magnification is a multiple of the shooting resolution, which determines the range of subsequent line coordinate determination.

[0109] After determining the preset shake factor and preset shooting magnification, these two factors are combined to filter coordinates from the initial coordinate set that meet the criteria. Subsequently, the filtered coordinates are arranged by column coordinates, ensuring that coordinates within the same row are arranged according to their physical location. This coordinate filtering and sorting based on the preset shake factor and preset shooting magnification significantly improves the reliability and orderliness of the initial coordinate set, reducing interference in subsequent calculations and providing an ordered data structure for subsequent detection.

[0110] S106. Check whether the spacing between adjacent column coordinates in the sorted initial coordinate set is within the preset spacing range;

[0111] In this embodiment, after sorting the filtered initial coordinate set according to column coordinates, the sorted initial coordinate set is traversed in column coordinate order, and the column spacing of each pair of adjacent coordinates in the initial coordinate set is calculated, i.e., the absolute value of the column coordinate difference. Subsequently, a preset spacing range is determined based on the physical pixel spacing and shooting magnification of the splicing screen. The typical value of the preset spacing range is 0.1~0.3. Finally, the column spacing of each pair of adjacent coordinates in the initial coordinate set is detected. The detection content is whether the column spacing of each pair of adjacent coordinates is within the preset spacing range. For example, when the preset spacing range is 0.1~0.3, if a pair of adjacent coordinates is detected with a column spacing of 0.5, and 0.5 is not within the range of 0.1~0.3, then it is determined that the spacing of adjacent column coordinates in the initial coordinate set is not within the preset spacing range; when the preset spacing range is 0.1~0.3, if a pair of adjacent coordinates is detected with a column spacing of 0.2, and 0.2 is within the range of 0.1~0.3, then it is determined that the spacing of adjacent column coordinates in the initial coordinate set is within the preset spacing range.

[0112] If the spacing between adjacent column coordinates in the initial coordinate set is not within the preset spacing range, then step S107 is executed; if the spacing between adjacent column coordinates in the initial coordinate set is within the preset spacing range, then the coordinate matrix of row clustering is directly generated.

[0113] S107. Remove the initial coordinates corresponding to the column coordinates, insert dark spot markers at the empty coordinate positions after removal, and perform the detection step of the next adjacent column coordinates until all column coordinates are detected, and generate the coordinate matrix of row clustering.

[0114] In this embodiment, when it is determined that the spacing between adjacent column coordinates in the initial coordinate set is not within the preset spacing range, the initial coordinates of the corresponding column coordinates are removed, and a dark spot marker is inserted in the empty position after removal. Then, the spacing between the next pair of adjacent column coordinates is detected until all column coordinates are detected, generating a coordinate matrix for row clustering. Specifically, when the spacing between currently detected adjacent column coordinates is not within the preset spacing range, there are unlit sub-pixel clusters within the spacing range of the currently adjacent column coordinates. It is necessary to remove the initial coordinates corresponding to these column coordinates with excessively large or small spacing, and insert a dark spot marker in the corresponding empty position after removal, for example, using special markers such as "null" or "dark" to represent dark spots.

[0115] After completing the spacing detection of adjacent column coordinates in a single row, the above process is repeated for all rows, eventually forming a coordinate matrix of row clusters containing dark spot markers. This coordinate matrix can reflect the true state of the display screen.

[0116] S108. Obtain the standard row in the coordinate matrix, and dynamically correct the coordinate matrix by using the standard row and a preset column spacing threshold. Based on the corrected coordinate matrix, obtain the positioning of the sub-pixel clusters of the splicing display screen.

[0117] In practical applications, this embodiment first requires obtaining a standard row from the coordinate matrix of row-oriented clustering. This standard row represents the coordinates of a theoretical row of pixel clusters, providing a unified reference for the coordinates of pixel clusters in the coordinate matrix. Subsequently, based on the standard row, expansion is performed in both row and column directions, forming an expansion region by combining a preset dithering coefficient. This aligns the coordinates in the coordinate matrix with the coordinates of the standard row, and the coordinates are adjusted in the column direction according to a preset column spacing range. The positions of dark spot markers in the coordinate matrix are filled with the coordinates from the standard row, thereby achieving dynamic correction. Finally, the positioning of sub-pixel clusters on the spliced ​​display screen can be obtained based on the dynamically corrected coordinate matrix.

[0118] By constructing a redundant positioning mechanism and combining it with a dynamic correction strategy, high-precision positioning of sub-pixel clusters in a spliced ​​display screen is achieved. First, dense template points are generated according to a preset multiple in the redundant positioning mechanism based on the total number of sub-pixel clusters. The excess of dense template points expands the search range and frequency for sub-pixel clusters, providing a more solid positioning foundation. Then, a Gaussian filtering algorithm is used to perform convolutional smoothing and noise reduction preprocessing on each sub-pixel cluster, which enhances the central features of the sub-pixel clusters, suppresses noise interference, and improves the clarity and reliability of the sub-pixel cluster features. Next, the preprocessed sub-pixel clusters are matched with the dense template points to generate an initial coordinate set. Then, the initial coordinate set is filtered by a preset jitter coefficient and a preset shooting magnification, and sorted by column coordinates. After sorting, the spacing between adjacent column coordinates in the initial coordinate set is checked to see if it is within a preset spacing range. If not, the corresponding initial coordinates are removed and dark spot markers are inserted, thus generating a row-oriented clustering coordinate matrix. This accurately identifies and eliminates coordinates with nonlinear offsets caused by pose deviations or splicing gaps, restoring the physical arrangement of the sub-pixel clusters. Finally, the standard rows in the coordinate matrix are obtained, and the coordinate matrix is ​​dynamically corrected based on a preset column spacing threshold to obtain accurate sub-pixel cluster positioning. The entire positioning process does not rely on the linear arrangement of a single screen, avoiding the overall misalignment of the coordinate matrix caused by pose deviations or splicing gaps when positioning methods based on the regular arrangement characteristics of a single screen are applied to a spliced ​​display screen. This improves the positioning accuracy and reliability of the spliced ​​display screen in complex scenarios and meets the application requirements of high-precision display scenarios.

[0119] Please see Figure 2 and combined Figure 5 , Figure 6 , Figure 7 , Figure 8 , Figure 9 as well as Figure 10 This application also provides another embodiment of a method for locating sub-pixel clusters in a spliced ​​display screen, the embodiment including:

[0120] S201. Obtain the target image in the splicing display screen. The target image includes several sub-pixel clusters.

[0121] S202. Construct a redundant positioning mechanism based on the total number of sub-pixel clusters, and generate dense template points according to the redundant positioning mechanism. The generation ratio of the redundant positioning mechanism is a preset multiple of the total number.

[0122] In an optional embodiment, the preset multiple is 1.5 to 2 times the total amount;

[0123] By generating 1.5 to 2 times the total number of dense template points through a redundant positioning mechanism, and allowing multiple dense template points to correspond to the same sub-pixel cluster, this "many-to-one" matching mode improves positioning fault tolerance. For example, when there are irregular offsets in the spliced ​​display screen, the same sub-pixel cluster can be covered by multiple dense template points. Using a preset multiple of 1.5 to 2 times the total number of dense template points not only avoids computational redundancy caused by an excessive number of dense template points, but also effectively covers irregular offset scenarios of the spliced ​​display screen, significantly improving the positioning reliability and anti-offset capability in spliced ​​display screen scenarios.

[0124] S203. Preprocess each sub-pixel cluster according to the Gaussian filtering algorithm. The preprocessing includes convolutional smoothing and noise reduction.

[0125] In this embodiment, steps S201 to S203 are similar to steps S101 to S103 in the previous embodiment, and will not be described again here.

[0126] S204. Obtain the pixel value of each sub-pixel cluster after preprocessing, and determine whether each pixel value is greater than the second pixel threshold in the preset triggering mechanism.

[0127] In this embodiment, after preprocessing, the pixel range of each sub-pixel cluster can be determined through image segmentation and region labeling. All sub-pixels within the pixel range are traversed, and the grayscale or RGB values ​​of all sub-pixels are obtained as pixel values. Subsequently, it is determined whether each pixel value is greater than the second pixel threshold in the preset triggering mechanism.

[0128] The second pixel threshold in the preset triggering mechanism is set based on the normal luminance and noise level of the splicing display screen. It is generally set to 2 to 3 times the noise peak value to ensure that subsequent operations are triggered only when the overall pixel value of the sub-pixel cluster is higher than the noise level.

[0129] If all pixel values ​​are greater than the second pixel threshold, then proceed to step S205; if one or more pixel values ​​are less than or equal to the second pixel threshold, then the subsequent dense template point matching operation will not be triggered to avoid invalid matching of non-luminous areas and reduce the waste of computing resources.

[0130] S205. Overlay the dense template points onto the preprocessed sub-pixel clusters for localization and matching to obtain the coordinates of each sub-pixel cluster;

[0131] In this embodiment, when it is determined that the value of each pixel is greater than the second pixel threshold in the preset triggering mechanism, the step of covering the preprocessed sub-pixel clusters with dense template points for localization and matching is performed to obtain the coordinates of each sub-pixel cluster. This step mainly uses the dense template points as the center to search for the brightness center or shape center within the sub-pixel cluster region, matches the dense template points to the center position of the nearest or most similar sub-pixel cluster, and records the coordinates of the current dense template points as the coordinates of the sub-pixel clusters. If multiple dense template points match the same sub-pixel cluster, all coordinates of the matching sub-pixel clusters are retained; if there are no valid sub-pixel clusters within the search range of the dense template points, the coordinates of the current dense template points are retained.

[0132] The entire positioning and matching process mainly utilizes the redundancy of dense template points and uses a "many-to-one" matching mode to ensure that each sub-pixel cluster is accurately matched by at least one dense template point, thereby improving the positioning coverage and accuracy of spliced ​​displays in complex scenarios.

[0133] S206. Clean up the coordinates and generate an initial coordinate set based on the cleaned coordinates;

[0134] In this embodiment, the coordinates obtained from the positioning and matching are cleaned up, and an initial coordinate set is generated based on the cleaned coordinates. The initial coordinate set generated after cleaning not only reflects the actual distribution of sub-pixel clusters, but also meets the accuracy requirements of subsequent operations.

[0135] Specifically, cleaning up the coordinates obtained from location matching includes:

[0136] (1) When multiple dense template points are matched to the same sub-pixel cluster and multiple duplicate coordinates are obtained, the duplicate coordinates are cleaned up.

[0137] First, iterate through all the coordinates generated by the matching of dense template points to determine if there are multiple coordinates pointing to the center of the same sub-pixel cluster. When it is determined that multiple dense template points are located and matched to the same sub-pixel cluster, resulting in multiple duplicate coordinates, retain one coordinate and delete the others. This avoids redundant data interfering with subsequent calculations. For example, when there is a slight offset in the splicing screen, multiple dense template points may simultaneously lock onto the same sub-pixel cluster, generating 6 identical (12, 9) coordinates. In this case, only one (12, 9) coordinate is retained, and coordinates identical to (12, 9) are deleted. This ensures that each sub-pixel cluster corresponds to only one valid coordinate, improving data simplicity and positioning accuracy.

[0138] (2) When the pixel value within the range matched by the dense template point is less than the first pixel threshold, the coordinates of the corresponding dense template point are cleared.

[0139] First, determine the matching range of the dense template points. Then, iterate through all pixels within the matching range of each dense template point to obtain all pixel values ​​within that range. Finally, compare these pixel values ​​with a first pixel threshold and perform cleanup based on the comparison result. If the pixel value is less than the first pixel threshold, clean up the coordinates of the corresponding dense template point; if the pixel value is greater than or equal to the first pixel threshold, retain the coordinates of the corresponding dense template point.

[0140] Furthermore, the first pixel threshold is set based on the image noise level and is mainly used to distinguish between effective luminous areas and noise. By comparing the first pixel threshold, the validity of coordinates within the dense template point matching range is determined, avoiding the inclusion of coordinates of non-luminous areas in the initial coordinate set and filtering out invalid data.

[0141] (3) When the dense template point is located at the center of the sub-pixel cluster, the pixel value of the sub-pixel cluster is obtained, and the coordinates of the sub-pixel cluster are cleaned or retained according to the pixel value.

[0142] The location of a dense template point at the center of a sub-pixel cluster is determined by the spatial relationship between the actual position of the dense template point and the outline of the sub-pixel cluster. When the dense template point is located at the center of the sub-pixel cluster, the pixel value of the sub-pixel cluster is obtained, and the coordinates of the sub-pixel cluster are selectively cleaned or retained based on the pixel value. This solves the coordinate verification problem when the preset triggering mechanism is not activated due to the uniform illumination of the sub-pixel cluster. For example, when the brightness of the sub-pixel cluster is consistent, the dense template point may directly locate to the center of the sub-pixel cluster without performing positioning matching. In this case, the pixel value of the sub-pixel cluster can be used to determine whether the sub-pixel cluster is a valid emitting point, preventing invalid coordinates from being mistakenly retained and valid coordinates from being mistakenly deleted.

[0143] S207. Sort the initial coordinates in the initial coordinate set in ascending order of row coordinates, and take the row coordinate of the first initial coordinate after sorting as the reference value.

[0144] In this embodiment, after the coordinate cleanup is completed and an initial coordinate set is generated, the initial coordinates in the initial coordinate set are sorted in ascending order by row coordinates. The sorting rule must be consistent with the actual layout direction of the splicing display screen. After sorting, the first initial coordinate in the ascending row coordinate sort is selected, and its row coordinate is extracted as the reference value. The reference value represents the theoretical row position of the current row and is used for the calculation of the subsequent effective row range.

[0145] Using the row coordinate of the first initial coordinate as the reference value is based on the regular distribution pattern of sub-pixel clusters in the row and column directions of the splicing display screen. For example, if the row coordinate deviation of sub-pixel clusters in the same row is small, the row coordinate of the first coordinate can be used as the representative position of that row.

[0146] S208. The effective line range is calculated based on the benchmark value and in combination with the preset shake coefficient and preset shooting magnification.

[0147] In this embodiment, the effective line range calculated based on the benchmark value and in combination with the preset jitter coefficient and preset shooting magnification can not only accommodate normal line direction fluctuations, such as splicing tilt and shooting angle errors, but also remove outlier line coordinates caused by abnormalities or mismatches.

[0148] In practical applications, the baseline value is used as the starting point of the effective line range. Then, the two ending points of the effective line range are determined by combining the preset jitter coefficient and the preset shooting magnification, forming a range interval. This range interval is the effective line range that encompasses normal line coordinate fluctuations.

[0149] Specifically, the effective row range is represented by the following formula:

[0150] ;

[0151] in, Indicates the baseline value. Indicates the range of valid rows. Indicates the preset shooting magnification. This indicates the preset jitter coefficient.

[0152] This method of calculating the effective row range avoids the misjudgment problem caused by the fixed row range in traditional methods. It can dynamically adjust the row range according to the actual offset of the splicing display screen, ensuring that the coordinates of sub-pixel clusters in the same row are accurately classified. Especially when there are multiple splicing displays, it can effectively reduce cross-row and column matching errors and improve the physical accuracy of the positioning results.

[0153] S209. Filter the row coordinates in the initial coordinate set by the effective row range, and sort the filtered initial coordinate set in ascending order of column coordinates.

[0154] In this embodiment, row clustering is mainly achieved by filtering out initial coordinates that are not within the effective row range, and then the initial coordinates are ordered by sorting the column coordinates in ascending order, so as to provide structured data for subsequent detection of the distance between adjacent columns.

[0155] Specifically, it is necessary to first traverse all initial coordinates in the initial coordinate set and determine whether the row coordinate of each initial coordinate is within the valid row range. If the row coordinate is not within the valid row range, the initial coordinate corresponding to the current row coordinate is removed; if the row coordinate is within the valid row range, the initial coordinate corresponding to the current row coordinate is retained. For example, if the initial coordinate set has initial coordinates A, B, and C, where the row coordinate of initial coordinate A is 95, the row coordinate of initial coordinate B is 105, and the row coordinate of initial coordinate C is 120, the calculated valid row range is [100, 118]. At this time, 95 is not within the range of [100, 118], 105 is within the range of [100, 118], and 120 exceeds the range of [100, 118]. Therefore, the initial coordinates A and C corresponding to row coordinates 95 and 120 are removed, and the initial coordinate B corresponding to row coordinate 105 is retained. The entire filtering process filters out the initial coordinates that match the actual layout of the splicing display screen by quantifying the deviation between the row coordinates and the effective row range, thus avoiding cross-row and column matching errors caused by row offset.

[0156] S210. Check whether the spacing between adjacent column coordinates in the sorted initial coordinate set is within the preset spacing range;

[0157] S211. Remove the initial coordinates corresponding to the column coordinates, insert dark spot markers at the empty coordinate positions after removal, and perform the detection step of the next adjacent column coordinates until all column coordinates are detected, and generate the coordinate matrix of row clustering.

[0158] In this embodiment, steps S210 to S211 are similar to steps S106 to S107 in the previous embodiment, and will not be described again here.

[0159] S212. Starting from the first row of the coordinate matrix, traverse row by row. If all coordinates in a row of the coordinate matrix are unmarked and the number of coordinates in that row is consistent with the preset theoretical number, then that row is taken as the standard row.

[0160] In this embodiment, after generating the coordinate matrix for row clustering, it is necessary to select standard rows from the coordinate matrix that can serve as references. These standard rows mainly provide a reference for subsequent coordinate matching and correction. First, the coordinate matrix is ​​traversed row by row in order from the first row to the last row. During the traversal, two judgments need to be made for the coordinates of each row: First, check whether all coordinates in the currently traversed row are free of dark spot markings to ensure that there are no unlit dead spots or dark spots in the row, thus avoiding interference from dark spots on the accuracy of the reference row; Second, count the number of coordinates in the currently traversed row and compare it with the preset theoretical number. The preset theoretical number is the number of pixel cluster coordinates that should be in each row based on the sub-screen resolution, the number of splicing, and the pixel cluster arrangement rules of the splicing display. For example, for an N×M splicing screen with a sub-screen resolution of n×m, if the number of pixel clusters in each row of a single sub-screen is n, then the preset theoretical number can be set to N×n, which can be adjusted according to the actual shooting magnification.

[0161] In practical applications, the process starts from the first row of the coordinate matrix. If there are dark spots in the first row or the number of coordinates in the first row is inconsistent with the preset theoretical number, the first row is temporarily stored and the process starts to traverse the second row. If there are still dark spots in the second row or the number of coordinates in the first row is inconsistent with the preset theoretical number, the second row is temporarily stored and the process starts to traverse the third row, until the standard row is selected.

[0162] When a row in the row clustering coordinate matrix simultaneously satisfies both the conditions of "no dark spots on any coordinate" and "the number of coordinates matches the preset theoretical number," it indicates that the pixel clusters in that row are completely arranged without defects, meeting the requirements for a baseline. This row will then be used as the standard row. Determining the standard row through these two selection criteria ensures its accuracy and reliability, providing a stable benchmark for subsequent topology reconstruction and compensation of the coordinate matrix. This avoids deviations in subsequent correction work caused by dark spots or abnormal coordinate numbers in the benchmark row.

[0163] In practical applications, it should be noted that if each row of coordinates is marked with a dark spot or has no dark spot but the number of coordinates is inconsistent with the preset theoretical number, step S213 needs to be executed.

[0164] S213. If each row of coordinates has a dark spot mark or no dark spot mark but the number of coordinates is inconsistent with the preset theoretical number, then modify the coordinate matrix of row clustering to the coordinate matrix of column clustering or use the coordinate rules of column coordinates in the coordinate matrix to generate standard rows.

[0165] In this embodiment, after traversing all rows of the coordinate matrix for row clustering, if each row is found to have dark spot markers, or if the number of coordinates in a row is inconsistent with the preset theoretical number even without dark spot markers, then standard rows are generated using two schemes. Specifically, the first scheme is to modify the coordinate matrix for row clustering to a coordinate matrix for column clustering: transform the clustering dimension of the coordinate matrix, and re-cluster using the arrangement features of pixel clusters in the column direction, i.e., rotate the coordinate matrix for row clustering by 90° to obtain the coordinate matrix for column clustering, and then generate standard rows; the second scheme is to generate standard rows using the coordinate patterns of column coordinates in the coordinate matrix: perform statistical analysis on all column coordinates in the coordinate matrix to explore the arrangement patterns of column coordinates, such as the interval features of column coordinates, the overall distribution trend, etc., and determine the normal arrangement pattern of column coordinates by calculating the average spacing of most column coordinates, the start and end range of column coordinates, etc., and then construct a complete column coordinate sequence based on this arrangement pattern, combine it with the preset theoretical number to generate the corresponding row coordinates, and finally form standard rows.

[0166] It should be noted that if the first approach, which modifies the coordinate matrix of row clustering to that of column clustering, still fails to generate standard rows, then the second approach is executed to generate standard rows using the coordinate patterns of the column coordinates in the coordinate matrix.

[0167] When each row of coordinates is marked with a dark spot or not marked with a dark spot but the number of coordinates is inconsistent with the preset theoretical number, the two standard row generation schemes effectively solve the problem of missing standard rows in extreme cases. The two schemes start from the perspectives of dimension transformation and pattern mining, respectively, and can deal with different types of abnormal situations, further improving the success rate and accuracy of sub-pixel cluster positioning of spliced ​​display screens.

[0168] S214. Dynamically correct the coordinate matrix by using standard rows and combining them with a preset column spacing threshold;

[0169] In this embodiment, step S214 is similar to step S108 in the previous embodiment, and will not be described again here.

[0170] S215. Take the first row of the coordinate matrix that has completed dynamic correction as the new standard row;

[0171] In this embodiment, the goal of dynamic correction is to ensure that the coordinate arrangement of each row of the coordinate matrix is ​​consistent with the actual physical arrangement of the sub-pixel clusters on the display screen. The first row, after dynamic correction, has been verified by referencing the standard row and the preset column spacing threshold, eliminating coordinate deviations caused by splicing gaps, pose deviations, or abnormal lamp activation. The accuracy and arrangement regularity of the coordinates in this row meet the requirements for being used as a benchmark. Therefore, after completing the dynamic correction of the first row, this row is directly defined as the new standard row.

[0172] The first row after dynamic correction has both the baseline characteristics of the initial standard row and incorporates the actual correction parameters for the current splicing scenario. Compared with the initial standard row, it is more in line with the overall arrangement of the current coordinate matrix, and can provide a more accurate reference for the correction of subsequent rows, avoiding subsequent correction errors caused by deviations from the actual scenario.

[0173] S216. Dynamically correct the next row in the coordinate matrix that needs to be dynamically corrected by using a new standard row and a preset column spacing threshold. When the row is dynamically corrected, use the row as the new standard row. Repeat this process until all rows in the coordinate matrix are dynamically corrected.

[0174] In this embodiment, based on the new standard row that is updated in real time, all rows of the coordinate matrix are corrected row by row. During the correction process, the coordinates of the next row are adjusted with reference to the coordinate arrangement of the new standard row and in combination with a preset column spacing threshold. If the deviation between the column coordinates of the next row and the corresponding column coordinates of the new standard row exceeds a preset range, or the spacing between adjacent column coordinates does not meet the preset column spacing threshold, then the column coordinates are corrected according to the column spacing rules and coordinate arrangement trends of the new standard row, to ensure that the column spacing and overall arrangement of the next row are consistent with the new standard row after correction.

[0175] Once a row has been dynamically corrected, it is immediately updated as a new standard row to provide a reference for the next row to be corrected. For example, when correcting the second row, the first row is used as the standard row, and the coordinates of the second row are adjusted so that the column spacing of the second row coordinates meets the preset column spacing threshold. After the second row is corrected, it becomes the new standard row. The third row is then corrected based on this, and so on, updating the standard row in real time to perform dynamic correction until all rows have been corrected.

[0176] This embodiment achieves precise correction of all rows of the coordinate matrix through a cyclical mechanism that corrects row by row and updates the standard row in real time. The correction of each row is based on the latest dynamically corrected standard row, effectively avoiding the accumulation of errors under a single reference, significantly improving the correction accuracy of the entire coordinate matrix, and ensuring that the final corrected coordinate matrix can accurately reflect the actual physical position of the sub-pixel clusters of the spliced ​​display screen.

[0177] S217. The positioning of the sub-pixel clusters of the splicing display screen is obtained based on the corrected coordinate matrix.

[0178] In this embodiment, step S217 is similar to step S108 in the previous embodiment, and will not be described again here.

[0179] In an optional embodiment, please refer to Figure 7 and Figure 8 and combined Figure 2When the standard row cannot be determined, the row clustering of the coordinate matrix is ​​modified to column clustering or the coordinate rules of the column coordinates are used to generate the standard row.

[0180] When the standard row cannot be determined, the row-oriented clustering is modified to the column-oriented clustering or the standard row is generated by using the rules of column coordinates. The situations in which the standard row cannot be determined include the coexistence of abnormal number of row coordinate points, dark points, abnormal bright points, etc., which cause the average value of row coordinates to be distorted.

[0181] Specifically, when row coordinates cannot be determined, the coordinate matrix is ​​rotated 90 degrees counterclockwise or clockwise to transform the row-oriented problem into a column-oriented problem. Then, the mean of the column coordinates is calculated as a standard column, and the coordinate matrix is ​​dynamically corrected based on this standard column. If modifying the row-oriented clustering of the coordinate matrix to column-oriented clustering still fails to generate a valid reference, the symmetry of the row and column arrangement of the splicing display screen is utilized. When row-oriented processing fails, the reference is reconstructed through column-oriented features, or the regularity of column coordinates is used to derive a standard row, ensuring that a valid standard row can still be generated in extreme abnormal scenarios, thus maintaining the stability of the positioning method.

[0182] The positioning system for sub-pixel clusters of the splicing display screen provided in this application is described in detail below. Please refer to [link / reference]. Figure 3 , Figure 3 An embodiment of the positioning system for sub-pixel clusters of a splicing display screen provided in this application includes:

[0183] The acquisition unit 301 is used to acquire a target image in the splicing display screen, the target image including several sub-pixel clusters;

[0184] The first generation unit 302 is used to construct a redundant positioning mechanism based on the total amount of sub-pixel clusters, and generate dense template points according to the redundant positioning mechanism. The generation ratio of the redundant positioning mechanism is a preset multiple of the total amount.

[0185] The processing unit 303 is used to preprocess each sub-pixel cluster according to the Gaussian filtering algorithm. The preprocessing includes convolutional smoothing and noise reduction.

[0186] The second generation unit 304 is used to match the preprocessed sub-pixel clusters with dense template points to generate an initial coordinate set;

[0187] The sorting unit 305 is used to filter the initial coordinate set by a preset shake coefficient and a preset shooting magnification, and sort the filtered initial coordinate set according to the column coordinates.

[0188] The detection unit 306 is used to detect whether the spacing between adjacent column coordinates in the sorted initial coordinate set is within a preset spacing range.

[0189] The third generation unit 307 is used to remove the initial coordinates corresponding to the column coordinates if not, insert dark spot markers at the empty coordinate positions after removal, and perform the detection step of the next adjacent column coordinates until all column coordinates are detected, and generate the coordinate matrix of row clustering.

[0190] The correction unit 308 is used to obtain the standard row in the coordinate matrix, and dynamically correct the coordinate matrix by using the standard row and combining it with the preset column spacing threshold, and obtain the positioning of the sub-pixel cluster of the splicing display screen according to the corrected coordinate matrix.

[0191] Optionally, sorting unit 305 is specifically used for:

[0192] Sort the initial coordinates in the initial coordinate set in ascending order of row coordinates, and take the row coordinate of the first initial coordinate after sorting as the reference value;

[0193] The effective line range is calculated based on the baseline value and in combination with the preset shake coefficient and preset shooting magnification.

[0194] The row coordinates in the initial coordinate set are filtered by the effective row range, and the filtered initial coordinate set is sorted in ascending order of column coordinates.

[0195] The valid row range is represented by the following formula:

[0196] ;

[0197] in, Indicates the baseline value. Indicates the range of valid rows. Indicates the preset shooting magnification. This indicates the preset jitter coefficient.

[0198] Optionally, the correction unit 308 is specifically used for:

[0199] Starting from the first row of the coordinate matrix, traverse row by row. If all coordinates in a row of the coordinate matrix are unmarked and the number of coordinates in that row is consistent with the preset theoretical number, then that row is taken as the standard row.

[0200] If each row of coordinates has a dark spot marker or no dark spot marker but the number of coordinates is inconsistent with the preset theoretical number, then the coordinate matrix of row clustering will be modified to the coordinate matrix of column clustering, or the coordinate pattern of column coordinates in the coordinate matrix will be used to generate standard rows.

[0201] Optionally, the second generating unit 304 is specifically used for:

[0202] The dense template points are overlaid on the preprocessed sub-pixel clusters for localization and matching, and the coordinates of each sub-pixel cluster are obtained.

[0203] Clean up the coordinates and generate an initial coordinate set based on the cleaned coordinates;

[0204] Cleaning up the coordinates includes:

[0205] When multiple dense template points match the same sub-pixel cluster and result in multiple duplicate coordinates, the duplicate coordinates are cleaned up.

[0206] When the pixel value within the range matched by the dense template point is less than the first pixel threshold, the coordinates of the corresponding dense template point are cleared.

[0207] When the dense template point is located at the center of the sub-pixel cluster, the pixel value of the sub-pixel cluster is obtained, and the coordinates of the sub-pixel cluster are cleaned or retained based on the pixel value.

[0208] Optionally, it also includes a judgment unit 309, specifically used for

[0209] Obtain the pixel value of each sub-pixel cluster after preprocessing, and determine whether each pixel value is greater than the second pixel threshold in the preset triggering mechanism;

[0210] If so, then perform the step of matching sub-pixel clusters with dense template points.

[0211] Optionally, it also includes an update unit 310, specifically used for:

[0212] Use the first row of the coordinate matrix that has undergone dynamic correction as the new standard row;

[0213] The next row in the coordinate matrix that needs to be dynamically corrected is dynamically corrected by using a new standard row and a preset column spacing threshold. When the row is dynamically corrected, it is used as the new standard row. This process is repeated until all rows in the coordinate matrix have been dynamically corrected.

[0214] Optionally, the preset multiplier is 1.5 to 2 times the total amount.

[0215] For details on the implementation method, please refer to [link / reference]. Figure 1 - Figure 2 Examples will not be described in detail here.

[0216] This application also provides a positioning device for sub-pixel clusters of a splicing display screen; please refer to [link to relevant documentation]. Figure 4 , Figure 4 One embodiment of the positioning device for sub-pixel clusters of a splicing display screen provided in this application includes:

[0217] Processor 401, memory 402, input / output unit 403, bus 404;

[0218] The processor 401 is connected to the memory 402, the input / output unit 403, and the bus 404;

[0219] The memory 402 stores a program, and the processor 401 calls the program to execute any of the above positioning methods.

[0220] This application also relates to a computer-readable storage medium on which a program is stored, characterized in that when the program is run on a computer, it causes the computer to perform any of the above positioning methods.

[0221] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0222] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0223] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0224] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0225] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method of positioning sub-pixel groups of tiled display screens, characterized in that, The application relates to a method for locating sub-pixel groups in a spliced display screen. The method comprises the following steps: acquiring a target image in the spliced display screen, the target image comprising a plurality of sub-pixel groups; constructing a redundancy positioning mechanism according to the total number of the sub-pixel groups, and generating dense template points according to the redundancy positioning mechanism, the generation ratio of the redundancy positioning mechanism being a preset multiple of the total number; preprocessing each of the sub-pixel groups according to a Gaussian filtering algorithm, the preprocessing comprising convolution smoothing and noise reduction; matching the preprocessed sub-pixel groups with the dense template points to generate an initial coordinate set; screening the initial coordinate set through a preset jitter coefficient and a preset shooting magnification, and sorting the screened initial coordinate set according to column coordinates; detecting whether the spacing of adjacent column coordinates in the sorted initial coordinate set is within a preset spacing range; if not, removing the initial coordinates corresponding to the column coordinates, inserting a dark point mark at the removed empty coordinate position, and performing the detection step of the next adjacent column coordinate until the detection of all the column coordinates is completed, thereby generating a row-oriented clustered coordinate matrix; acquiring a standard row in the coordinate matrix, and dynamically correcting the coordinate matrix through the standard row and in combination with a preset column spacing threshold, and obtaining the positioning of the sub-pixel groups in the spliced display screen according to the corrected coordinate matrix; the matching of the preprocessed sub-pixel groups with the dense template points to generate an initial coordinate set comprises: positioning and matching the dense template points on the preprocessed sub-pixel groups to obtain the coordinates of each of the sub-pixel groups; cleaning the coordinates, and generating an initial coordinate set according to the cleaned coordinates; the cleaning of the coordinates comprises: when a plurality of the dense template points are positioned and matched to the same sub-pixel group to obtain a plurality of repeated coordinates, the repeated coordinates are cleaned; when the pixel value in the matching range of the dense template point is less than a first pixel threshold, the coordinates of the corresponding dense template point are cleaned; 2. The positioning method according to claim 1, characterized in that, when the dense template point is located at the center of the sub-pixel group, the pixel value of the sub-pixel group is acquired, and the coordinates of the sub-pixel group are cleaned or retained according to the pixel value. the screening of the initial coordinate set through a preset jitter coefficient and a preset shooting magnification, and the sorting of the screened initial coordinate set according to column coordinates comprises: sorting the initial coordinates in the initial coordinate set in ascending order of row coordinates, and taking the row coordinate of the first initial coordinate after sorting as a reference value; calculating an effective row range based on the reference value and in combination with a preset jitter coefficient and a preset shooting magnification; screening the row coordinates in the initial coordinate set through the effective row range, and sorting the screened initial coordinate set in ascending order of column coordinates; ； wherein, represents the reference value, represents the valid row range, represents the preset photographing magnification, represents the preset shake coefficient.

3. The positioning method of claim 1, wherein, the effective row range is represented by the following formula: the acquisition of a standard row in the coordinate matrix comprises: starting from the first row of the coordinate matrix, traversing row by row, if all the coordinates of one row of the coordinate matrix are not marked with a dark point, and the number of the coordinates of the row is consistent with a preset theoretical number, the row is taken as a standard row. If all the coordinates in each row are marked with dark spots or are not marked with dark spots but the number of the coordinates is inconsistent with the preset theoretical number, the row of the coordinate matrix is modified to the column of the coordinate matrix or the standard row is generated by using the coordinate rule of the column coordinates in the coordinate matrix.

4. The positioning method of claim 1, wherein, After the preprocessing of each of the sub-pixel groups according to the Gaussian filter algorithm, the method further comprises: obtaining the pixel value of each of the preprocessed sub-pixel groups, and determining whether each of the pixel values is greater than a second pixel threshold in a preset trigger mechanism; if yes, performing the step of matching the sub-pixel group with the dense template point.

5. The positioning method according to any one of claims 1 to 4, characterized in that, After the dynamic modification of the coordinate matrix by the standard row and in combination with the preset column spacing threshold, the method further comprises: taking the first row in the coordinate matrix that has completed the dynamic modification as a new standard row; performing dynamic modification on the next row in the coordinate matrix that needs dynamic modification by the new standard row and in combination with the preset column spacing threshold, and taking the row as the new standard row when the dynamic modification of the row is completed, and repeating sequentially until all the rows in the coordinate matrix have completed the dynamic modification.

6. The positioning method according to any one of claims 1 to 4, characterized in that, The preset multiple is 1.5 times to 2 times of the total amount.

7. A positioning system for tiled display screen sub-pixel clusters, characterized by, A device for performing the positioning method according to any one of claims 1-6, comprising: an obtaining unit configured to obtain a target image in a spliced display screen, the target image comprising a plurality of sub-pixel groups; a first generating unit configured to construct a redundant positioning mechanism according to a total amount of the sub-pixel groups, and generate dense template points according to the redundant positioning mechanism, a generation ratio of the redundant positioning mechanism being a preset multiple of the total amount; a processing unit configured to preprocess each of the sub-pixel groups according to a Gaussian filter algorithm, the preprocessing comprising convolution smoothing and noise reduction; a second generating unit configured to match the preprocessed sub-pixel groups with the dense template points to generate an initial coordinate set; an ordering unit configured to screen the initial coordinate set by a preset dithering coefficient and a preset shooting magnification, and order the screened initial coordinate set according to column coordinates; a detecting unit configured to detect whether the spacing between adjacent column coordinates in the ordered initial coordinate set is within a preset spacing range; a third generating unit configured to, if not, remove the initial coordinates corresponding to the column coordinates, insert dark spot marks at the positions of the removed empty coordinates, and perform the detection step of the next adjacent column coordinates until the detection of all the column coordinates is completed, to generate a row-clustering coordinate matrix; a modifying unit configured to obtain a standard row in the coordinate matrix, and perform dynamic modification of the coordinate matrix by the standard row and in combination with a preset column spacing threshold, to obtain the positioning of the sub-pixel groups of the spliced display screen according to the modified coordinate matrix.

8. A positioning device for tiling sub-pixel groups of a display screen, characterized in that, comprise: a processor, a memory, an input / output unit, and a bus; the processor is connected with the memory, the input / output unit, and the bus; the memory stores a program, and the processor invokes the program to execute the positioning method according to any one of claims 1-6. comprise: a processor, a memory, an input / output unit, and a bus; the processor is connected with the memory, the input / output unit, and the bus; the memory stores a program, and the processor invokes the program to execute the positioning method according to any one of claims 1-6.

9. A computer readable storage medium having a program recorded thereon, the program, when executed on a computer, performing the positioning method according to any one of claims 1 to 6.

Images