Automatic measurement method and system for cell size of honeycomb curtain

Abstract

The present application relates to the field of image processing, and more particularly to a kind of automatic measurement method and system of honeycomb curtain cell size.The method includes the following steps: first, extract a plurality of candidate boundary points on a plurality of rays, and based on the prior information such as boundary uniqueness representing the thickness characteristics of honeycomb curtain edge and gradient peak symmetry representing gradient symmetry, calculate the local feature deviation cost of each candidate point;Then, all candidate points are constructed into a continuous path graph, and the path weight of the graph is integrated with the local feature deviation cost and the contour continuity deviation cost;Finally, the global optimal edge line is determined by searching the shortest path in the graph, and size measurement is carried out based on the edge line.The present application converts edge extraction into a graph search problem with physical constraints, which can effectively resist strong interference, significantly improve the accuracy, robustness of edge extraction and the accuracy of subsequent size measurement.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of image processing, and in particular to a method and system for automatically measuring the size of a honeycomb blind cell. BACKGROUND

[0002] In the automated production process of honeycomb blinds, accurate measurement of cell size is a key link to ensure product quality. Traditional measurement methods rely on manual sampling, which is inefficient and inconsistent. Therefore, using machine vision technology to achieve automated and high-precision online measurement is urgently needed in the industry.

[0003] Existing machine vision measurement methods typically include image acquisition, edge detection, and geometric fitting. Among them, the radial profile method is a commonly used measurement technique for hole targets. This method starts from the center of the hole and emits multiple rays in all directions. It analyzes the one-dimensional profile of the gray value on each ray and locates the edge by finding the gradient peak or the farthest point.

[0004] However, the traditional radial profile method, especially when relying on overly simple decision rules such as the farthest peak, has significant shortcomings in the context of honeycomb blind cell measurement. Honeycomb blinds under certain lighting conditions can produce complex shadows, highlights, and noise textures. These disturbances also manifest as significant gradient peaks in the gray profile, forming false edges. Traditional methods cannot effectively handle these false edges and noise textures, making it easy to mislocate the edge and resulting in jagged extracted profiles.

[0005] In summary, the traditional radial profile method cannot simultaneously address issues such as shadows, highlights, and noise textures due to its simple decision rules, making it difficult to meet the demand for accurate measurement of honeycomb blind cell size. Therefore, how to design a robust edge extraction strategy to overcome the above shortcomings has become the focus of the present application. SUMMARY

[0006] To solve the problem of how to overcome the interference of shadows, highlights, and noise textures and achieve high-precision automatic measurement of cell size, the present application provides a method and system for automatically measuring the size of a honeycomb blind cell.

[0007] In a first aspect, the present application provides a method for automatically measuring the size of a honeycomb blind cell, which adopts the following technical solution:

[0008] A method for automatically measuring the size of a honeycomb blind cell, comprising the steps of:

[0009] a. Obtain a honeycomb blind cell image;

[0010] b. extracting a plurality of candidate boundary points on a plurality of rays starting from a preset center point in the cellular shade cell image, and calculating a local feature deviation cost of each candidate boundary point based on a condition that the candidate boundary point meets a preset local feature of the edge of the cellular shade cell;

[0011] c. constructing a continuous path graph containing the candidate boundary points based on each candidate boundary point and the local feature deviation cost thereof, wherein a weight of a path in the continuous path graph is positively correlated with the local feature deviation cost;

[0012] d. searching for an optimal path with the lowest total weight in the continuous path graph, and connecting candidate boundary points corresponding to nodes on the optimal path to obtain an edge line of the cellular shade cell;

[0013] e. measuring a size of the cellular shade cell based on the edge line of the cellular shade cell.

[0014] The application realizes comprehensive consideration of the advantages and disadvantages of each point itself and the connection smoothness with adjacent points by calculating the local feature deviation cost of the candidate points and constructing the continuous path graph. This makes the final edge line a globally optimal solution rather than a simple connection of locally optimal points, so as to effectively resist strong interference such as local noise, shadow and highlight, and even in the area where part of the edge features are not clear, the correct contour can be found by relying on the advantages of the overall path, effectively improving the accuracy and robustness of edge extraction.

[0015] Preferably, the preset local features at least include:

[0016] a boundary uniqueness for representing the thickness feature of the edge of the cellular shade cell, and a gradient peak symmetry for representing the symmetry of the gradient distribution of the edge of the cellular shade cell.

[0017] The application extracts prior information unique to the edge of the cellular shade cell, which has a clear physical meaning: the boundary uniqueness corresponds to the physical thickness, and the gradient peak symmetry corresponds to the clear boundary transition. This enables the model to have strong discrimination ability and accurately distinguish real edge points with these physical features from shadow, highlight or texture artifacts without these features.

[0018] Preferably, the calculation of the boundary uniqueness includes:

[0019] defining a boundary window, an outer adjacent window and an inner adjacent window on the gray value sequence of the ray with the candidate boundary point as the center, and quantifying the boundary uniqueness based on the difference between the gray mean value of the boundary window and the gray mean values of the outer adjacent window and the inner adjacent window.

[0020] The present application can reliably identify the edge region with certain width and unique gray value by comparing the gray difference between the boundary window and the inner and outer adjacent windows, thereby effectively filtering out the shadow or highlight boundary without physical thickness and sharp gray change, and significantly enhancing the anti-interference ability of the algorithm.

[0021] Preferably, the calculation of the gradient peak symmetry comprises:

[0022] The derivative values on the inner and outer sides of the candidate boundary point are obtained, and the gradient peak symmetry is quantified based on the relationship between the sum of the derivative values on the inner and outer sides and the sum of the absolute values thereof.

[0023] The present application can effectively identify such symmetry, thereby excluding the single-sided gradient artifact without symmetry caused by uneven illumination or surface texture, and further improving the accuracy of the candidate boundary point screening.

[0024] Preferably, the path weight of the continuous path graph connecting two candidate boundary points on adjacent rays is positively related to the continuity deviation cost represented by the distance difference of the two candidate boundary points to the center point.

[0025] The present application takes into account that the edges of the honeycomb curtain cells are continuous and smooth in physics, and thus ensures that the edge line corresponding to the shortest path searched finally is more smooth and natural in shape, and conforms to the real geometric shape of the object, by punishing the connection of adjacent rays with too large radius jump.

[0026] Preferably, the step of searching for an optimal path with the lowest total weight in the continuous path graph is realized by applying the Dijkstra algorithm.

[0027] Preferably, the extraction method of the candidate boundary point comprises:

[0028] The threshold segmentation is performed on the honeycomb curtain cell image to determine the foreground region, the centroid of the foreground region is obtained as the center point, a plurality of rays are made from the center point, and the gray value sequence on each ray is obtained; the local extreme point with the derivative absolute value of the gray value sequence greater than a preset gradient threshold value is determined as the candidate boundary point.

[0029] Preferably, the step of measuring the size of the honeycomb curtain cell comprises:

[0030] The radial distance sequence of all edge points on the edge line is obtained;

[0031] The peak value detection is performed on the radial distance sequence to extract a plurality of edge vertices and a plurality of edge midpoints;

[0032] An outer diameter is calculated based on the distance between the edge vertices, and an inner diameter is calculated based on the distance between the edge midpoints.

[0033] Preferably, the method for constructing the continuous path graph further comprises:

[0034] When there is no candidate boundary point on a certain ray, a virtual node is created for the ray to ensure the closure of the continuous path graph and the integrity of the path search.

[0035] In a second aspect, the present application provides an automatic measurement system for the size of a honeycomb curtain cell, which adopts the following technical solution:

[0036] An automatic measurement system for the size of a honeycomb curtain cell comprises a processor and a memory, and the memory stores computer program instructions which, when executed by the processor, implement the above-mentioned automatic measurement method for the size of a honeycomb curtain cell.

[0037] By adopting the above technical solution, the above-mentioned automatic measurement method for the size of a honeycomb curtain cell is generated into a computer program and stored in the memory to be loaded and executed by the processor, so that a terminal device is made according to the memory and the processor, and use is facilitated.

[0038] The present application has the following technical effects:

[0039] The present application fundamentally overcomes the defects of traditional methods which are susceptible to interference of shadows, highlights, noise and textures by converting edge extraction into a graph theory shortest path problem combining local physical features and global smoothness constraints. It can penetrate local artifacts and find a globally optimal edge contour that best conforms to physical reality.

[0040] Further, the present application extracts prior information unique to the edges of honeycomb curtain cells and having clear physical meaning, so that the model has strong discrimination ability and can accurately distinguish real edge points having these physical features from shadow, highlight or texture artifacts that do not have these features. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 is a flowchart of a method in an automatic measurement method for the size of a honeycomb curtain cell provided by Embodiment 1 of the present application;

[0042] Figure 2 is a honeycomb curtain cell image provided by Embodiment 2 of the present application;

[0043] Figure 3 is a contrast image of a honeycomb curtain cell edge image provided by Embodiment 2 of the present application. DETAILED DESCRIPTION

[0044] Embodiment 1:

[0045] Embodiment 1 of the present application discloses an automatic measurement method for the cell size of a honeycomb curtain, referring to Figure 1 , comprising steps S1-S5:

[0046] S1: collecting a honeycomb curtain cell image.

[0047] Specifically, a camera is used to collect the honeycomb curtain cell image, and the honeycomb curtain cell image is subjected to grayscale processing and denoising processing to obtain a grayscale image. For the sake of subsequent description, the grayscale image is still referred to as the honeycomb curtain cell image.

[0048] S2: analyzing the gradient values of the pixels in the honeycomb curtain cell image to obtain candidate boundary points; calculating the likelihood of each candidate boundary point being a true boundary point based on the characteristics of the candidate boundary points conforming to the edge points of the honeycomb curtain cell; and calculating the local feature deviation cost of each candidate boundary point based on the likelihood of each candidate boundary point being a true boundary point.

[0049] It should be noted that the traditional radial profile method obtains an edge line in the following process: starting from a preset center point, drawing a radial line outward, calculating the gradient of each pixel on the radial line, and finding the pixel corresponding to the last peak point whose gradient value exceeds the threshold value as the boundary point. The profile obtained by connecting all the boundary points together. This method is easy to regard shadow boundary points, highlight boundary points or points on interference textures as boundary points, and thus the method cannot accurately obtain the edge line of the honeycomb curtain cell.

[0050] It should be further noted that, in order to exclude the interference of shadow and highlight boundary points and better extract the edge line of the honeycomb curtain cell, the prior information that the boundary points of the honeycomb curtain cell are different from the shadow and highlight boundary points should be considered. According to Figure 2 , the prior information of the honeycomb curtain cell relative to the shadow or highlight boundary points includes: 1. The honeycomb curtain cell has a certain thickness, thereby causing a unique gray area at the boundary position thereof; 2. The gradient difference at the boundary of the honeycomb curtain cell is large; and 3. The gradient symmetry at the boundary of the honeycomb curtain cell is large.

[0051] S20: analyzing the gradient values of the pixels in the honeycomb curtain cell image to obtain candidate boundary points.

[0052] Preferably, as an example, the analysis of the gradient values of the pixels in the honeycomb curtain cell image to obtain candidate boundary points includes:

[0053] The Otsu threshold value method is used to perform threshold segmentation on the honeycomb curtain cell image to obtain a foreground region and a background region. The foreground region mainly contains the internal information of the honeycomb curtain cell, and the background region mainly contains the external information of the honeycomb curtain cell.

[0054] The centroid of the foreground region is acquired as an initial center point, a ray is made outward from the initial center point, the gray value of all pixels on each ray is acquired to obtain a gray value sequence of each ray, the gray value sequence of each ray is subjected to Gaussian smoothing processing, and the local extreme point with a derivative absolute value greater than a preset gradient threshold is taken as a candidate boundary point.

[0055] It can be understood that the honeycomb blind cells are roughly segmented by the threshold segmentation manner, and the initial center point is located, and the manner is relatively accurate compared with the traditional manner of setting the initial center point.

[0056] S21: The possibility that each candidate boundary point is a real boundary point is calculated based on the feature condition that the candidate boundary point meets the edge point of the honeycomb blind cell, and the local feature deviation cost of each candidate boundary point is calculated based on the possibility that each candidate boundary point is a real boundary point.

[0057] S210: The possibility that each candidate boundary point is a real boundary point is calculated based on the feature condition that the candidate boundary point meets the edge point of the honeycomb blind cell.

[0058] Preferably, as an example, the local feature deviation cost of each candidate boundary point is calculated based on the possibility that each candidate boundary point is a real boundary point, and the local feature deviation cost of each candidate boundary point includes:

[0059] The boundary uniqueness of each candidate boundary point is calculated.

[0060] A preset first length window is acquired in the gray value sequence as a boundary window with the candidate boundary point as the center, a preset second length window is acquired as an outer adjacent window outside the adjacent boundary window, and a preset third length window is acquired as an inner adjacent window inside the adjacent boundary window.

[0061] The boundary uniqueness satisfies a relationship formula:

[0062]

[0063] Wherein, The gray value mean of the boundary window is denoted as μb, The gray value mean of the outer adjacent window is denoted as μo, The gray value mean of the inner adjacent window is denoted as μi, The gray value standard deviation of the boundary window is denoted as σb, and min() represents the minimum value, The boundary uniqueness of each candidate boundary is denoted as U.

[0064] It can be understood that, due to the certain thickness of the honeycomb curtain cell, the honeycomb curtain cell has a boundary continuation area. The boundary of the shadow and highlight area has no thickness, and it will not have a boundary continuation area. The honeycomb curtain cell has a unique gray value at the boundary continuation area, the gray value inside is similar, and the boundary continuation area has a large difference in gray value with other adjacent areas, so a boundary degree feature index is constructed based on the two information to describe the thickness characteristics of the boundary of the honeycomb curtain cell. reflecting the gray value information of the thickness area of the honeycomb curtain cell, indicating the gray value of the two adjacent areas on both sides of the thickness area of the honeycomb curtain cell, reflecting the uniqueness of the thickness area of the honeycomb curtain cell compared to the two adjacent areas.

[0065] Calculate the gradient peak symmetry of each candidate boundary point:

[0066]

[0067] wherein, denotes the derivative of the outside pixel point at the preset position of the candidate boundary point, denotes the derivative of the inside pixel point at the preset position of the candidate boundary point, and | | denotes the absolute value symbol, denotes the preset zero prevention parameter, denotes the gradient peak symmetry.

[0068] It can be understood that the edge of the honeycomb curtain cell has a distribution symmetry, so the gradient peak symmetry can better describe the edge information of the honeycomb curtain cell. If the gradient is symmetrical, the sum of the two derivatives will be biased towards 0, that is, biased towards 0, so the smaller, the greater the gradient peak symmetry.

[0069] Calculate the possibility of each candidate boundary point being a real boundary point:

[0070]

[0071] wherein, denotes the reference weight of the boundary uniqueness, denotes the reference weight of the gradient peak symmetry, denotes the reference weight of the gradient amplitude, denotes the gradient amplitude, denotes the possibility of each candidate boundary point being a real boundary point.

[0072] S211: Calculate the local feature deviation cost of each candidate boundary point based on the possibility of each candidate boundary point being a real boundary point.

[0073] Preferably, as an example, the local feature deviation cost of each candidate boundary point is calculated based on the possibility of each candidate boundary point being a true boundary point, and includes:

[0074]

[0075] wherein, represents the local feature deviation cost of each candidate boundary point.

[0076] S3: constructing a continuous path graph based on the candidate boundary points, and setting the path weight in the continuous path graph based on the local feature deviation cost.

[0077] It should be noted that the above only obtains discrete candidate boundary points, and the candidate boundary points need to be connected into lines in the following, so as to facilitate subsequent screening of the edge lines that best meet the honeycomb blind hole edge features.

[0078] It should be further noted that the directed graph can realize the connection of discrete points into paths, and thus the directed graph can be used to connect the discrete points into lines.

[0079] Preferably, as an example, the continuous path graph is constructed, including:

[0080] First, a virtual START node is initialized.

[0081] Next, each ray is traversed, and if there is a candidate boundary point on the ray, the candidate boundary point and its radius are created as a real node in the graph. If there is no candidate boundary point on the ray, a virtual node is created for the ray.

[0082] Subsequently, an initial ray is selected, the virtual START node and the nodes on the initial ray are constructed into directed edges, and the comprehensive weight of each directed edge is set to be equal to the local feature deviation cost of the corresponding node on the initial ray.

[0083] Then, starting from the initial ray, directed edges are established between the nodes of each two adjacent rays, and the comprehensive weight of each directed edge is set.

[0084] Finally, the nodes of the last ray and the nodes on the initial ray are constructed into directed edges, and the comprehensive weight of each directed edge is set, to obtain the continuous path graph.

[0085] It can be understood that by constructing the directed graph, the discrete points can be connected into several closed path lines, providing a basis for subsequent selection of edge lines.

[0086] It should be added that the setting method of the comprehensive weight of the directed edge between the nodes of two rays includes:

[0087]

[0088] wherein, represents a reference weight of the local feature deviation cost, used for determining a reference degree of the local feature deviation cost, represents the local feature deviation cost of the start node of the directed edge, represents a reference weight of the continuity deviation cost, used for determining a reference degree of the continuity deviation cost, represents the distance from the start node of the directed edge to the initial center point, represents the distance from the end node of the directed edge to the initial center point, represents the continuity deviation cost, represents the comprehensive weight of the directed edge.

[0089] It can be understood that, since the edges of the honeycomb shade cells are smooth and continuous, the continuity of the edges needs to be considered when extracting the edges of the honeycomb shade cells.

[0090] Further, the greater the local feature deviation cost, the greater the degree that the node does not conform to the features of the honeycomb shade cell edge points; the greater the continuity deviation cost, the worse the continuity of the two nodes of the directed edge. The comprehensive weight can effectively measure the condition that the directed edge conforms to the honeycomb shade cell edge line by comprehensively analyzing the local feature deviation cost and the continuity deviation cost.

[0091] In particular, when calculating the comprehensive weight of the directed edge between the nodes of the two rays, if one of the nodes of the two rays is a virtual node, the comprehensive weight of the directed edge between the nodes of the two rays is set to a preset value. The embodiment needs to ensure that the preset value is as large as possible, and the specific setting method of the preset value is not limited in the embodiment.

[0092] S4: extracting the shortest path in the continuous path graph to obtain the edge line of the honeycomb shade cell.

[0093] It should be noted that the above steps connect all discrete candidate boundary points into several closed lines, and the edge line that most conforms to the edge features of the honeycomb shade cell needs to be selected from the several closed lines.

[0094] Preferably, as an example, the step of extracting the shortest path in the continuous path graph to obtain the edge line of the honeycomb shade cell includes:

[0095] applying the Dijkstra algorithm to analyze the continuous path graph, and extracting an optimal path from the START node to the last ray with the lowest total weight. The candidate boundary points corresponding to the nodes on the optimal path are connected in order to obtain the edge line of the honeycomb shade cell.

[0096] It can be understood that the comprehensive weight of each directed edge on the continuous path graph reflects the situation that the path conforms to the honeycomb curtain cell edge line, and the greater the comprehensive weight, the less the path conforms to the honeycomb curtain cell edge line, so that the optimal path with the lowest total weight extracted by the Dijkstra algorithm can better conform to the honeycomb curtain cell edge line.

[0097] S5: measuring the size of the honeycomb curtain cell based on the edge line of the honeycomb curtain cell.

[0098] Preferably, as an example, measuring the size of the honeycomb curtain cell based on the edge line of the honeycomb curtain cell comprises:

[0099] First, the centroid of all edge points on the edge line of the honeycomb curtain cell is obtained;

[0100] Then, the Euclidean distance of each edge point to the centroid is calculated and recorded as the radial distance, and the radial distances of all edge points are arranged in order to obtain a radial distance sequence;

[0101] After that, a peak detection algorithm is applied to the radial distance sequence to extract the edge points corresponding to the six most significant local maximum values, which are recorded as edge vertices, and the edge points corresponding to the six most significant local minimum values, which are recorded as edge midpoints;

[0102] For any edge vertex, the edge vertex and the farthest edge vertex form a vertex pair, and a plurality of vertex pairs are obtained.

[0103] For any edge midpoint, the edge midpoint and the farthest edge midpoint form a midpoint pair.

[0104] The Euclidean distance of the two edge vertices in the vertex pair is taken as the outer diameter of the vertex pair, and the average of the outer diameters of all vertex pairs is taken as the final outer diameter of the honeycomb curtain cell.

[0105] The Euclidean distance of the two edge midpoints in the midpoint pair is taken as the inner diameter of the midpoint pair, and the average of the inner diameters of all midpoint pairs is taken as the final inner diameter of the honeycomb curtain cell.

[0106] Embodiment 1 of the present application also discloses an automatic measurement system for the size of a honeycomb curtain cell, comprising a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement an automatic measurement method for the size of a honeycomb curtain cell according to the present application.

[0107] The above system also includes a communication bus and a communication interface and other components well known to those skilled in the art, the settings and functions of which are known in the art, and therefore will not be described here.

[0108] In this application, the aforementioned memory can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device. For example, the computer-readable storage medium can be any suitable magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory, a dynamic random access memory, a static random access memory, an enhanced dynamic random access memory, a high bandwidth memory, a hybrid memory cube, or the like, or any other medium that can be used to store the desired information and that can be accessed by an application, module, or both. Any such computer storage media can be part of the device or accessible or connectable thereto.

[0109] Embodiment 2

[0110] The embodiment is a specific embodiment of a method for automatically measuring the size of a honeycomb curtain cell.

[0111] To illustrate the technical effects of Embodiment 1 of the present application, the technical effects of the technical solutions will be described below in combination with some experimental process data.

[0112] Experimental goal: verify the accuracy of the proposed algorithm in extracting the honeycomb curtain cell edge.

[0113] Figure 2 The image is a honeycomb curtain cell image. As can be seen from the image, there is shadow information outside the honeycomb curtain cell and other honeycomb curtain cell structure information interference. At the same time, there is shadow information interference inside the honeycomb curtain cell.

[0114] Figure 3 The image is a honeycomb curtain cell edge image comparison chart. The left side of the image shows the honeycomb curtain cell edge image obtained using the method in Embodiment 1, and the right side of the image shows the honeycomb curtain cell edge image obtained using the traditional radial profile method. As can be seen from the image, the traditional radial profile method is disturbed by false edges and noise textures, resulting in the extracted honeycomb curtain cell edge line being jagged and having a large difference from the true honeycomb curtain cell edge. The honeycomb curtain cell edge extracted by the method of the embodiment closely matches the true honeycomb curtain cell edge.

Claims

1. An automatic measurement method for the size of honeycomb curtain mesh, characterized in that, Including the following steps: a. Obtain the honeycomb screen grid image; b. Extract multiple candidate boundary points along multiple rays originating from the preset center point in the honeycomb mesh image, and calculate the local feature deviation cost of each candidate boundary point based on the fact that the candidate boundary points conform to the preset local features of the honeycomb mesh edge. c. Based on each candidate boundary point and its local feature deviation cost, construct a continuous path graph containing the candidate boundary points, wherein the weight of the path in the continuous path graph is positively correlated with the local feature deviation cost; d. Search for the optimal path with the lowest total weight in the continuous path graph, and connect the candidate boundary points corresponding to the nodes on the optimal path to obtain the edge line of the honeycomb screen. e. Measure the size of the honeycomb mesh based on the edge line of the honeycomb mesh; The preset local features include at least: boundary uniqueness for characterizing the thickness characteristics of the honeycomb mesh edge, and gradient peak symmetry for characterizing the gradient distribution symmetry of the honeycomb mesh edge; The calculation of boundary uniqueness includes: defining a boundary window, an outer adjacent window, and an inner adjacent window on the gray value sequence of the ray with the candidate boundary point as the center; and quantifying the boundary uniqueness based on the difference between the gray mean of the boundary window and the gray mean of the outer and inner adjacent windows.

2. The automatic measurement method for the size of honeycomb curtain mesh according to claim 1, characterized in that, The calculation of the gradient peak symmetry includes: Obtain the derivative values ​​on both sides of the candidate boundary point; and quantify the gradient peak symmetry based on the relationship between the sum of the derivative values ​​on both sides and the sum of their respective absolute values.

3. The automatic measurement method for the size of honeycomb curtain mesh according to claim 1, characterized in that, The path weight connecting two candidate boundary points on adjacent rays in the continuous path graph is also positively correlated with the continuity deviation cost represented by the difference in distance from the two candidate boundary points to the center point.

4. The automatic measurement method for the size of honeycomb curtain mesh according to claim 1, characterized in that, The step of searching for the optimal path with the lowest total weight in the continuous path graph is implemented by applying Dijkstra's algorithm.

5. The automatic measurement method for the size of honeycomb curtain mesh according to claim 1, characterized in that, The method for extracting candidate boundary points includes: The honeycomb grid image is segmented using a threshold to determine the foreground region; the centroid of the foreground region is obtained as the center point; multiple rays are emanated from the center point, and the gray value sequence on each ray is obtained; the local extreme points where the absolute value of the derivative of the gray value sequence is greater than a preset gradient threshold are determined as the candidate boundary points.

6. The automatic measurement method for the size of honeycomb curtain mesh according to claim 1, characterized in that, The step of measuring the size of the honeycomb screen includes: Obtain the radial distance sequence of all edge points on the edge line; Peak detection is performed on the radial distance sequence to extract multiple edge vertices and multiple edge midpoints; The outer diameter is calculated based on the distance between the edge vertices, and the inner diameter is calculated based on the distance between the edge midpoints.

7. The automatic measurement method for the size of honeycomb curtain mesh according to claim 1, characterized in that, The method for constructing the continuous path graph further includes: When there are no candidate boundary points on a certain ray, a virtual node is created for that ray to ensure the closure of the continuous path graph and the integrity of the path search.

8. An automatic measurement system for the size of honeycomb curtain mesh, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement an automatic measurement method for honeycomb screen aperture size according to any one of claims 1-7.

Images