Intelligent measurement method and system based on multi-view image acquisition analysis

Abstract

The application provides an intelligent measurement method and system based on multi-view image acquisition analysis, relates to the technical field of image processing, and comprises the following steps: obtaining a standard three-dimensional model of a target to be measured; performing two-dimensional image slicing on the standard three-dimensional model to obtain monitoring angle target size information; constructing an optimization balance formula according to the minimum detectable expected size, size tolerance information and monitoring angle target size information; optimizing camera pixels, light source layout types and illumination control parameters, and when the optimization balance formula is satisfied, outputting recommended camera models, recommended light source layout types and recommended illumination control parameters, synchronously positioning the target to be measured, performing image acquisition through the optimization result, constructing a comparison three-dimensional model; comparing the comparison three-dimensional model with the standard three-dimensional model to obtain an identification area that does not satisfy the size tolerance information and send the identification area to a user end. The technical problem that the image acquisition configuration is manually configured and has weak migration in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, specifically to an intelligent measurement method and system based on multi-view image acquisition and analysis. Background Technology

[0002] Using machine vision to detect workpiece dimensional deviations is a widely used application. In order to achieve accurate recognition in machine vision software, the image acquisition process in hardware is particularly important. To detect images with minimal distortion and appropriate accuracy, it is necessary to select suitable image acquisition equipment and determine appropriate image acquisition parameters. Currently, the above process is usually customized by image acquisition managers, which has the disadvantage of relying on the personal professional experience of the managers and having weak transferability.

[0003] In summary, existing technologies suffer from weak portability due to the manual configuration of image acquisition settings. Summary of the Invention

[0004] This application provides an intelligent measurement method and system based on multi-view image acquisition and analysis, aiming to solve the technical problem in the prior art where the configuration of image acquisition is manually configured, resulting in weak transferability.

[0005] In view of the above problems, this application provides an intelligent measurement method and system based on multi-view image acquisition and analysis.

[0006] The first aspect disclosed in this application provides an intelligent measurement method based on multi-view image acquisition and analysis, comprising: obtaining a standard three-dimensional model of the target to be measured, wherein the standard three-dimensional model has a digital positioning posture label, an image acquisition angle label, a minimum detectable expected size, and size tolerance information; performing two-dimensional image slicing on the standard three-dimensional model using the image acquisition angle label to obtain target size information from a monitoring perspective; constructing an optimized balance formula based on the minimum detectable expected size, the size tolerance information, and the target size information from the monitoring perspective; optimizing camera pixels, light source layout type, and lighting control parameters, and when the optimized balance formula is satisfied, outputting a recommended camera model, a recommended light source layout type, and recommended lighting control parameters, wherein camera pixels refer to the number of pixels on the longer side of the image; synchronously positioning the target to be measured based on the digital positioning posture, acquiring images using the recommended camera model, the recommended light source layout type, and the recommended lighting control parameters, and constructing a comparative three-dimensional model; performing comparative analysis based on the comparative three-dimensional model and the standard three-dimensional model, obtaining identified areas that do not meet the size tolerance information, and sending them to the user terminal.

[0007] Another aspect of this application discloses an intelligent measurement system based on multi-view image acquisition and analysis, comprising: a standard model acquisition module for acquiring a standard three-dimensional model of the target to be measured, wherein the standard three-dimensional model has digital positioning attitude labels, image acquisition angle labels, minimum detectable expected size, and size tolerance information; a two-dimensional image slicing module for performing two-dimensional image slicing on the standard three-dimensional model using the image acquisition angle labels to obtain target size information from the monitoring viewpoint; an optimized balance construction module for constructing an optimized balance based on the minimum detectable expected size, the size tolerance information, and the target size information from the monitoring viewpoint; and a parameter optimization module. The system is used to optimize camera pixels, light source layout type, and lighting control parameters. When the optimization balance is satisfied, it outputs a recommended camera model, recommended light source layout type, and recommended lighting control parameters. The camera pixels refer to the number of pixels on the longer side of the image. The comparison model construction module is used to synchronously locate the target to be measured based on the digital positioning posture, and to acquire images using the recommended camera model, recommended light source layout type, and recommended lighting control parameters to construct a comparison 3D model. The comparison analysis module is used to perform comparison analysis based on the comparison 3D model and the standard 3D model, and to obtain the marked areas that do not meet the dimensional tolerance information and send them to the user terminal.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] Because it employs a standard 3D model of the target to be measured, i.e., a 3D model of the target's conceptual model in a preset fixed state, which has digital positioning attitude labels, image acquisition angle labels, minimum detectable expected size, and size tolerance information; based on the image acquisition angle labels, the standard 3D model can be sliced ​​into 2D images to obtain the target size information from the monitoring perspective; then, based on the minimum detectable expected size, the size tolerance information, and the target size information from the monitoring perspective, an optimized balance formula is constructed; furthermore, based on the optimized balance formula, camera pixels, light source layout type, and lighting control parameters are optimized to obtain recommended camera models, recommended light source layout types, and recommended lighting control parameters, and then a real-time acquisition comparison 3D model is constructed and compared with the standard 3D model to obtain the identified areas that do not meet the size tolerance information and send them to the user terminal. By constructing an optimized balance formula for hardware parameters based on the basic state information of the target to be measured, and using the optimized balance formula to achieve optimized configuration of camera pixels, light source layout type, and lighting control parameters, the automatic selection of camera type and the determination of acquisition control parameter technology density are achieved, thereby improving the technical effect of transferability.

[0010] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0011] Figure 1 This application provides a possible flowchart of an intelligent measurement method based on multi-view image acquisition and analysis for embodiments of the present application;

[0012] Figure 2 This application provides a schematic diagram of a possible two-dimensional image slicing process in an intelligent measurement method based on multi-view image acquisition and analysis.

[0013] Figure 3 This application provides a schematic diagram of a possible structure for an intelligent measurement system based on multi-view image acquisition and analysis.

[0014] Explanation of reference numerals in the attached figures:

[0015] The module includes: Standard Model Acquisition Module 100, Two-Dimensional Image Slicing Module 200, Optimized Balanced Construction Module 300, Parameter Optimization Module 400, Comparison Model Construction Module 500, and Comparison Analysis Module 600. Detailed Implementation Example 1

[0016] like Figure 1 As shown, embodiments of this application provide an intelligent measurement method based on multi-view image acquisition and analysis, including:

[0017] A standard three-dimensional model of the target to be measured is obtained, wherein the standard three-dimensional model has digital positioning attitude labels, image acquisition angle labels, minimum detectable expected size and dimensional tolerance information.

[0018] In this embodiment, the target to be measured is a workpiece whose dimensional deviation needs to be measured. During visual inspection of the dimensional deviation, the workpiece needs to be fixed on a machine tool in a specified posture. Therefore, a digital twin machine tool is constructed based on the machine tool used to fix the target, and then a simulated three-dimensional workpiece model is constructed based on the conceptual model of the target. Finally, the simulated three-dimensional workpiece model is fixed on the digital twin machine tool in a specified posture to obtain a standard three-dimensional model. That is, the standard three-dimensional model represents the ideal dimensional state of the target under the specified fixed posture. Further, the digital positioning posture label refers to the specified fixed posture; the image acquisition angle label refers to the preset angle data for visual inspection, which includes multiple angles, at least six directions: front / back, left / right, and up / down; the minimum detectable expected size refers to the accuracy level of the visual inspection, and the unit can be cm, mm, nm, etc., without limitation, and can be adjusted by the user according to the actual situation; the dimensional tolerance information refers to the maximum allowable dimensional deviation, and the unit can be cm, mm, nm, etc., without limitation, and can be adjusted by the user according to the actual situation. By using digital positioning posture tags and image acquisition angle tags, parameters can be provided for the actual fixation of the workpiece and the image acquisition angle in subsequent steps. The expected accuracy and allowable deviation of the size detection can be determined by the minimum detectable expected size and size tolerance information, which facilitates the subsequent construction of an optimized balance system.

[0019] By using image acquisition angle labels, two-dimensional image slices are performed on the standard three-dimensional model to obtain target size information from the monitoring perspective.

[0020] In this embodiment, a two-dimensional image slice refers to a two-dimensional projection image of a standard three-dimensional model determined from an image acquisition angle label. By matching the size data of the standard three-dimensional model to the two-dimensional projection image, the size information of the two-dimensional projection image can be determined. Preferably, the two-dimensional projection image is generally a polygon, and its maximum side length is taken as the target size information of the monitoring viewpoint. Using the maximum side length as the target size information of the monitoring viewpoint can determine the integrity of the image data acquired from the image acquisition angle label and ensure the accuracy of subsequent detection.

[0021] An optimized balance formula is constructed based on the minimum detectable expected size, the size tolerance information, and the target size information of the monitoring viewpoint.

[0022] Furthermore, based on the minimum detectable expected size, the size tolerance information, and the target size information of the monitoring viewpoint, an optimized balance formula is constructed, including:

[0023] The first desired accuracy is calculated by applying the first constraint multiple between the minimum detectable expected size and the detection accuracy.

[0024] The second desired accuracy is calculated by loading the dimensional tolerance information and the second constraint multiple of the detection accuracy.

[0025] Extract the maximum value of the first desired precision and the second desired precision, and set it as the desired precision;

[0026] The ratio of the desired accuracy to the number of effective pixels is set as the first balancing parameter.

[0027] The ratio of the target size information from the monitoring perspective to the camera pixels is set as the second balancing parameter.

[0028] The optimized balance formula is constructed by setting the first balance parameter equal to the second balance parameter.

[0029] In the embodiments of this application, the optimized balance equation refers to the constraint equation used to constrain the parameters of the image acquisition device and the image acquisition control parameters.

[0030] Formula 1: Resolution = Field of view / Camera pixels. For example, the field of view is the minimum field of view for monitoring the size of the target in order to ensure the integrity of the acquired image. The camera pixels are the hardware parameters of the camera. For example, the hardware parameters are 1800 pixels * 1000 pixels, where 1800 pixels is the camera pixel. That is, the larger pixel side is selected as the number of pixels.

[0031] Furthermore, we retrieve Equation 2: Accuracy = Resolution * Number of Effective Pixels. The number of effective pixels refers to the number of pixels that can be stably distinguished after the acquired image is magnified by a preset factor, which determines the accuracy. Substituting Equation 1 into Equation 2, we obtain Equation 3: Accuracy = (Target Size Information at Monitoring Viewpoint / Camera Pixels) * Number of Effective Pixels. Preferably, the number of effective pixels can be determined based on historical data of the selected camera model.

[0032] Furthermore, we retrieve Equation 4: Minimum detectable expected size = First constraint multiple * Accuracy; and Equation 5: Dimensional tolerance information = Second constraint multiple * Accuracy. Here, both the first and second constraint multiples are user-preset multiplier factors that can be adjusted according to the actual scenario. A larger constraint multiple results in a more stringent accuracy level, but also a greater computational load. Based on Equations 4 and 5, and the aforementioned minimum detectable expected size and dimensional tolerance information, we can calculate the first and second expected accuracies. Then, we select the maximum value of the first and second expected accuracies as the expected accuracy. The maximum accuracy value actually refers to the minimum size. For example, if the first expected accuracy is 0.1mm and the second expected accuracy is 0.07mm, then the second expected accuracy is the maximum accuracy value, but the size is smaller than the first expected accuracy.

[0033] Furthermore, after determining the desired accuracy, substituting into Equation 3: Accuracy = (Target Size Information at Monitoring Viewpoint / Camera Pixels) * Number of Effective Pixels, we obtain Equation 6: Desired Accuracy / Number of Effective Pixels = Target Size Information at Monitoring Viewpoint / Camera Pixels. Here, both the desired accuracy and the target size information at the monitoring viewpoint are known quantities. The number of effective pixels determines the constraints on the image acquisition control parameters, and the camera pixels serve as the camera's constraint. The desired accuracy / number of effective pixels is the first balancing parameter, and the target size information at the monitoring viewpoint / camera pixels is the second balancing parameter. Therefore, Equation 6 is an optimized balancing equation. When the selected camera, the number of effective pixels in the camera control parameters, and the camera pixels satisfy this equation, and the acquired image accuracy meets the desired accuracy (i.e., image accuracy is greater than or equal to the desired accuracy), and the camera's field of view meets the target size information at the monitoring viewpoint (i.e., camera field of view is greater than or equal to the target size information at the monitoring viewpoint), then it is considered to meet the expectation.

[0034] The system optimizes camera pixels, light source layout type, and lighting control parameters. When the optimization balance is satisfied, it outputs a recommended camera model, a recommended light source layout type, and recommended lighting control parameters. The camera pixels refer to the number of pixels on the longer side of the image.

[0035] In this embodiment, different camera models have different camera pixel values; the light source layout types are various, including low-angle linear arrays, area light sources, backlights, coaxial lights, bright-field illumination, ring lights, and strip lights, all of which are conventional lighting methods used in the prior art. The lighting control parameters refer to the specific control parameters of the camera, such as focal length, and the distribution range of control parameters varies for each camera. Therefore, an optimal solution can be randomly constructed for the above three parameters. Any solution includes a selected camera model, a selected light source layout type, and a selected lighting control parameter. Then, it is determined whether the solution satisfies the optimization balance formula. That is, when the effective number of pixels of the selected camera and camera control parameters and the camera pixels satisfy the optimization balance formula, the acquired image accuracy card meets the expected accuracy, and the camera field of view meets the target size information of the monitoring angle, the output is a recommended camera model, a recommended light source layout type, and recommended lighting control parameters. The camera pixels refer to the number of pixels on the longer side of the image. Through the optimization balance formula, the existing camera model, light source layout type, and lighting control can be automatically optimized, achieving the technical objective of ensuring image quality and improving the transferability of image acquisition.

[0036] Based on the digital positioning attitude, the target to be measured is synchronously positioned, and images are acquired using the recommended camera model, the recommended light source layout type, and the recommended lighting control parameters to construct a comparative 3D model.

[0037] Based on the comparison analysis of the 3D model and the standard 3D model, the identified areas that do not meet the dimensional tolerance information are sent to the user terminal.

[0038] In this embodiment, given the recommended camera model, recommended light source layout type, and recommended lighting control parameters, the target to be measured can be synchronously positioned on the actual machine tool according to the digital positioning posture. Then, image data corresponding to the recommended camera model, recommended light source layout type, and recommended lighting control parameters are acquired from image acquisition angle labels. This same processing is repeated at multiple angles to construct a comparative 3D model. Furthermore, the comparative 3D model and the standard 3D model are distributed in a unified coordinate system with the machine tool as the reference object, identifying the areas where the dimensional deviations do not meet the dimensional tolerance information—that is, areas where the dimensional deviations are greater than the dimensional tolerance information. These identified areas are marked on the comparative 3D model and sent to the user terminal for visualization, facilitating subsequent adjustments by the user.

[0039] Furthermore, such as Figure 2 As shown, by using image acquisition angle labels, two-dimensional image slicing is performed on the standard three-dimensional model to obtain target size information from the monitoring viewpoint, including:

[0040] Construct a virtual plane that is perpendicular to the image acquisition angle label.

[0041] The projection of the standard 3D model onto the virtual plane is captured and set as a 2D image from the monitoring perspective.

[0042] Size analysis is performed on the two-dimensional image from the monitoring perspective to generate target size information from the monitoring perspective.

[0043] In this embodiment, the virtual plane is a plane perpendicular to the image acquisition angle label. The projection of the standard three-dimensional model onto the virtual plane is captured and set as a two-dimensional image of the monitoring viewpoint. Then, the length of the longest side of the two-dimensional image of the monitoring viewpoint is analyzed and set as the target size information of the monitoring viewpoint.

[0044] Furthermore, based on the two-dimensional image from the monitoring perspective, size analysis is performed to generate the target size information from the monitoring perspective, including:

[0045] The two-dimensional images from the monitoring perspective are deployed in a two-dimensional coordinate system to obtain a two-dimensional image coordinate set. Based on the two-dimensional image coordinate set, the fitting starting point is extracted:

[0046] Based on the two-dimensional image coordinate set, extract the point with the minimum vertical coordinate.

[0047] When there are multiple minimum points in the vertical coordinate, the minimum point in the horizontal coordinate of the minimum point in the vertical coordinate is extracted and set as the starting point of the fitting.

[0048] When there is only one minimum point of the ordinate, the minimum point of the ordinate is set as the starting point of the fitting.

[0049] Based on the fitted starting point, lines are drawn to the two-dimensional image coordinate set, and the angle between the line and the positive x-axis is calculated to obtain the deviation angle set. The distance between the lines is also calculated to obtain the deviation distance set.

[0050] The two-dimensional image coordinate set is sorted from largest to smallest according to the set of deviation angles. When the deviation angles are the same, the two-dimensional image coordinate set is sorted from smallest to largest according to the set of deviation distances, generating a two-dimensional image coordinate sorting result.

[0051] Based on the sorting results of the two-dimensional image coordinates, the minimum convex polygon is obtained by fitting the minimum convex polygon using the convex hull algorithm.

[0052] Extract the maximum side length of the minimum rectangle covering the minimum convex polygon, and set it as the target size information of the monitoring viewpoint.

[0053] Specifically, when the 2D image from the monitoring perspective is an irregular, complex shape, it is difficult to directly determine the longest side length. It is necessary to fit the circumscribed rectangle of this complex shape and extract the maximum side length of the circumscribed rectangle as the target size information from the monitoring perspective to ensure the image's field of view. Preferably, the process for determining the circumscribed rectangle is as follows:

[0054] To determine the bounding rectangle, we first need to determine the smallest convex polygon of the complex shape image, as follows:

[0055] The two-dimensional image from the monitoring perspective is deployed in a two-dimensional coordinate system to extract a two-dimensional image coordinate set. Then, a fitting starting point is selected from the two-dimensional image coordinate set; that is, the minimum point of the vertical coordinate is extracted based on the two-dimensional image coordinate set. When there are multiple minimum points of the vertical coordinate, the minimum point of the horizontal coordinate of the minimum point of the vertical coordinate is extracted and set as the fitting starting point. When there is only one minimum point of the vertical coordinate, the minimum point of the vertical coordinate is set as the fitting starting point. Based on the fitting starting point, a line is drawn connecting it to each coordinate in the two-dimensional image coordinate set, and the angle between the line and the positive horizontal axis is calculated and stored as a set of deviation angles. The Euclidean distance of each line is also calculated and stored as a set of deviation distances. The two-dimensional image coordinate set is sorted from largest to smallest based on the set of deviation angles. When the deviation angles are the same, the two-dimensional image coordinate set is sorted from smallest to largest based on the set of deviation distances, generating a sorted result for the two-dimensional image coordinates. Based on the sorted result of the two-dimensional image coordinates, a minimum convex polygon is fitted using a convex hull algorithm to obtain the minimum convex polygon.

[0056] Preferably, when determining the minimum convex polygon using the convex hull algorithm, the first coordinate of the fitted starting point coordinates and the sorted result of the two-dimensional image coordinates is used. First, the first and second corner points of the minimum convex polygon are assumed. Then, the sorted result of the two-dimensional image coordinates is traversed to determine whether each point belongs to a corner point of the convex polygon. Details are as follows:

[0057] Assuming the last corner of a convex polygon is Q, and the second-to-last corner is R, the new point to be determined is P. If the cross product of vectors QP and RQ is positive, then point P will be considered part of the convex hull because it does not violate the convexity of the convex hull. In this case, point P is set as the last corner of the convex polygon, and Q is set as the second-to-last corner, and the process continues to traverse other points. If the cross product is negative, point Q needs to be removed from the convex hull, i.e., the second-to-last and last corners are updated, and then point P is re-analyzed. When all known corners have been considered, and the cross product results are all negative, and point P is still not a corner, it means that point P is not on the convex hull. Based on the original last corner being Q and the second-to-last corner being R, subsequent coordinate points are considered. When any cross product result is positive, the removed corner is permanently removed, point P is set as the last corner, and the process continues to traverse other points. When the entire 2D image coordinate set has been analyzed, a sequence of corner points is obtained. Connecting these points sequentially yields the smallest convex polygon.

[0058] Extract the smallest rectangle covering the smallest convex polygon, i.e., the maximum side length of the circumscribed rectangle, and set it as the target size information of the monitoring viewpoint to ensure that the image field of view meets the detection requirements.

[0059] Furthermore, the camera pixel count, light source layout type, and illumination control parameters are optimized. When the optimized balance is satisfied, a recommended camera model, recommended light source layout type, and recommended illumination control parameters are output, including:

[0060] Obtain the set of camera pixel counts.

[0061] Obtain the set of light source layout types.

[0062] Obtain the constraint range of lighting control parameters.

[0063] Based on the set of camera pixel counts, the set of light source layout types, and the constraint range of lighting control parameters, configure the recommended camera model, the recommended light source layout type, and the recommended lighting control parameters that satisfy the optimized balance.

[0064] In this embodiment, the camera pixel count set is a set of pixel counts corresponding to a user-preset camera model. The light source layout type set refers to a set of layout types preset by the user, including but not limited to: low-angle linear array, area light source, backlight, coaxial light, bright field illumination, ring light source, and strip light source. The lighting control parameter constraint range refers to the specific control parameters of the camera, such as the range of values ​​for parameters like shooting distance and focal length. Based on the camera pixel count set, the light source layout type set, and the lighting control parameter constraint range, a solution can be randomly configured. That is, a camera pixel count is randomly selected from the camera pixel count set, a light source layout type is randomly selected from the light source layout type set, and a set of lighting control parameters is randomly determined from the lighting control parameter constraint range based on the camera model corresponding to the camera pixel count, thus generating a solution. When the effective pixel count and camera pixel count of the selected camera and camera control parameters corresponding to the solution satisfy this equation, and the acquired image accuracy meets the expected accuracy (i.e., the image accuracy is greater than or equal to the expected accuracy), and the camera field of view meets the target size information of the monitoring angle (i.e., the camera field of view is greater than or equal to the target size information of the monitoring angle), it is considered to meet the expectation. These are then set as the recommended camera model, the recommended light source layout type, and the recommended lighting control parameters.

[0065] Furthermore, based on the set of camera pixel counts, the set of light source layout types, and the constraint range of lighting control parameters, the recommended camera model, the recommended light source layout type, and the recommended lighting control parameters that satisfy the optimized balance are configured, including:

[0066] The first camera pixel count, the first light source layout type, and the first lighting control parameters are obtained by randomly selecting values ​​from the set of camera pixel counts, the set of light source layout types, and the constraint range of lighting control parameters.

[0067] Historical data is collected based on the number of pixels in the first camera, the layout type of the first light source, and the first lighting control parameters to obtain a valid historical calibration dataset of pixels.

[0068] A central tendency analysis is performed on the historical calibration dataset of the effective pixels to obtain the first effective pixel feature value.

[0069] Substitute the first camera pixel count and the first effective pixel feature value into the optimized balance formula, and determine whether the deviation between the first balance parameter and the second balance parameter is greater than or equal to the deviation threshold.

[0070] If the value is greater than the specified value, update the number of pixels in the first camera, the layout type of the first light source, and the first lighting control parameters.

[0071] If the number of pixels in the first camera, the first light source layout type, and the first lighting control parameter are less than or equal to the recommended camera model, the recommended light source layout type, and the recommended lighting control parameter, respectively.

[0072] In this embodiment, since the optimization balance formula is: Desired accuracy / Number of effective pixels = Target size information of monitoring viewpoint / Camera pixels, and each solution corresponds to a camera model that can be matched one-to-one with the corresponding / camera pixel, the number of effective pixels for each solution needs to be configured.

[0073] Preferably, the algorithm process for configuring the effective pixel count of any solution is as follows:

[0074] By randomly selecting values ​​from the set of camera pixel counts, the set of light source layout types, and the constraint range of lighting control parameters, the first camera pixel count, the first light source layout type, and the first lighting control parameters are obtained. These three parameters constitute a randomly configured solution. Since the light source layout type, camera model, and camera control parameters are determined, historical data with the same light source layout type, camera model, and camera control parameters can be collected using these constraints to determine a valid historical pixel calibration dataset containing multiple valid pixel count records. The steps of central tendency analysis are as follows: Cluster analysis is performed on the valid historical pixel calibration dataset according to the user-preset valid pixel count deviation. Clusters with a number of pixels within a cluster less than the user-preset threshold are deleted. Mean analysis is then performed on the remaining valid historical pixel calibration dataset to obtain the first valid pixel feature value.

[0075] Furthermore, the first camera pixel count and the first effective pixel feature value are substituted into the optimization balancing formula to determine whether the deviation between the first balancing parameter and the second balancing parameter is greater than or equal to a deviation threshold. If it is greater, the balancing formula is not satisfied, and the first camera pixel count, the first light source layout type, and the first lighting control parameter are updated; if it is less than or equal to, the balancing formula is satisfied, and the first camera pixel count, the first light source layout type, and the first lighting control parameter are set as the recommended camera model, the recommended light source layout type, and the recommended lighting control parameter.

[0076] Furthermore, if the number of pixels in the first camera, the first light source layout type, and the first lighting control parameter are less than or equal to the recommended camera model, the recommended light source layout type, and the recommended lighting control parameter, respectively, the following is included:

[0077] If the first camera pixel count, the first light source layout type, and the first lighting control parameters are less than or equal to the first resolution group, the first camera pixel count, the first light source layout type, and the first lighting control parameters are added to the initial resolution group.

[0078] When the initial solution group is greater than or equal to the preset number, the initial solution group is traversed to perform minimum energy consumption sorting to obtain the recommended camera model, the recommended light source layout type and the recommended lighting control parameters.

[0079] In this embodiment, when the first camera pixel count, the first light source layout type, and the first lighting control parameters are less than or equal to a certain value, they are added to the initial solution group. Then, solutions conforming to the balance formula are configured. When the initial solution group is greater than or equal to a preset number, the initial solution group is traversed to perform minimum energy consumption sorting to obtain the recommended camera model, the recommended light source layout type, and the recommended lighting control parameters. Since there may be multiple solutions satisfying the optimization balance formula, selecting the solution with lower energy consumption is beneficial for energy saving.

[0080] In summary, the intelligent measurement method and system based on multi-view image acquisition and analysis provided in this application have the following technical effects:

[0081] 1. This method employs a standard 3D model of the target to be measured, i.e., a 3D model of the target's conceptual model in a preset fixed state. This model contains digital positioning attitude labels, image acquisition angle labels, minimum detectable expected size, and dimensional tolerance information. Based on the image acquisition angle labels, the standard 3D model can be sliced ​​into 2D images to obtain the target size information from the monitoring perspective. Then, based on the minimum detectable expected size, the dimensional tolerance information, and the target size information from the monitoring perspective, an optimized balance is constructed. Furthermore, based on the optimized balance, camera pixels, light source layout type, and lighting control parameters are optimized to obtain recommended camera models, recommended light source layout types, and recommended lighting control parameters. Then, a real-time acquisition comparison 3D model is constructed and compared with the standard 3D model. Identified areas that do not meet the dimensional tolerance information are sent to the user terminal. This technical solution, by constructing an optimized balance of hardware parameters based on the target's basic state information, and using this optimized balance to optimize the configuration of camera pixels, light source layout type, and lighting control parameters, achieves automated selection of camera type and determination of acquisition control parameter technology density, thus improving portability. Example 2

[0082] Based on the same inventive concept as the intelligent measurement method based on multi-view image acquisition and analysis in the foregoing embodiments, such as Figure 3 As shown, this application provides an intelligent measurement system based on multi-view image acquisition and analysis, including:

[0083] The standard model acquisition module 100 is used to acquire a standard three-dimensional model of the target to be measured, wherein the standard three-dimensional model has digital positioning attitude labels, image acquisition angle labels, minimum detectable expected size and size tolerance information;

[0084] The two-dimensional image slicing module 200 is used to slice the standard three-dimensional model into two-dimensional images using image acquisition angle labels to obtain target size information from the monitoring perspective.

[0085] An optimized balance construction module 300 is used to construct an optimized balance based on the minimum detectable expected size, the size tolerance information, and the target size information of the monitoring viewpoint.

[0086] The parameter optimization module 400 is used to optimize camera pixels, light source layout type and lighting control parameters. When the optimization balance is satisfied, it outputs recommended camera model, recommended light source layout type and recommended lighting control parameters. The camera pixels refer to the number of pixels on the longer side of the image.

[0087] The comparison model construction module 500 is used to synchronously locate the target to be measured based on the digital positioning posture, and to construct a comparison 3D model by acquiring images through the recommended camera model, the recommended light source layout type and the recommended lighting control parameters.

[0088] The comparison analysis module 600 is used to perform comparison analysis based on the comparison 3D model and the standard 3D model, and to obtain the identified areas that do not meet the dimensional tolerance information and send them to the user terminal.

[0089] Furthermore, the two-dimensional image slicing module 200 performs the following steps: constructing a virtual plane perpendicular to the image acquisition angle label; cropping the projection of the standard three-dimensional model onto the virtual plane and setting it as a two-dimensional image of the monitoring perspective; and performing size analysis based on the two-dimensional image of the monitoring perspective to generate the target size information of the monitoring perspective.

[0090] Furthermore, the two-dimensional image slicing module 200 performs the following steps: deploying the two-dimensional image from the monitoring perspective onto a two-dimensional coordinate system to obtain a two-dimensional image coordinate set; extracting a fitting starting point based on the two-dimensional image coordinate set; extracting the minimum point of the vertical coordinate based on the two-dimensional image coordinate set; when there are multiple minimum points of the vertical coordinate, extracting the minimum point of the horizontal coordinate of the minimum points of the vertical coordinate and setting it as the fitting starting point; when there is only one minimum point of the vertical coordinate, setting the minimum point of the vertical coordinate as the fitting starting point; and connecting the fitting starting point to the two-dimensional image coordinate set. The process involves calculating the angle between the connecting line and the positive x-axis to obtain a set of deviation angles, and calculating the distance between the connecting lines to obtain a set of deviation distances. The two-dimensional image coordinate set is then sorted from largest to smallest according to the set of deviation angles. When the deviation angles are the same, the two-dimensional image coordinate set is sorted from smallest to largest according to the set of deviation distances to generate a two-dimensional image coordinate sorting result. Based on the two-dimensional image coordinate sorting result, a minimum convex polygon is fitted using a convex hull algorithm to obtain the minimum convex polygon. The maximum side length of the minimum rectangle covering the minimum convex polygon is extracted and set as the target size information of the monitoring viewpoint.

[0091] Furthermore, the optimized balance construction module 300 execution steps include: loading the minimum detectable expected size and the first constraint multiple of the detection accuracy, and calculating the first expected accuracy; loading the size tolerance information and the second constraint multiple of the detection accuracy, and calculating the second expected accuracy; extracting the maximum value of the first expected accuracy and the second expected accuracy, and setting it as the expected accuracy; setting the ratio of the expected accuracy to the number of effective pixels as the first balance parameter; setting the ratio of the target size information of the monitoring viewpoint to the camera pixels as the second balance parameter; and constructing the optimized balance by setting the first balance parameter equal to the second balance parameter.

[0092] Furthermore, the parameter optimization module 400 performs the following steps: obtaining a set of camera pixel counts; obtaining a set of light source layout types; obtaining a range of lighting control parameter constraints; and configuring a recommended camera model, a recommended light source layout type, and recommended lighting control parameters that satisfy the optimization balance based on the set of camera pixel counts, the set of light source layout types, and the range of lighting control parameter constraints.

[0093] Furthermore, the parameter optimization module 400 performs the following steps: randomly selecting values ​​from the set of camera pixel counts, the set of light source layout types, and the constraint range of the lighting control parameters to obtain a first camera pixel count, a first light source layout type, and a first lighting control parameter; collecting historical data based on the first camera pixel count, the first light source layout type, and the first lighting control parameter to obtain a historical calibration dataset of effective pixels; performing central tendency analysis on the historical calibration dataset of effective pixels to obtain a first effective pixel feature value; substituting the first camera pixel count and the first effective pixel feature value into the optimization balance formula to determine whether the deviation between the first balance parameter and the second balance parameter is greater than or equal to a deviation threshold; if greater, updating the first camera pixel count, the first light source layout type, and the first lighting control parameter; if less than or equal to, setting the first camera pixel count, the first light source layout type, and the first lighting control parameter as the recommended camera model, the recommended light source layout type, and the recommended lighting control parameter.

[0094] Furthermore, the parameter optimization module 400 performs the following steps: if the number of first camera pixels, the first light source layout type, and the first lighting control parameters are less than or equal to the number of preset values, the first camera pixel count, the first light source layout type, and the first lighting control parameters are added to the initial solution group; when the initial solution group is greater than or equal to a preset number, the initial solution group is traversed to perform minimum energy consumption sorting to obtain the recommended camera model, the recommended light source layout type, and the recommended lighting control parameters.

[0095] In summary, any step of the method described above can be stored as a computer instruction or program in an unrestricted computer memory, and can be called and identified by an unrestricted computer processor to implement any method in the embodiments of this application, without any additional restrictions.

[0096] Furthermore, the "first" or "second" mentioned above may not only represent a sequential relationship, but may also represent a specific concept, and / or refer to the individual or collective selection of multiple elements. Clearly, those skilled in the art can make various modifications and variations to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. An intelligent measurement method based on multi-view image acquisition analysis, characterized in that, The method comprises the following steps: obtaining a standard three-dimensional model of a target to be measured, wherein the standard three-dimensional model has a digital positioning pose label, an image acquisition angle label, a minimum detectable expected size, and size tolerance information; performing two-dimensional image slicing on the standard three-dimensional model through the image acquisition angle label to obtain monitoring view angle target size information; constructing an optimization balance formula according to the minimum detectable expected size, the size tolerance information, and the monitoring view angle target size information; optimizing a camera pixel, a light source layout type, and an illumination control parameter, and outputting a recommended camera model, a recommended light source layout type, and a recommended illumination control parameter when the optimization balance formula is satisfied, wherein the camera pixel refers to the number of image longer side pixels; synchronously positioning the target to be measured based on the digital positioning pose, performing image acquisition through the recommended camera model, the recommended light source layout type, and the recommended illumination control parameter, and constructing a comparison three-dimensional model; performing comparison analysis based on the comparison three-dimensional model and the standard three-dimensional model, and obtaining an identification area that does not satisfy the size tolerance information and sending the identification area to a user end.

2. The method of claim 1, wherein, The monitoring view angle target size information is obtained by performing two-dimensional image slicing on the standard three-dimensional model through the image acquisition angle label, and the method comprises the following steps: constructing a virtual plane perpendicular to the image acquisition angle label; intercepting a projection of the standard three-dimensional model on the virtual plane and setting the projection as a monitoring view angle two-dimensional image; performing size analysis according to the monitoring view angle two-dimensional image to generate the monitoring view angle target size information.

3. The method of claim 2, wherein, The monitoring view angle target size information is generated by performing size analysis according to the monitoring view angle two-dimensional image, and the method comprises the following steps: deploying the monitoring view angle two-dimensional image on a two-dimensional coordinate system to obtain a two-dimensional image coordinate set; extracting a fitting starting point according to the two-dimensional image coordinate set; extracting a longitudinal coordinate minimum value point according to the two-dimensional image coordinate set; when there are multiple longitudinal coordinate minimum value points, extracting a transverse coordinate minimum value point of the longitudinal coordinate minimum value point as the fitting starting point; when there is only one longitudinal coordinate minimum value point, setting the longitudinal coordinate minimum value point as the fitting starting point; connecting the fitting starting point with the two-dimensional image coordinate set respectively, calculating the included angle between the connection and the transverse coordinate positive axis to obtain an offset angle set, and calculating the distance of the connection to obtain an offset distance set; sorting the two-dimensional image coordinate set from large to small according to the offset angle set, and sorting the two-dimensional image coordinate set from small to large according to the offset distance set when the offset angles are the same, to generate a two-dimensional image coordinate sorting result; performing minimum convex polygon fitting on the two-dimensional image coordinate sorting result through a convex hull algorithm to obtain a minimum convex polygon; extracting the maximum side length of the minimum rectangle covering the minimum convex polygon as the monitoring view angle target size information.

4. The method of claim 1, wherein, The optimization balance formula is constructed according to the minimum detectable expected size, the size tolerance information, and the monitoring view angle target size information, and the method comprises the following steps: loading a first constraint multiple of the minimum detectable expected size and detection accuracy to calculate a first expected accuracy; Load the size tolerance information and the second constraint multiple of detection precision to calculate a second expected precision; Extract the maximum value of the first expected precision and the second expected precision as an expected precision; Set the ratio of the expected precision to the number of effective pixels as a first balance parameter; Set the ratio of the monitoring angle target size information to the camera pixel as a second balance parameter; Construct the optimization balance formula when the first balance parameter is equal to the second balance parameter.

5. The method of claim 4, wherein, Optimize the camera pixel, light source layout type, and illumination control parameter, and output the recommended camera model, recommended light source layout type, and recommended illumination control parameter when the optimization balance formula is met, including: Obtain a camera pixel number set; Obtain a light source layout type set; Obtain an illumination control parameter constraint interval; According to the camera pixel number set, the light source layout type set, and the illumination control parameter constraint interval, configure the recommended camera model, the recommended light source layout type, and the recommended illumination control parameter that meet the optimization balance formula.

6. The method of claim 5, wherein, According to the camera pixel number set, the light source layout type set, and the illumination control parameter constraint interval, configure the recommended camera model, the recommended light source layout type, and the recommended illumination control parameter that meet the optimization balance formula, including: Randomly select a first camera pixel number, a first light source layout type, and a first illumination control parameter from the camera pixel number set, the light source layout type set, and the illumination control parameter constraint interval; According to the first camera pixel number, the first light source layout type, and the first illumination control parameter, collect historical data to obtain an effective pixel historical calibration data set; Conduct centralized trend analysis on the effective pixel historical calibration data set to obtain a first effective pixel characteristic value; Substitute the first camera pixel number and the first effective pixel characteristic value into the optimization balance formula to determine whether the deviation between the first balance parameter and the second balance parameter is greater than or equal to a deviation threshold value; If greater, update the first camera pixel number, the first light source layout type, and the first illumination control parameter; If less than or equal to, set the first camera pixel number, the first light source layout type, and the first illumination control parameter as the recommended camera model, the recommended light source layout type, and the recommended illumination control parameter.

7. The method of claim 6, wherein, If less than or equal to, set the first camera pixel number, the first light source layout type, and the first illumination control parameter as the recommended camera model, the recommended light source layout type, and the recommended illumination control parameter, including: If less than or equal to, add the first camera pixel number, the first light source layout type, and the first illumination control parameter to an initial solution group; When the initial solution group is greater than or equal to a preset number, sort the initial solution group to obtain the recommended camera model, the recommended light source layout type, and the recommended illumination control parameter.

8. An intelligent measurement system based on multi-view image acquisition analysis, characterized in that, Including: A standard model obtaining module is configured to obtain a standard three-dimensional model of a target to be measured, wherein the standard three-dimensional model has a digital positioning pose label, an image acquisition angle label, a minimum detectable expected size, and size tolerance information; A two-dimensional image slice module is configured to perform two-dimensional image slicing on the standard three-dimensional model by using the image acquisition angle label to obtain monitoring visual angle target size information; An optimization balance type construction module is configured to construct an optimization balance type according to the minimum detectable expected size, the size tolerance information, and the monitoring visual angle target size information; A parameter optimization module is configured to optimize camera pixels, light source layout types, and illumination control parameters, and output recommended camera models, recommended light source layout types, and recommended illumination control parameters when the optimization balance type is satisfied, wherein the camera pixels refer to the number of long-side pixels of an image; A comparison model construction module is configured to synchronize positioning of the target to be measured based on the digital positioning pose, perform image acquisition by using the recommended camera models, the recommended light source layout types, and the recommended illumination control parameters, and construct a comparison three-dimensional model; A comparison analysis module is configured to perform comparison analysis based on the comparison three-dimensional model and the standard three-dimensional model, and send an identification area that does not satisfy the size tolerance information to a user terminal.

Images