A mark point coordinate detection method based on image sequences and related devices

Abstract

The application discloses a Mark point coordinate detection method based on an image sequence and a related device, and is used for improving display screen Mark point positioning precision. Image acquisition of N frames in succession is performed on a to-be-positioned object provided with a Mark point; Mark point coordinates of a photographed image sequence are extracted to generate a candidate coordinate point set; according to each candidate coordinate point, a gray feature of a photographed image where the candidate coordinate point is located is extracted to generate a gray feature set; a position deviation amount of each candidate coordinate point is calculated according to the candidate coordinate point set to generate a position deviation amount set; a gray consistency parameter of each candidate coordinate point is calculated according to the gray feature set; a multi-dimension confidence index of each candidate coordinate point in the candidate coordinate point set is calculated according to the position deviation amount set and the gray consistency parameter set; the candidate coordinate point set is screened according to the multi-dimension confidence index set, and a target coordinate of the Mark point is generated by using a screened candidate coordinate point subset.

Description

Technical Field

[0001] This application relates to the field of display screen inspection, and in particular to a method and apparatus for detecting Mark point coordinates based on image sequences. Background Technology

[0002] With technological innovation, new display technologies are constantly being updated and iterated. As a core link in the display industry chain, display quality inspection technology continues to receive attention, especially the defect detection of micron-level pixels on the display surface, which has become a key factor determining the quality of display products. Before defect detection, it is necessary to accurately locate the position of the pixels on the display screen under test to improve the quality of subsequent defect detection and brightness compensation.

[0003] However, the pixel precision of new displays is gradually increasing, and the arrangement of pixels is becoming increasingly sophisticated and complex. This makes mark point detection and positioning more difficult in the automatic optical inspection of new displays (such as OLED, Micro-LED, and flexible foldable screens). Conventional mark point detection and positioning uses a single-frame detection method. The alignment system acquires a single frame image and identifies the mark point position directly as the alignment reference. However, this method is highly susceptible to fluctuations in lighting, noise, scratches, blemishes, or glass reflections. Furthermore, for high-precision, structurally complex new displays, it is more prone to unstable detection results. When mark points are partially missing or blurred, false detections and missed detections are likely, affecting alignment accuracy. To address these issues, a multi-frame averaging method has emerged. The alignment system continuously acquires multiple frames of images of the display under test and performs a simple spatial averaging of the mark point position results for each frame to reduce the impact of single-frame anomalies. While this method improves robustness to some extent and reduces the impact of single-point errors in a single frame, it lacks quality assessment for each frame, making it unable to effectively eliminate low-quality or erroneous candidate points. Furthermore, the average value is still skewed when there are a few obviously abnormal frames or a large number of inconspicuously abnormal frames. To address these issues, existing technologies use template matching enhancement methods for Mark point detection and localization. Mark points are filtered based on a calculated template similarity threshold using grayscale template matching or feature matching. This method can reduce some noise interference, but it relies on template design and optical consistency, making the detection results highly susceptible to the acquisition environment. It can be seen that the similarity threshold also cannot comprehensively consider the stability of the detected point position, making it difficult to eliminate abnormal points and reducing the accuracy of Mark point localization on the display screen. Summary of the Invention

[0004] This application discloses a Mark point coordinate detection method and related apparatus based on image sequences, which can be used to improve the Mark point positioning accuracy of a display screen.

[0005] In a first aspect, embodiments of this application provide a method for detecting Mark point coordinates based on image sequences, comprising: The object to be aligned with the Mark point is imaged continuously for N frames to generate a sequence of captured images. In each captured image, the point in the image corresponding to the Mark point is regarded as a candidate point. Extract the coordinates of the Mark point in each captured image from the captured image sequence to generate a set of candidate coordinate points; Based on each candidate coordinate point, grayscale features are extracted from the captured image to generate a grayscale feature set. Calculate the position deviation for each candidate coordinate point based on the candidate coordinate point set, and generate a position deviation set. Calculate the gray-level consistency parameter for each candidate coordinate point based on the gray-level feature set, and generate a gray-level consistency parameter set. Calculate a multi-dimensional confidence index for each candidate coordinate point in the candidate coordinate point set based on the set of positional deviation and the set of grayscale consistency parameters, and generate a multi-dimensional confidence index set. The candidate coordinate point set is filtered based on the multi-dimensional confidence index set, and the target coordinates of the Mark point are generated using the filtered subset of candidate coordinate points.

[0006] Optionally, the step of calculating the position deviation for each candidate coordinate point based on the set of candidate coordinate points and generating the set of position deviations specifically includes: Generate a cluster center coordinate based on the coordinates of all candidate coordinate points in the candidate coordinate point set; The positional deviation is calculated for candidate coordinate points in each frame of the captured image based on the coordinates of the cluster center, and a set of positional deviations is generated. The formula is as follows:

[0007] in, The coordinates of the cluster center Let i be the i-th candidate coordinate point. Let be the positional deviation of the i-th candidate coordinate point.

[0008] Optionally, the grayscale features in the grayscale feature set include the image signal-to-noise ratio; the specific steps for generating the grayscale feature set by extracting grayscale features from the captured image for each candidate coordinate point include: For each frame of captured image, an image signal-to-noise ratio (SNR) is generated from the effective image signal and noise signal. The SNR is then correlated with the corresponding candidate coordinate points to generate a grayscale feature set.

[0009] Optionally, the grayscale features in the grayscale feature set include template matching similarity scores; the specific steps for generating the grayscale feature set by extracting grayscale features from the captured image for each candidate coordinate point include: A local feature acquisition region is generated based on the morphological parameters of each candidate coordinate point; Based on the gray values ​​of pixels in the local feature acquisition area and the gray values ​​of corresponding pixels in the template image, a template matching similarity score is generated for each candidate coordinate point.

[0010] Optionally, the grayscale features in the grayscale feature set include local region contrast or image grayscale gradient intensity; the specific steps for generating the grayscale feature set by extracting grayscale features from the captured image for each candidate coordinate point include: A local feature acquisition region is generated based on the morphological parameters of each candidate coordinate point; The local region contrast or gray-level gradient intensity is generated based on the gray-level values ​​in the local feature acquisition area; the gray-level gradient intensity includes horizontal gray-level gradient intensity and vertical gray-level gradient intensity.

[0011] Optionally, the step of calculating a gray-level consistency parameter for each candidate coordinate point based on the gray-level feature set and generating a gray-level consistency parameter set specifically includes: Generate a gray-level feature mean based on the gray-level feature set; Based on the grayscale feature mean and the consistency function, grayscale consistency parameters are calculated for the corresponding candidate coordinate points in each frame of the captured image, generating a set of grayscale consistency parameters. The consistency function is as follows:

[0012] in, Let be the arithmetic mean of the grayscale features of all N candidate coordinate points. It refers to the grayscale features corresponding to the candidate coordinate points in the i-th captured image.

[0013] Optionally, the step of calculating a multi-dimensional confidence index for each candidate coordinate point in the candidate coordinate point set based on the set of positional deviation and the set of grayscale consistency parameters, and generating the multi-dimensional confidence index set specifically includes: Determine the maximum deviation based on the set of positional deviations; Calculate the normalized position deviation for each candidate coordinate point in the candidate coordinate point set based on the maximum deviation, the position normalization function, and the set of position deviations, and generate the normalized position deviation set. By performing a weighted sum based on the normalized position deviation set and the grayscale consistency parameter set, a multi-dimensional confidence index is calculated for each candidate coordinate point in the candidate coordinate point set, generating a multi-dimensional confidence index set.

[0014] Optionally, the steps of filtering the candidate coordinate point set based on the multi-dimensional confidence index set and generating target coordinates for the Mark point using the filtered subset of candidate coordinate points specifically include: Each candidate coordinate point is sorted according to the multi-dimensional confidence index set, and candidate coordinate points that meet the preset screening conditions are selected as the reliable coordinate point set. The target coordinates of the Mark point are generated by taking a weighted average of the reliable coordinate point set and the corresponding index in the multi-dimensional confidence index set. The formula for generating the target coordinates of the Mark point is as follows:

[0015] Where K is the number of candidate coordinate points, For reliable coordinate point coordinates, Let be the multi-dimensional confidence index of the i-th candidate coordinate point. The target coordinates are given.

[0016] Optionally, after the step of acquiring N consecutive frames of images of the object to be aligned with Mark points to generate an image sequence, and before the step of extracting the Mark point coordinates of each image in the image sequence to generate a candidate coordinate point set, the Mark point coordinate detection method further includes: Median filtering is applied to the captured images in the image sequence for noise reduction.

[0017] Secondly, embodiments of this application provide a Mark point coordinate detection device based on an image sequence, comprising: The acquisition unit acquires N consecutive frames of images of the object to be aligned with Mark points, generating a sequence of captured images. In each captured image, the point in the image corresponding to the Mark point is considered as a candidate point. The candidate coordinate point set generation unit is used to extract the Mark point coordinates of each captured image in the captured image sequence and generate a candidate coordinate point set; The grayscale feature set generation unit is used to extract grayscale features from the captured image for each candidate coordinate point and generate a grayscale feature set. The position deviation set generation unit is used to calculate the position deviation for each candidate coordinate point based on the candidate coordinate point set and generate a position deviation set. The grayscale consistency parameter set generation unit is used to calculate the grayscale consistency parameter for each candidate coordinate point based on the grayscale feature set, and generate the grayscale consistency parameter set. The multi-dimensional confidence index generation unit is used to calculate the multi-dimensional confidence index for each candidate coordinate point in the candidate coordinate point set based on the set of position deviation and the set of grayscale consistency parameters, and generate a multi-dimensional confidence index set. The target coordinate generation unit is used to filter the set of candidate coordinate points based on a multi-dimensional confidence index set, and to generate target coordinates for the Mark point using the filtered subset of candidate coordinate points.

[0018] Optionally, the position deviation set generation unit is specifically used to include: Generate a cluster center coordinate based on the coordinates of all candidate coordinate points in the candidate coordinate point set; The positional deviation is calculated for candidate coordinate points in each frame of the captured image based on the coordinates of the cluster center, and a set of positional deviations is generated. The formula is as follows:

[0019] in, The coordinates of the cluster center Let i be the i-th candidate coordinate point. Let be the positional deviation of the i-th candidate coordinate point.

[0020] Optionally, the grayscale features in the grayscale feature set may include the image signal-to-noise ratio; The grayscale feature set generation unit is specifically used for including: For each frame of captured image, an image signal-to-noise ratio (SNR) is generated from the effective image signal and noise signal. The SNR is then correlated with the corresponding candidate coordinate points to generate a grayscale feature set.

[0021] Optionally, the grayscale features in the grayscale feature set include template matching similarity scores; The grayscale feature set generation unit is specifically used for including: A local feature acquisition region is generated based on the morphological parameters of each candidate coordinate point; Based on the gray values ​​of pixels in the local feature acquisition area and the gray values ​​of corresponding pixels in the template image, a template matching similarity score is generated for each candidate coordinate point.

[0022] Optionally, the gray-level features in the gray-level feature set may include local region contrast or image gray-level gradient intensity; The grayscale feature set generation unit is specifically used for including: A local feature acquisition region is generated based on the morphological parameters of each candidate coordinate point; The local region contrast or gray-level gradient intensity is generated based on the gray-level values ​​in the local feature acquisition area; the gray-level gradient intensity includes horizontal gray-level gradient intensity and vertical gray-level gradient intensity.

[0023] Optionally, the grayscale consistency parameter set generation unit is specifically used to include: Generate a gray-level feature mean based on the gray-level feature set; Based on the grayscale feature mean and the consistency function, grayscale consistency parameters are calculated for the corresponding candidate coordinate points in each frame of the captured image, generating a set of grayscale consistency parameters. The consistency function is as follows:

[0024] in, Let be the arithmetic mean of the grayscale features of all N candidate coordinate points. It refers to the grayscale features corresponding to the candidate coordinate points in the i-th captured image.

[0025] Optionally, the multi-dimensional confidence index generation unit is specifically used to include: Determine the maximum deviation based on the set of positional deviations; Calculate the normalized position deviation for each candidate coordinate point in the candidate coordinate point set based on the maximum deviation, the position normalization function, and the set of position deviations, and generate the normalized position deviation set. By performing a weighted sum based on the normalized position deviation set and the grayscale consistency parameter set, a multi-dimensional confidence index is calculated for each candidate coordinate point in the candidate coordinate point set, generating a multi-dimensional confidence index set.

[0026] Optionally, the target coordinate generation unit is specifically used to include: Each candidate coordinate point is sorted according to the multi-dimensional confidence index set, and candidate coordinate points that meet the preset screening conditions are selected as the reliable coordinate point set. The target coordinates of the Mark point are generated by taking a weighted average of the reliable coordinate point set and the corresponding index in the multi-dimensional confidence index set. The formula for generating the target coordinates of the Mark point is as follows:

[0027] Where K is the number of candidate coordinate points, For reliable coordinate point coordinates, Let be the multi-dimensional confidence index of the i-th candidate coordinate point. The target coordinates are given.

[0028] Optionally, the Mark point coordinate detection device also includes: The median filtering noise reduction processing unit is used to perform median filtering noise reduction processing on the captured images in the captured image sequence generated by the acquisition unit.

[0029] As can be seen from the above technical solutions, the embodiments of this application have the following advantages: In this application, firstly, N consecutive frames of images are acquired on the object to be aligned, with Mark points, generating a sequence of captured images. In each captured image, the point corresponding to the Mark point is considered a candidate point. The coordinates of the Mark point in each captured image are extracted, generating a set of candidate coordinate points. For each candidate coordinate point, grayscale features are extracted from its corresponding captured image, generating a set of grayscale features. The positional deviation is calculated for each candidate coordinate point based on the candidate coordinate point set, generating a set of positional deviations. A grayscale consistency parameter is calculated for each candidate coordinate point based on the grayscale feature set, generating a set of grayscale consistency parameters. A multi-dimensional confidence index is calculated for each candidate coordinate point in the candidate coordinate point set based on the positional deviation set and the grayscale consistency parameter set, generating a set of multi-dimensional confidence indices. The candidate coordinate point set is then filtered based on the multi-dimensional confidence index set, and the filtered subset of candidate coordinate points is used to generate target coordinates for the Mark point.

[0030] By continuously acquiring multiple frames of images from the display screen under test, numerous noise interferences are generated, causing deviations in the position and grayscale features of the Mark points. These noises include those caused by sudden changes in ambient brightness, as well as noise from the display screen and the industrial camera. Initial positioning of the Mark points in each frame is then performed to obtain candidate coordinate points. Next, grayscale feature acquisition processing is performed on different regions of each frame based on these candidate coordinate points. After acquisition, the positional deviation and grayscale consistency parameter set of each candidate coordinate point are analyzed. This involves performing positional stability and grayscale consistency analysis on the candidate coordinate points detected in the multi-frame image sequence, constructing a comprehensive confidence scoring model. This model allows for the screening of candidate coordinate points based on coordinate position and grayscale features, eliminating those that do not meet the criteria. Finally, target coordinates are generated for each Mark point based on a multi-dimensional confidence index set and the candidate coordinate points that meet the criteria. By analyzing candidate coordinate points obtained from multiple frames of captured images from multiple dimensions, and by using a joint scoring system of position deviation and grayscale consistency, dynamic screening of candidate points is achieved. Through a multi-dimensional data fusion mechanism, outliers are eliminated, single-frame noise interference is avoided, and the positioning accuracy of Mark points on the display screen is improved. Attached Figure Description

[0031] 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.

[0032] Figure 1 This is a schematic diagram of the Mark point coordinate detection method based on image sequences in this application; Figure 2 A schematic diagram of the method for generating a set of position deviations for this application; Figure 3 A schematic diagram of the method for generating a grayscale feature set for this application; Figure 4 Another schematic diagram illustrating the method for generating a grayscale feature set for this application; Figure 5 Another schematic diagram illustrating the method for generating a grayscale feature set for this application; Figure 6 A schematic diagram of the method for generating a set of grayscale consistency parameters for this application; Figure 7 A schematic diagram illustrating the method for calculating the multi-dimensional confidence index in this application; Figure 8 A schematic diagram of the method for generating the coordinates of the target Mark point in this application; Figure 9 A schematic diagram illustrating the image processing method for this application; Figure 10 This is a schematic diagram of the Mark point coordinate detection device based on image sequence according to this application. Detailed Implementation

[0033] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0034] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0035] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0036] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0037] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0038] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0039] In existing Mark point localization methods, single-frame detection acquires a single frame image through an alignment system and identifies the Mark point position detected in that frame, directly using it as the alignment reference. However, this method is susceptible to fluctuations in lighting, noise, scratches, blemishes, or glass reflections. Furthermore, for high-precision, structurally complex new displays, it is more prone to unstable detection results. When Mark points are partially missing or blurred, false detections and missed detections can easily occur, affecting alignment accuracy. To address these issues, existing technologies employ multi-frame averaging methods. The alignment system continuously acquires multiple frames of images of the display under test, performing a simple spatial averaging of the Mark point position results in each frame to reduce the impact of single-frame anomalies. While this method improves robustness to some extent and reduces the impact of single-point errors in a single frame, it lacks quality assessment for each frame's results. This makes it difficult to effectively eliminate low-quality or erroneous candidate points, and the average value can still be skewed when there are a few obviously abnormal frames or a large number of inconspicuously abnormal frames. To address the aforementioned issues, existing technologies employ template matching enhancement methods for Mark point detection and localization. These methods use grayscale template matching or feature matching to filter Mark points based on a calculated template similarity threshold. While this approach can reduce some noise interference, it relies heavily on template design and optical consistency, making the detection results highly susceptible to environmental factors. Furthermore, the similarity threshold cannot comprehensively consider the stability of the detected point's position, making it difficult to eliminate outliers and ultimately reducing the accuracy of Mark point localization on the display screen.

[0040] Based on this, this application discloses a Mark point coordinate detection method and related apparatus based on image sequences, which can improve the Mark point positioning accuracy of a display screen.

[0041] This method constructs a comprehensive credibility scoring model by performing multi-dimensional feature analysis (including positional stability and grayscale consistency) on candidate coordinate points detected in a multi-frame image sequence. Based on this score, the model is sorted and weighted to automatically select the most reliable detection results and finally output a Mark point coordinate with high accuracy and good stability, which significantly improves the performance of the visual alignment system under complex working conditions.

[0042] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0043] The method described in this application can be applied to servers, devices, terminals, or other devices with logical processing capabilities; therefore, this application does not limit its application. For ease of description, the following description uses a terminal as the executing entity.

[0044] Please see Figure 1 This application provides an embodiment of a Mark point coordinate detection method based on image sequences, including: 101. Acquire N consecutive frames of images of the object to be aligned with Mark points to generate a sequence of captured images, wherein the points in each captured image corresponding to the Mark points are considered as candidate points; In this embodiment, the terminal inputs the positioning image into the display screen under test. Multiple Mark points are set on the positioning image. After the display screen under test is lit up, an industrial camera continuously captures N frames of images of the display screen. The captured images will also display the Mark points, but these Mark points are not necessarily accurate and need to be initially located and then filtered. Specifically, during the alignment process, N frames (N≥3) of images containing the target Mark points are continuously captured, and the points in each captured image corresponding to the Mark points are considered as candidate points.

[0045] 102. Extract the coordinates of the Mark point of each captured image in the captured image sequence and generate a set of candidate coordinate points; In this embodiment, the terminal performs Mark point detection and localization on each frame of captured images using a preset Mark point detection algorithm to obtain initial candidate Mark point positions (candidate coordinate points). In N frames of captured images, each Mark point corresponds to N candidate coordinate points. The terminal adaptively generates a local grayscale acquisition area centered on the candidate coordinate points. Specifically, Mark point detection and localization extracts the coordinates of the target Mark position in each frame of the captured image. As candidate coordinate points, grayscale features will be selected and collected based on these candidate coordinate points.

[0046] 103. Based on each candidate coordinate point, perform grayscale feature extraction on the captured image to generate a grayscale feature set; In this embodiment, the terminal determines the grayscale acquisition area based on the candidate coordinate points. This grayscale acquisition area may be the entire effective display area of ​​the captured image, or it may be a acquisition area within the effective display area centered on the candidate coordinate points. The acquired grayscale features are defined as follows: ,in, This refers to the gray-level features corresponding to the target candidate coordinates in the i-th captured image, forming the gray-level feature set of the target candidate coordinates. grayscale features The grayscale feature set can be one or more of the following: template matching similarity score, local region contrast, signal-to-noise ratio (SNR), or image grayscale gradient intensity, depending on the type of display screen being tested. The specific generation of the grayscale feature set will be described in detail in subsequent embodiments.

[0047] 104. Calculate the position deviation for each candidate coordinate point based on the candidate coordinate point set, and generate a position deviation set; In this embodiment, the terminal calculates the position deviation for each candidate coordinate point based on the candidate coordinate point set, and generates a position deviation set. Specifically, it sets the N candidate coordinate points corresponding to a Mark point. The coordinates are integrated, and the deviations between the integrated coordinates and the N candidate coordinate points are compared and analyzed to generate a set of position deviations. The specific method for generating the set of position deviations will be explained in subsequent embodiments.

[0048] 105. Calculate the gray-level consistency parameter for each candidate coordinate point based on the gray-level feature set, and generate a gray-level consistency parameter set; In this embodiment, the terminal calculates a grayscale consistency parameter for each candidate coordinate point based on the grayscale feature set, and generates a grayscale consistency parameter set. Specifically, the grayscale features corresponding to all candidate coordinate points are integrated, and a consistency analysis is performed on each grayscale feature and the integrated grayscale parameter to generate a grayscale consistency parameter set. The specific method for generating the grayscale consistency parameter set will be described in detail in subsequent embodiments.

[0049] 106. Calculate the multi-dimensional confidence index for each candidate coordinate point in the candidate coordinate point set based on the set of positional deviation and the set of grayscale consistency parameters, and generate a multi-dimensional confidence index set. In this embodiment, the terminal calculates a multi-dimensional confidence index for each candidate coordinate point in the candidate coordinate point set based on the set of position deviation and the set of grayscale consistency parameters. The larger the position deviation of a candidate coordinate point, the smaller the multi-dimensional confidence index, while the larger the grayscale consistency parameter, the larger the multi-dimensional confidence index. The specific method for calculating the multi-dimensional confidence index will be described in detail in subsequent embodiments.

[0050] 107. Filter the candidate coordinate point set based on the multi-dimensional confidence index set, and use the filtered candidate coordinate point subset to generate target coordinates for the Mark point.

[0051] After the terminal obtains the multi-dimensional confidence index, it filters each candidate coordinate point, selects points that meet the preset acceptance conditions from the N candidate coordinate points of the Mark point, removes points with too large differences, and finally generates the target Mark point coordinates based on the filtered points. The specific method of generating the target Mark point coordinates will be described in detail in the subsequent embodiments.

[0052] In this embodiment, firstly, N consecutive frames of images are acquired on the object to be aligned, with Mark points, generating a sequence of captured images. Points in each captured image corresponding to the Mark points are considered candidate points. The coordinates of the Mark points in each captured image are extracted, generating a set of candidate coordinate points. For each candidate coordinate point, grayscale features are extracted from its corresponding captured image, generating a set of grayscale features. The positional deviation is calculated for each candidate coordinate point based on the candidate coordinate point set, generating a set of positional deviations. A grayscale consistency parameter is calculated for each candidate coordinate point based on the grayscale feature set, generating a set of grayscale consistency parameters. A multi-dimensional confidence index is calculated for each candidate coordinate point in the candidate coordinate point set based on the positional deviation set and the grayscale consistency parameter set, generating a set of multi-dimensional confidence indices. The candidate coordinate point set is then filtered based on the multi-dimensional confidence index set, and the filtered subset of candidate coordinate points is used to generate target coordinates for the Mark points.

[0053] By continuously acquiring multiple frames of images from the display screen under test, numerous noise interferences are generated, causing deviations in the position and grayscale features of the Mark points. These noises include those caused by sudden changes in ambient brightness, as well as noise from the display screen and the industrial camera. Initial positioning of the Mark points in each frame is then performed to obtain candidate coordinate points. Next, grayscale feature acquisition processing is performed on different regions of each frame based on these candidate coordinate points. After acquisition, the positional deviation and grayscale consistency parameter set of each candidate coordinate point are analyzed. This involves performing positional stability and grayscale consistency analysis on the candidate coordinate points detected in the multi-frame image sequence, constructing a comprehensive confidence scoring model. This model allows for the screening of candidate coordinate points based on coordinate position and grayscale features, eliminating those that do not meet the criteria. Finally, target coordinates are generated for each Mark point based on a multi-dimensional confidence index set and the candidate coordinate points that meet the criteria. By analyzing candidate coordinate points obtained from multiple frames of captured images from multiple dimensions, and by using a joint scoring system of position deviation and grayscale consistency, dynamic screening of candidate points is achieved. Through a multi-dimensional data fusion mechanism, outliers are eliminated, single-frame noise interference is avoided, and the positioning accuracy of Mark points on the display screen is improved.

[0054] Please see Figure 2 This application provides an embodiment of a method for generating a set of position deviations, comprising: 201. Generate a cluster center coordinate based on the coordinates of all candidate coordinate points in the candidate coordinate point set; 202. Calculate the positional deviation of candidate coordinate points in each frame of the captured image based on the coordinates of the cluster center, and generate a set of positional deviations. The formula is as follows:

[0055] in, The coordinates of the cluster center Let i be the i-th candidate coordinate point. Let be the positional deviation of the i-th candidate coordinate point.

[0056] In this embodiment, the terminal can choose to directly calculate the arithmetic mean of the N candidate coordinates of the Mark point to generate the cluster center coordinates. .

[0057] Next, the terminal uses the cluster center coordinates. Calculate the positional deviation of N candidate coordinate points In a conventional display screen, the terminal can calculate each candidate point. To the candidate point cluster center The Euclidean distance is used to measure the deviation of the frame position from the overall position, and the formula is as follows:

[0058] Please see Figure 3 This application provides an embodiment of a method for generating a grayscale feature set, wherein the grayscale features in the grayscale feature set include the image signal-to-noise ratio; 301. Generate the image signal-to-noise ratio (SNR) for each frame of captured image from the effective image signal and the noise signal, and associate the image SNR with the corresponding candidate coordinate points to generate a grayscale feature set.

[0059] In this embodiment, the signal-to-noise ratio (SNR) is used to characterize the intensity relationship between the effective image signal and the noise component in the whole frame image, and it can be calculated by image gray-scale statistical characteristics or noise estimation methods.

[0060] In this embodiment, the terminal generates the image signal-to-noise ratio based on the effective image signal and invalid noise signal of the entire captured image. Generate image signal-to-noise ratio Subsequently, all candidate coordinate points of the same captured image are bound to the image signal-to-noise ratio and become one of the grayscale features of the candidate coordinate points.

[0061] Please see Figure 4 This application provides an embodiment of a method for generating a grayscale feature set, wherein the grayscale features in the grayscale feature set include template matching similarity scores; 401. Generate a local feature acquisition region based on the morphological parameters of each candidate coordinate point; In this embodiment, in order to perform grayscale acquisition on each candidate coordinate point, a local feature acquisition area is usually generated by taking the candidate coordinate point as the center and using the morphological parameters (shape and size, etc.) of the Mark point on the positioning screen.

[0062] 402. Based on the gray values ​​of pixels in the local feature acquisition area and the gray values ​​of corresponding pixels in the template image, generate a template matching similarity score for each candidate coordinate point.

[0063] Template matching similarity score is an evaluation metric calculated based on the gray-level distribution or structural features of a local region, used to characterize the degree of similarity between the region to be detected and a preset template. The terminal generates a template matching similarity score for each candidate coordinate point based on the gray-level values ​​of pixels in the template image (preset template) and the gray-level values ​​of pixels in the actual local feature acquisition region, thus generating a gray-level feature set. The terminal retrieves the template image from the database and analyzes the gray-level values ​​of pixels in the template image's Mark point region with the gray-level values ​​of pixels in the actual acquired local feature acquisition region. The Mark point region of the template image has the same shape and size as the local feature acquisition region. It should be noted that the Mark point region of the template image is a preset reference region, and the positional relationship between each pixel in the Mark point region and the corresponding region in the template image is predetermined.

[0064] Please see Figure 5 This application provides an embodiment of a method for generating a grayscale feature set, wherein the grayscale features in the grayscale feature set include local region contrast or image grayscale gradient intensity; 501. Generate a local feature acquisition region based on the morphological parameters of each candidate coordinate point; In this embodiment, in order to perform grayscale acquisition on each candidate coordinate point, a local feature acquisition area is usually generated by taking the candidate coordinate point as the center and using the morphological parameters (shape and size, etc.) of the Mark point on the positioning screen.

[0065] 502. Generate local region contrast or gray-level gradient intensity based on the gray-level values ​​in the local feature acquisition area; gray-level gradient intensity includes horizontal gray-level gradient intensity and vertical gray-level gradient intensity.

[0066] Local area contrast refers to the degree of difference in grayscale values ​​within a specific local area of ​​an image. It can be measured by calculating the difference between the maximum and minimum grayscale values ​​within that area, or by other statistical methods (such as standard deviation). In this embodiment, the local area contrast is set to... .

[0067] Image grayscale gradient intensity refers to the rate of change of grayscale values ​​in an image. It can be obtained by calculating the grayscale gradient of each pixel in the image (i.e., the rate of change of the grayscale value at that point in the horizontal and vertical directions). Grayscale gradient intensity reflects the drasticness of grayscale changes in the image. In this embodiment, the grayscale gradient intensity is set to... .

[0068] The terminal correlates the template matching similarity score, local region contrast, gray-level gradient intensity, and signal-to-noise ratio of each candidate point, and finally integrates all the template matching similarity scores, local region contrast, gray-level gradient intensity, and signal-to-noise ratio to generate a gray-level feature set.

[0069] Please see Figure 6 This application provides an embodiment of a method for generating a set of grayscale consistency parameters, comprising: 601. Generate a gray-level feature mean based on the gray-level feature set; 602. Calculate gray-level consistency parameters for the corresponding candidate coordinate points in each frame of the captured image based on the gray-level feature mean and consistency function, generating a set of gray-level consistency parameters. The consistency function is as follows:

[0070] in, Let be the arithmetic mean of the grayscale features of all N candidate coordinate points. It refers to the grayscale features corresponding to the candidate coordinate points in the i-th captured image.

[0071] In this embodiment, the terminal generates a gray-level feature mean for each Mark point based on the gray-level features of the gray-level feature set. Then calculate the grayscale features of each candidate point. With grayscale feature mean The degree of consistency, specifically, can be achieved by averaging the grayscale features of N values ​​(grayscale features, such as N template matching similarity scores) in a grayscale feature item for a Mark point. Calculate, and then sum the N grayscale features with the mean of the grayscale features respectively. Perform grayscale consistency The calculation is repeated to determine the grayscale consistency parameter for each grayscale feature. Then collect the grayscale consistency parameters of all Mark points. The data is integrated to generate a set of grayscale consistency parameters. The consistency function is designed so that a larger value indicates better consistency. The consistency function is as follows:

[0072] in For all N candidate points, grayscale features The arithmetic mean.

[0073] Please see Figure 7 This application provides an embodiment of a method for calculating a multi-dimensional confidence index, comprising: 701. Determine the maximum deviation based on the set of positional deviations; 702. Calculate the normalized position deviation for each candidate coordinate point in the candidate coordinate point set based on the maximum deviation, the position normalization function, and the set of position deviations, and generate a set of normalized position deviations. 703. Calculate the multi-dimensional confidence index for each candidate coordinate point in the candidate coordinate point set by weighting the normalized position deviation set and the gray-scale consistency parameter set, and generate a multi-dimensional confidence index set.

[0074] The terminal first determines the maximum deviation based on the set of position deviations, that is, it determines the position deviation of all candidate points. maximum value Based on the maximum deviation, the position normalization function, and the set of position deviations, the normalized position deviation is calculated for each candidate coordinate point in the candidate coordinate point set, generating a set of normalized position deviations. The position normalization function is shown below:

[0075] Position deviation The normalization function is used to normalize the data. It is mapped to an interval of [0, 1], and the larger the value, the more reliable the location.

[0076] Next, the terminal calculates a multi-dimensional confidence index for each candidate coordinate point in the candidate coordinate point set based on the normalized position deviation set and the grayscale consistency parameter set, generating a multi-dimensional confidence index set. Specifically, the terminal calculates a comprehensive confidence score for each candidate point table. This score represents the positional deviation. and grayscale consistency Weighted sum:

[0077] Here, α and β are weighting coefficients, and they satisfy α + β = 1. The weighting coefficients can be adjusted according to the actual application scenario. For example, if higher positional stability is required, α can be increased; if higher image quality consistency is required, β can be increased.

[0078] Please see Figure 8 This application provides an embodiment of a method for generating the coordinates of a target Mark point, comprising: 801. Sort each candidate coordinate point according to the multi-dimensional confidence index set, and select the candidate coordinate points that meet the preset screening conditions as the reliable coordinate point set; 802. Generate the target coordinates of the Mark point by taking a weighted average of the reliable coordinate point set and the corresponding index in the multi-dimensional confidence index set. The formula for generating the target coordinates of the Mark point is as follows:

[0079] Where K is the number of candidate coordinate points, For reliable coordinate point coordinates, Let be the multi-dimensional confidence index of the i-th candidate coordinate point. The target coordinates are given.

[0080] In this embodiment, the terminal sorts the N candidate coordinate points corresponding to each Mark point according to the multi-dimensional confidence index set. In this embodiment, descending order is used to filter out the candidate coordinate points that meet the preset filtering conditions and use them as a reliable coordinate point set. Specifically, the top candidate coordinate points can be filtered out to generate a reliable coordinate point set, and the set is then calculated based on the confidence score. Sort all candidate points in N frames in descending order. Select the top K (K≤N) candidate points as the high-reliability coordinate point set. The parameter K can be fixed as the top percentage (e.g., top 50%), or it can be set to all points with scores greater than a certain threshold.

[0081] Finally, the terminal calculates the target Mark point coordinates for each Mark point based on the reliability coordinate point set and the multi-dimensional confidence index set, i.e., it performs weighted fusion positioning. Specifically, the terminal calculates the precise coordinates of the final target Mark point based on the selected high-reliability coordinate point set. A weighted average is performed using confidence scores as weights; points with higher scores contribute more to the final result. The formula is as follows:

[0082] This method can perform multi-dimensional evaluation and analysis of candidate coordinate points on each frame of captured image, and select sufficiently credible candidate coordinate points from the dimensions of position and grayscale features. Finally, these candidate coordinate points are weighted and fused based on multi-dimensional confidence index, which further increases the accuracy of target coordinates.

[0083] Please see Figure 9 This application provides an embodiment of a method for processing captured images, comprising: 901. Perform median filtering noise reduction on the captured images in the captured image sequence.

[0084] In this embodiment, after acquiring multiple frames of captured images, the terminal needs to perform median filtering and noise reduction processing on the captured images to reduce some interference before positioning and improve the accuracy of subsequent positioning.

[0085] Please see Figure 10 This application provides an embodiment of a Mark point coordinate detection device based on image sequences, comprising: The acquisition unit 1001 acquires images of the object to be aligned with Mark points for N consecutive frames, generating a sequence of captured images, wherein the points in each captured image corresponding to the Mark points are regarded as candidate points. The candidate coordinate point set generation unit 1002 is used to extract the Mark point coordinates of each captured image in the captured image sequence and generate a candidate coordinate point set. The grayscale feature set generation unit 1003 is used to extract grayscale features from the captured image based on each candidate coordinate point and generate a grayscale feature set. The position deviation set generation unit 1004 is used to calculate the position deviation for each candidate coordinate point based on the candidate coordinate point set and generate a position deviation set. The grayscale consistency parameter set generation unit 1005 is used to calculate the grayscale consistency parameter for each candidate coordinate point based on the grayscale feature set and generate the grayscale consistency parameter set. The multi-dimensional confidence index generation unit 1006 is used to calculate the multi-dimensional confidence index for each candidate coordinate point in the candidate coordinate point set based on the set of position deviation and the set of grayscale consistency parameters, and generate a multi-dimensional confidence index set. The target coordinate generation unit 1007 is used to filter the candidate coordinate point set according to the multi-dimensional confidence index set, and use the filtered candidate coordinate point subset to generate target coordinates for the Mark point.

[0086] Optionally, the position deviation set generation unit 1004 is specifically used to include: Generate a cluster center coordinate based on the coordinates of all candidate coordinate points in the candidate coordinate point set; The positional deviation is calculated for candidate coordinate points in each frame of the captured image based on the coordinates of the cluster center, and a set of positional deviations is generated. The formula is as follows:

[0087] in, The coordinates of the cluster center Let i be the i-th candidate coordinate point. Let be the positional deviation of the i-th candidate coordinate point.

[0088] Optionally, the grayscale features in the grayscale feature set may include the image signal-to-noise ratio; The grayscale feature set generation unit 1003 is specifically used for including: For each frame of captured image, an image signal-to-noise ratio (SNR) is generated from the effective image signal and noise signal. The SNR is then correlated with the corresponding candidate coordinate points to generate a grayscale feature set.

[0089] Optionally, the grayscale features in the grayscale feature set include template matching similarity scores; The grayscale feature set generation unit 1003 is specifically used for including: A local feature acquisition region is generated based on the morphological parameters of each candidate coordinate point; Based on the gray values ​​of pixels in the local feature acquisition area and the gray values ​​of corresponding pixels in the template image, a template matching similarity score is generated for each candidate coordinate point.

[0090] Optionally, the gray-level features in the gray-level feature set may include local region contrast or image gray-level gradient intensity; The grayscale feature set generation unit 1003 is specifically used for including: A local feature acquisition region is generated based on the morphological parameters of each candidate coordinate point; The local region contrast or gray-level gradient intensity is generated based on the gray-level values ​​in the local feature acquisition area; the gray-level gradient intensity includes horizontal gray-level gradient intensity and vertical gray-level gradient intensity.

[0091] Optionally, the grayscale consistency parameter set generation unit 1005 is specifically used to include: Generate a gray-level feature mean based on the gray-level feature set; Based on the grayscale feature mean and the consistency function, grayscale consistency parameters are calculated for the corresponding candidate coordinate points in each frame of the captured image, generating a set of grayscale consistency parameters. The consistency function is as follows:

[0092] in, Let be the arithmetic mean of the grayscale features of all N candidate coordinate points. It refers to the grayscale features corresponding to the candidate coordinate points in the i-th captured image.

[0093] Optionally, the multi-dimensional confidence index generation unit 1006 is specifically used for including: Determine the maximum deviation based on the set of positional deviations; Calculate the normalized position deviation for each candidate coordinate point in the candidate coordinate point set based on the maximum deviation, the position normalization function, and the set of position deviations, and generate the normalized position deviation set. By performing a weighted sum based on the normalized position deviation set and the grayscale consistency parameter set, a multi-dimensional confidence index is calculated for each candidate coordinate point in the candidate coordinate point set, generating a multi-dimensional confidence index set.

[0094] Optionally, the target coordinate generation unit 1007 is specifically used to include: Each candidate coordinate point is sorted according to the multi-dimensional confidence index set, and candidate coordinate points that meet the preset screening conditions are selected as the reliable coordinate point set. The target coordinates of the Mark point are generated by taking a weighted average of the reliable coordinate point set and the corresponding index in the multi-dimensional confidence index set. The formula for generating the target coordinates of the Mark point is as follows:

[0095] Where K is the number of candidate coordinate points, For reliable coordinate point coordinates, Let be the multi-dimensional confidence index of the i-th candidate coordinate point. The target coordinates are given.

[0096] Optionally, the Mark point coordinate detection device also includes: The median filtering noise reduction processing unit is used to perform median filtering noise reduction processing on the captured images in the captured image sequence generated by the acquisition unit.

[0097] 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.

[0098] 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.

[0099] 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.

[0100] 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.

[0101] 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 for detecting Mark point coordinates based on image sequences, characterized in that, include: The object to be aligned with the Mark point is subjected to continuous image acquisition of N frames to generate a sequence of captured images, wherein the point in each captured image corresponding to the Mark point is regarded as a candidate point; Extract the coordinates of the Mark point in each captured image from the captured image sequence to generate a set of candidate coordinate points; Based on each candidate coordinate point, grayscale features are extracted from the captured image to generate a grayscale feature set. Calculate the position deviation for each candidate coordinate point based on the set of candidate coordinate points, and generate a set of position deviations. Calculate the grayscale consistency parameter for each candidate coordinate point based on the grayscale feature set, and generate a grayscale consistency parameter set; Calculate a multi-dimensional confidence index for each candidate coordinate point in the candidate coordinate point set based on the set of positional deviations and the set of grayscale consistency parameters, and generate a multi-dimensional confidence index set. The candidate coordinate point set is filtered based on the multi-dimensional confidence index set, and the target coordinates of the Mark point are generated using the filtered subset of candidate coordinate points.

2. The Mark point coordinate detection method according to claim 1, characterized in that, The step of calculating the position deviation for each candidate coordinate point based on the candidate coordinate point set and generating a set of position deviations specifically includes: Generate a cluster center coordinate based on the coordinates of all candidate coordinate points in the candidate coordinate point set; Based on the cluster center coordinates, the positional deviation is calculated for candidate coordinate points in each frame of the captured image, and the set of positional deviations is generated, as shown in the following formula: in, The coordinates of the cluster center Let i be the i-th candidate coordinate point. Let be the positional deviation of the i-th candidate coordinate point.

3. The Mark point coordinate detection method according to claim 2, characterized in that, The grayscale features in the grayscale feature set include the image signal-to-noise ratio; the step of extracting grayscale features from the captured image based on each candidate coordinate point to generate the grayscale feature set specifically includes: For each frame of the captured image, an image signal-to-noise ratio is generated from the effective image signal and the noise signal, and the image signal-to-noise ratio is associated with the corresponding candidate coordinate points to generate a grayscale feature set.

4. The Mark point coordinate detection method according to claim 2, characterized in that, The grayscale features in the grayscale feature set include template matching similarity scores; The step of extracting grayscale features from the captured image based on each candidate coordinate point to generate a grayscale feature set specifically includes: A local feature acquisition region is generated based on the morphological parameters of each candidate coordinate point; Based on the gray values ​​of pixels in the local feature acquisition area and the gray values ​​of corresponding pixels in the template image, a template matching similarity score is generated for each candidate coordinate point.

5. The Mark point coordinate detection method according to claim 2, characterized in that, The gray-level features in the gray-level feature set include local region contrast or image gray-level gradient intensity; the step of extracting gray-level features from the captured image based on each candidate coordinate point to generate the gray-level feature set specifically includes: A local feature acquisition region is generated based on the morphological parameters of each candidate coordinate point; The local region contrast or gray-level gradient intensity is generated based on the gray-level values ​​in the local feature acquisition area; the gray-level gradient intensity includes horizontal gray-level gradient intensity and vertical gray-level gradient intensity.

6. The Mark point coordinate detection method according to any one of claims 1 to 5, characterized in that, The step of calculating a gray-level consistency parameter for each candidate coordinate point based on the gray-level feature set and generating a gray-level consistency parameter set specifically includes: Generate a gray-scale feature mean value based on the gray-scale feature set; Based on the grayscale feature mean and consistency function, grayscale consistency parameters are calculated for the corresponding candidate coordinate points in each frame of the captured image, generating a set of grayscale consistency parameters. The consistency function is as follows: in, Let be the arithmetic mean of the grayscale features of all N candidate coordinate points. It refers to the grayscale features corresponding to the candidate coordinate points in the i-th captured image.

7. The Mark point coordinate detection method according to any one of claims 1 to 5, characterized in that, The steps of calculating a multi-dimensional confidence index for each candidate coordinate point in the candidate coordinate point set based on the set of positional deviations and the set of grayscale consistency parameters, and generating a multi-dimensional confidence index set, specifically include: The maximum deviation is determined based on the set of positional deviations. Based on the maximum deviation, the position normalization function, and the set of position deviations, a normalized position deviation is calculated for each candidate coordinate point in the candidate coordinate point set, generating a set of normalized position deviations. A weighted sum is performed based on the normalized position deviation set and the grayscale consistency parameter set to calculate a multi-dimensional confidence index for each candidate coordinate point in the candidate coordinate point set, thereby generating a multi-dimensional confidence index set.

8. The Mark point coordinate detection method according to any one of claims 1 to 5, characterized in that, The step of filtering the candidate coordinate point set based on the multi-dimensional confidence index set and generating target coordinates for the Mark point using the filtered subset of candidate coordinate points specifically includes: Each candidate coordinate point is sorted according to the set of multi-dimensional confidence indices, and candidate coordinate points that meet the preset screening conditions are selected as a reliable coordinate point set. The target coordinates of the Mark point are generated by taking a weighted average of the reliable coordinate point set and the corresponding index in the multi-dimensional confidence index set. The formula for generating the target coordinates of the Mark point is as follows: Where K is the number of candidate coordinate points, For reliable coordinate point coordinates, Let be the multi-dimensional confidence index of the i-th candidate coordinate point. The target coordinates are given.

9. The Mark point coordinate detection method according to any one of claims 1 to 5, characterized in that, After the step of acquiring N consecutive frames of images of the object to be aligned with Mark points to generate an image sequence, and before the step of extracting the Mark point coordinates of each image in the image sequence to generate a candidate coordinate point set, the Mark point coordinate detection method further includes: Median filtering is applied to the captured images in the captured image sequence to reduce noise.

10. A Mark point coordinate detection device based on image sequences, characterized in that, include: The acquisition unit acquires N consecutive frames of images of the object to be aligned, which has a Mark point, and generates a sequence of captured images. In each captured image, the point in the image corresponding to the Mark point is regarded as a candidate point. The candidate coordinate point set generation unit is used to extract the Mark point coordinates of each captured image in the captured image sequence and generate a candidate coordinate point set. The grayscale feature set generation unit is used to extract grayscale features from the captured image for each candidate coordinate point and generate a grayscale feature set. The position deviation set generation unit is used to calculate the position deviation for each candidate coordinate point based on the candidate coordinate point set, and generate a position deviation set. The grayscale consistency parameter set generation unit is used to calculate the grayscale consistency parameter for each candidate coordinate point based on the grayscale feature set, and generate the grayscale consistency parameter set. A multi-dimensional confidence index generation unit is used to calculate a multi-dimensional confidence index for each candidate coordinate point in the candidate coordinate point set based on the set of position deviations and the set of grayscale consistency parameters, and to generate a multi-dimensional confidence index set. The target coordinate generation unit is used to filter the candidate coordinate point set according to the multi-dimensional confidence index set, and generate target coordinates for the Mark point using the filtered candidate coordinate point subset.

Images