Method and device for identifying microstructure of 2.5-dimensional braided composite materials based on machine vision

Abstract

The present application provides a method and device for identifying the microstructure of a 2.5-dimensional woven composite material based on machine vision. The method provided by the present application includes: identifying the target woven composite material based on the top view and the structural tensor information identification method of the target woven composite material to obtain a first material classification result; determining the pixel range of the warp and weft yarns of the target woven composite material from the main view based on the first material classification result of the top view, and determining the number of pixels in each independent area based on the pixel range, thereby optimizing the first material classification result and obtaining a second material classification result; obtaining a third material classification result based on the main view recognition result and the side view recognition of the overlapping warp yarns, and combining the second material classification result and the third material classification result to obtain the recognition result of the microstructure of the target woven composite material. The method and device provided by the present application are used to accurately and efficiently identify the microstructure of woven composite materials.

Description

Technical Field

[0001] The present application relates to the technical field of composite material microstructure, and in particular to a method and device for identifying the microstructure of a 2.5-dimensional woven composite material based on machine vision. Background Art

[0002] 2.5-dimensional woven composites are usually composed of a two-dimensional woven fabric and a non-woven structure (such as an interlayer insert or filler of the woven fabric). This structure gives the composite material properties between two-dimensional and three-dimensional. Compared with traditional woven composites, 2.5-dimensional woven composites have higher strength, stiffness and wear resistance. At the same time, they are easier to prepare, have relatively low cost, and can more easily realize some specific structural designs.

[0003] With the development of woven composite materials, the identification of their mesostructure has become a research focus. The mesostructure of woven composite materials can improve product quality and promote the development of woven composite materials.

[0004] The existing mesostructure recognition method uses manual marking to mark the woven composite materials in the image, which not only requires high professional quality of the marking personnel, but is also very time-consuming and laborious. The existing mesostructure recognition method based on computer vision algorithm cannot effectively complete the recognition of woven composite materials because the yarns in the woven composite materials overlap with each other and cannot distinguish the boundaries between different yarns. Summary of the invention

[0005] In view of this, the present application provides a 2.5-dimensional woven composite material mesostructure recognition method and device based on machine vision, which are used to accurately and efficiently identify the mesostructure of the woven composite material.

[0006] Specifically, the present application is implemented through the following technical solutions:

[0007] The first aspect of the present application provides a method for identifying a 2.5-dimensional braided composite material microstructure based on machine vision, the method comprising:

[0008] Identify the target woven composite material based on the top view and the structural tensor information identification method of the target woven composite material to obtain a first material classification result, wherein the first material classification result at least includes warp yarns, weft yarns and pores;

[0009] Determine the pixel ranges of the warp yarns and weft yarns of the target woven composite material from the main view based on the first material classification result of the top view, respectively, and determine the number of pixels in each independent area based on the pixel range, thereby optimizing the first material classification result to obtain a second material classification result;

[0010] Based on the front view recognition result and the warp yarns that overlap with the side view recognition result, a third material classification result is obtained, and the recognition result of the microstructure of the target woven composite material is obtained by combining the second material classification result and the third material classification result;

[0011] Wherein, a preset number of control points are selected on the warp yarns and the weft yarns based on the main view, and the second material classification result is optimized based on the correspondence between the control points in the main view and those in the side view.

[0012] The second aspect of the present application provides a 2.5-dimensional braided composite material microstructure recognition device based on machine vision, the device comprising a processing module, an optimization module and a recognition module;

[0013] The processing module is used to identify the target woven composite material based on the top view of the target woven composite material and the structural tensor information identification method, and obtain a first material classification result, wherein the first material classification result at least includes warp yarns, weft yarns and pores;

[0014] The optimization module is used to determine the pixel ranges of the warp yarns and weft yarns of the target woven composite material from the main view respectively based on the first material classification result of the top view, and determine the number of pixels in each independent area based on the pixel range, so as to optimize the first material classification result and obtain the second material classification result;

[0015] The recognition module is used to obtain a third material classification result based on the front view recognition result and the side view recognition result of the overlapping warp yarns, and obtain the recognition result of the target woven composite material mesostructure by combining the second material classification result and the third material classification result;

[0016] Wherein, a preset number of control points are selected on the warp yarns and the weft yarns based on the main view, and the second material classification result is optimized based on the correspondence between the control points in the main view and those in the side view.

[0017] The third aspect of the present application provides a 2.5-dimensional woven composite material microstructure recognition device based on machine vision, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of any one of the methods provided in the first aspect of the present application are implemented.

[0018] The fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of the methods provided in the first aspect of the present application.

[0019] The present application provides a 2.5-dimensional woven composite material micro-structure recognition method and device based on machine vision. First, the target woven composite material is identified based on the top view and structure tensor information recognition method of the target woven composite material to obtain a first material classification result, which includes at least warp, weft and pores. Then, based on the first material classification result of the top view, the pixel range of the warp and weft of the target woven composite material is determined from the main view, and the number of pixels in each independent area is determined based on the pixel range, so as to optimize the first material classification result and obtain the second material classification result. Then, based on the main view recognition result and the side view recognition result, the overlapping warp is identified to obtain the third material classification result. Combined with the second material classification result and the third material classification result, the recognition result of the micro-structure of the target woven composite material is obtained. In this way, the first material classification result is combined with the top view and the structure tensor information, and rich structural features can be extracted to ensure the accuracy and reliability of the first material classification result. Further, combined with the information of the number of pixels, the first material classification result can be accurately optimized based on morphology, ensuring the recognition effect of the second material classification result. Furthermore, by combining the recognition result of the main view and the information of the side view, the complex microstructure with a relatively high degree of overlap can be accurately identified, and the third material classification result can be obtained, thereby ensuring the accuracy of the third material classification recognition result. Furthermore, the final recognition result is obtained through the second material classification result and the third material classification result, thereby improving the accuracy of the recognition of the microstructure of the woven composite material, and the recognition result can be obtained accurately and efficiently. In general, the method and device provided by the present invention utilize the image information of each view in the three views, comprehensively consider the influence of the views of each angle of the material on the recognition result, and improve the accuracy of structure recognition; at the same time, in the process of using each view, the information determined by other views is used as a constraint, that is, the image information of each view does not exist independently, but exists in a mutually constrained relationship, thereby narrowing the scope that needs to be considered when using one view to recognize the structure, reducing the amount of calculation in the recognition process, and improving the recognition efficiency, while avoiding the mutual exclusion of the recognition results of each view, thereby improving the accuracy and utilization rate of the recognition results. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A flowchart of Example 1 of a method for identifying the microstructure of a 2.5-dimensional braided composite material based on machine vision provided in this application;

[0021] Figure 2 A schematic diagram of a set of three-view embodiment 1 of the target woven composite material provided by the present application;

[0022] Figure 3 A schematic diagram of a first embodiment of the key point position provided in this application;

[0023] Figure 4 A schematic diagram of Example 1 of the third material classification result provided in this application;

[0024] Figure 5 A schematic diagram of the identification results of the target woven composite material mesostructure provided for this application;

[0025] Figure 6 A flowchart of Example 2 of a method for identifying the microstructure of a 2.5-dimensional braided composite material based on machine vision provided in this application;

[0026] Figure 7 A flowchart of Example 3 of a method for identifying the microstructure of a 2.5-dimensional braided composite material based on machine vision provided in this application;

[0027] Figure 8 A schematic diagram of the finite element model provided for this application;

[0028] Fig. 9 A hardware structure diagram of the device in which the 2.5-dimensional braided composite material microstructure recognition device based on machine vision is located in this application;

[0029] Fig.10 This is a structural schematic diagram of Example 1 of a 2.5-dimensional woven composite material microstructure recognition device based on machine vision provided in the present application. DETAILED DESCRIPTION

[0030] Here, exemplary embodiments will be described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application.

[0031] The terms used in this application are only for the purpose of describing specific embodiments and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in this article refers to and includes any or all possible combinations of one or more associated listed items.

[0032] It should be understood that although the terms first, second, third, etc. may be used in the present application to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0033] The present application provides a 2.5-dimensional woven composite material microstructure recognition method and device based on machine vision, which are used to accurately and efficiently recognize the microstructure of the woven composite material.

[0034] Specific embodiments are given below to introduce the technical solution of the present application in detail.

[0035] Figure 1 This is a flow chart of the first embodiment of the method for identifying the microscopic structure of a 2.5-dimensional braided composite material based on machine vision provided in this application. Please refer to Figure 1 , the method provided in this embodiment may include:

[0036] S101. Identify a target woven composite material based on a top view and a structural tensor information identification method of the target woven composite material to obtain a first material classification result, wherein the first material classification result at least includes warp yarns, weft yarns, and pores.

[0037] Specifically, the target braided composite material is a composite material formed by interweaving fiber bundles. It is formed by weaving two or more fibers arranged in different directions and then solidified with an adhesive. In this embodiment, a 2.5D braided composite material can be selected as the target braided composite material.

[0038] Furthermore, the three views of the target braided composite material include a top view, a front view and a side view, and the internal structure of the target braided composite material can be fully understood through the three views.

[0039] Optionally, the target woven composite material is scanned based on an XCT device to obtain a plurality of groups of three-view images, wherein the plurality of groups of three-view images are derived from scanning results of the target woven composite material of different thicknesses.

[0040] Specifically, when the target woven composite material is scanned by an XCT device, three views corresponding to the target woven composite material at different depths can be obtained. Through multiple sets of three views, the stacking structure of the target woven composite material in three directions can be comprehensively obtained, so as to fully understand its structural characteristics. Figure 2 This is a schematic diagram of a set of three-view embodiments of the target woven composite material provided by this application. Figure 2 , the target woven composite material includes a top view, a front view and a side view.

[0041] Furthermore, the structure tensor information is a second-order symmetric matrix that represents the gradient changes in the local area of ​​the image. Through the structure tensor information, the texture, edge, direction and other information in the image can be analyzed to obtain the corresponding results.

[0042] In the specific implementation, a specific embodiment is given below to introduce in detail the process of obtaining the first material classification result:

[0043] (1) Obtaining gradient information and direction information of the target woven composite material based on the top view.

[0044] Specifically, the gradient information of the target woven composite material represents the grayscale or color change rate at different positions in the top view, and the gradient information of the target woven composite material can be used to characterize the degree of change of the image at different positions. In specific implementation, the larger the gradient information of the top view, the more drastic the change of the image at this position.

[0045] Furthermore, the directional information of the target woven composite material represents the direction of change of different points in the top view. It should be noted that the gradient information of the target woven composite material can be represented as a vector, and its direction corresponds to the directional information of the target woven composite material. The texture direction and structural characteristics of different regions in the image can be determined by the directional information of the target woven composite material.

[0046] In specific implementation, the gradient information and direction information of the target woven composite material may be extracted by using a traditional information extraction method, or the gradient information and direction information of the target woven composite material may be extracted by using deep learning, which is not limited in this embodiment.

[0047] (2) Classifying the target woven composite material based on the gradient information and the direction information to obtain the first material classification result.

[0048] Specifically, when the target woven composite material is classified, the specific types obtained by classification can be determined according to actual needs, and this is not limited in this embodiment. For example, in one embodiment, the target woven composite material can be classified, and the fiber bundles passing along the weft yarns in the target woven composite material are classified as warp yarns, and the fiber bundles passing along the warp direction in the target woven composite material are classified as weft yarns, and the gaps or air areas in the target woven composite material are divided into pores to obtain warp yarns, weft yarns and pores.

[0049] In this step, a first material classification result of the completed classification is obtained based on the gradient information and direction information of the top view.

[0050] The 2.5-dimensional woven composite material microstructure recognition method based on machine vision provided in this embodiment first obtains the gradient information and direction information of the target woven composite material based on the top view; then, based on the gradient information and direction information, the target woven composite material is classified to obtain the first material classification result. In this way, by obtaining the gradient information and direction information from the top view, a reliable classification result can be obtained, which ensures the recognition accuracy of the first material classification result.

[0051] S102. Based on the first material classification result of the top view, the pixel ranges of the warp yarns and weft yarns of the target woven composite material are determined from the main view respectively, and the number of pixels in each independent area is determined based on the pixel ranges, so as to optimize the first material classification result and obtain the second material classification result.

[0052] Specifically, the pixel range of the warp and weft of the target woven composite material is the interval of pixel values ​​of different parts of the warp and weft of the target woven composite material. In other words, the pixel ranges of different parts of the warp and weft of the target woven composite material are different, and the warp and weft have different corresponding pixel ranges.

[0053] Furthermore, the pixel value in each pixel range is the number of pixels in the pixel range. In specific implementation, the pixel value in each pixel range can be obtained based on the pixel value acquisition method, and then the pixel value is used as the number of pixels in the corresponding pixel range.

[0054] When this step is implemented specifically, a specific embodiment is given below to detail the process of optimizing the first material classification result and obtaining the second material classification result by the number of pixels in each independent area:

[0055] (1) Based on the first material classification result of the top view, the pixel ranges of the warp yarn and the weft yarn of the target woven composite material are determined from the main view respectively, and the number of pixels in each independent area is determined based on the pixel range, so as to obtain a second material classification result.

[0056] In specific implementation, it is necessary to screen out the pixel range of the pores and then determine the remaining area as the pixel range of the warp and weft. It should be noted that the method for determining the pixel range of the warp and weft of the target woven composite material can be selected according to actual needs, and this is not limited in this embodiment. For example, in one embodiment, the maximum inter-class variance method can be used to segment the main view image, and the segmented image is processed by the maximum inter-class variance method to obtain two categories with the largest variance.

[0057] Further, after obtaining the pixel range corresponding to the warp yarn and the pixel range corresponding to the weft yarn in the main view, the first material classification result is optimized and corrected based on the number of pixels within the pixel range to obtain the second classification result. It should be noted that the method for obtaining the number of pixels can be selected according to actual needs, and this is not limited in this embodiment.

[0058] (2) Based on the warp yarns and weft yarns identified in the top view, a pixel range of the corresponding warp yarns and weft yarns is determined from the front view as the identification range in the front view.

[0059] In specific implementation, in this step, since the pixel range of the warp yarn and the pixel range of the weft yarn are quite different, the warp yarn and the weft yarn in the top view can be identified, and then the position corresponding to the pixel range in the main view is obtained, and these ranges are determined as the identification range.

[0060] It should be noted that, since the main view and the top view have a corresponding relationship, the pixel range of the warp yarn and the weft yarn in the main view can be determined by the warp yarn and the weft yarn identified in the top view.

[0061] (3) Obtaining a second material classification result based on the identification range.

[0062] In specific implementation, in this step, the area outside the recognition range is determined as a pore, and the pixel range of the warp yarn and the pixel range of the weft yarn in the recognition range are added to obtain the second material classification result.

[0063] The 2.5-dimensional woven composite material microstructure recognition method based on machine vision provided in this embodiment first determines the pixel ranges of the warp and weft of the target woven composite material from the main view based on the first material classification result of the top view, and determines the number of pixels in each independent area based on the pixel range, thereby obtaining the second material classification result, and then determines the pixel ranges of the corresponding warp and weft from the main view based on the warp and weft identified in the top view as the recognition range in the main view, and finally obtains the second material classification result based on the recognition range. In this way, the first material classification result of the top view is processed by the maximum inter-class variance method. Since the pixels of the pores are quite different from those of other parts, the pixel range of the pore part can be accurately obtained. In this way, the second material classification result can be obtained by combining the pixel range of the warp and the pixel range of the weft.

[0064] Another specific embodiment is given below, which details the process of obtaining the second material classification result based on the recognition range recognition:

[0065] (1) The recognition range of the main view is divided into at least one independent area, and the independent area is determined as a warp yarn category or a weft yarn category according to the number of warp yarn pixels of the warp yarn and the number of weft yarn pixels of the weft yarn in each independent area.

[0066] Specifically, an independent region refers to a single region that is divided, and the independent regions do not overlap with each other and cover the entire recognition range. It should be noted that each region contains a number of pixels, and these pixels are regarded as a whole during the analysis process.

[0067] In specific implementation, the area and position of the independent area are set according to actual needs, and are not limited in this embodiment. For example, in one embodiment, the actual size of the warp and weft yarns can be used as the area of ​​the independent area, and the main view can be evenly divided into 50 independent areas.

[0068] Furthermore, the pixels in the independent regions are distinguished to obtain warp pixels and weft pixels. For example, in one embodiment, an appropriate image feature extraction method can be used to extract texture features of the image, and then a classifier is used to classify the texture features to obtain warp pixels and weft pixels.

[0069] Furthermore, the independent regions are divided into different categories based on the number of warp pixels and weft pixels. In specific implementation, the category of the independent region can be determined based on a comparison between the preset value set in advance and the number. For example, in one embodiment, the part with a large number of pixels can be directly determined as the category of the independent region. For another example, in another embodiment, when the number of one of the warp pixels or the weft pixels is less than the preset value, the independent region can be determined as another category.

[0070] (2) Correcting the first material classification result based on the warp yarn category and the weft yarn category to determine the second material classification result.

[0071] In specific implementation, in this step, based on processing the first material classification result by the warp yarn category and the weft yarn category, a corrected result is obtained, and the corrected result is determined as the second material classification result.

[0072] The 2.5-dimensional woven composite material microstructure recognition method based on machine vision provided in this embodiment first divides the recognition range of the main view into at least one independent area, and determines the independent area as a warp category or a weft category according to the number of warp yarn pixels and weft yarn pixels in each independent area; then corrects the first material classification result based on the warp yarn category and the weft yarn category, and determines the second material classification result. In this way, the independent area is divided into different categories based on the number of warp yarn pixels and weft yarn pixels, and then the second material classification result is corrected, and the first material classification result can be further corrected, which ensures the accuracy of the warp yarn category and the weft yarn category in the second material classification result, and facilitates the accurate recognition of the microstructure of the woven composite material.

[0073] S103, obtaining a third material classification result based on the front view recognition result and the warp yarns that overlap with the side view recognition result, and combining the second material classification result and the third material classification result to obtain a recognition result of the target woven composite material mesostructure;

[0074] Wherein, a preset number of control points are selected on the warp yarns and the weft yarns based on the main view, and the second material classification result is optimized based on the correspondence between the control points in the main view and those in the side view.

[0075] Specifically, the warp yarns on the front view and the side view are identified, and after the identification results are obtained, the overlap of the two results is calculated based on the overlap acquisition method. It should be noted that the overlap can be defined as the overlap ratio of the warp yarn pixels in two regions.

[0076] In specific implementation, when warp overlap is detected on the main view, the corresponding warp can be directly determined as the overlapped warp; further, if warp overlap is not detected on the main view, the warp overlap is continued to be detected on the side view. If warp overlap is detected in the side view, the corresponding warp can also be determined as the overlapped warp. In this way, first determine whether there are overlapping warp yarns through the main view. When it is impossible to determine through the main view, continue to determine the overlapping warp yarns based on the side view, while ensuring the efficiency of the overlapping warp yarn detection, the reliability of the detection is guaranteed. Another specific embodiment is given below, which details the process of obtaining the third material classification result based on the main view recognition result and the side view recognition of the overlapping warp yarns:

[0077] (1) Based on the relative positions of the warp yarns and the weft yarns on the main view, control points on the warp yarns and the weft yarns are marked.

[0078] Specifically, the position of the warp yarn is represented by the control point. In specific implementation, for example, in one embodiment, the control points can be determined as the leftmost, leftmost and middle parts of the warp yarn edge. Figure 3 This is a schematic diagram of the control point embodiment 1 provided in this application. Please refer to Figure 3 , in the main view, each weft yarn divides the warp yarn, and 3 control points are confirmed on one edge.

[0079] It should be noted that when identifying control points on warp and weft yarns, the specific number of control points is determined according to the number of warp yarns, and at least 3 control points can be identified for each warp yarn. For example, in one embodiment, 5 warp yarns are identified in the main view, and there are 15 control points.

[0080] Furthermore, in order to distinguish the warp yarns, the control points at the upper edge of the warp yarns and the control points at the lower edge of the warp yarns can be marked. In specific implementation, for example, in one embodiment, the odd-numbered control points can be determined as the upper edge of the warp yarns, and the even-numbered control points can be determined as the lower edge of the warp yarns.

[0081] (2) Mapping the control points on the main view to the side view to obtain the control points in the side view.

[0082] Specifically, since the same warp and weft in the main view and the side view have corresponding characteristics, the control points in the main view can be mapped to the side view to obtain the control points of the same warp or weft. At the same time, by comparing the control points in the main view with the control points in the side view, the control point positions can be made more accurate.

[0083] (3) Identifying edge points of the warp yarns whose overlap is greater than a threshold based on the control points in the side view.

[0084] Specifically, the specific value of the threshold is determined according to actual needs, and is not limited in this embodiment. In specific implementation, the control points in the side view are compared with the control points in the main view for coincidence, and the edge points whose coincidence meets the requirements are obtained. In other words, the control points at the same position in the main view and the side view are determined as the edge points of the warp yarn.

[0085] (4) Using the edge point as a new edge point to determine the warp edge.

[0086] In specific implementation, in this step, the edge points are connected into lines, which are determined as warp yarn edges.

[0087] Furthermore, the morphological structure element can be used to erode the entire image so that the upper and lower edges of the warp yarn are flat. It should be noted that the warp yarn edge can be defined as a pore, and its pixel value is assigned to 0.

[0088] The 2.5-dimensional woven composite material microstructure recognition method based on machine vision provided in this embodiment first marks the control points on the warp and weft based on the relative position of the warp and weft in the main view; then maps the control points on the main view to the side view to obtain the control points in the side view; then identifies the edge points of the warp with a coincidence greater than a threshold based on the control points in the side view; finally, uses the edge points as new edge points to determine the warp edges. In this way, the initial control points are determined based on the relative position of the warp and weft, and then the edge points can be accurately and efficiently obtained through the control points. After traversing the main view and the side view, all the edge points of the warp can be detected, and finally the accurate warp edges are obtained to divide each warp.

[0089] It should be noted that a preset number of control points are selected on the warp yarns and the weft yarns based on the main view, and the second material classification result is optimized based on the correspondence between the control points in the main view and those in the side view.

[0090] Specifically, the preset number is a specific value determined according to actual needs, and is not limited in this embodiment. For example, in one embodiment, the preset number can be set to a multiple of 3.

[0091] It should be noted that, the second material classification result is obtained by optimizing the first material classification result according to the pixel distribution of the target woven composite material image, the third material classification result is determined according to the control point, the third material classification result completes the division of each warp yarn, and the first material classification result, the second material classification result and the third material classification result have the same classification label range. For example, in one embodiment, the classification label ranges of the first material classification result, the second material classification result and the third material classification result are all divided into warp yarn, weft yarn and pore.

[0092] A specific embodiment is given below to describe in detail the process of combining the second material classification result and the third material classification result:

[0093] (1) Calculate the center of gravity of each warp yarn based on the warp yarn edges.

[0094] Specifically, the center of gravity of each warp yarn can be calculated through the warp yarn edge, and the method for obtaining the center of gravity can adopt the traditional center of gravity calculation method, which will not be described here.

[0095] (2) Sort the warp yarns based on the horizontal coordinate and the vertical coordinate of the center of gravity in the main view to obtain a sorting result.

[0096] Specifically, when sorting, the warp yarns may be first divided into a plurality of columns according to the order of the ordinate, and then each column of the warp yarns may be marked according to the order of the abscissa.

[0097] (3) Marking the warp yarns according to the sorting result.

[0098] In this step, the sorting result is determined as the third material classification result. Figure 4 This is a schematic diagram of Example 1 of the third material classification result provided in this application. Please refer to Figure 4 , divide the warp yarns into two columns according to the horizontal axis, and then mark each column of warp yarns according to the vertical axis.

[0099] The 2.5-dimensional woven composite material microstructure recognition method based on machine vision provided in this embodiment first obtains the center of gravity of each bundle of warp yarns based on the edge of the warp yarns; then sorts the warp yarns based on the horizontal and vertical coordinates of the center of gravity in the main view of all three-view images to obtain the sorting result; finally, the sorting result is determined as the third material classification result. In this way, the warp yarns are sorted by the center of gravity of the warp yarns, and the third material classification result can be accurately obtained.

[0100] Furthermore, after obtaining the third material classification result, the second material classification result and the third material classification result are combined to obtain the recognition result of the microstructure of the target braided composite material. Figure 5 Schematic diagram of the identification results of the microstructure of the target woven composite material provided in this application. Please refer to Figure 5 In the recognition results, the warp yarn, weft yarn and pores can be clearly seen.

[0101] The 2.5-dimensional woven composite material microstructure recognition method based on machine vision provided in this embodiment first identifies the target woven composite material based on the top view and the structural tensor information recognition method of the target woven composite material to obtain the first material classification result, which includes at least warp, weft and pores, and then determines the pixel range of the warp and weft of the target woven composite material from the main view based on the first material classification result of the top view, and determines the number of pixels in each independent area based on the pixel range, thereby optimizing the first material classification result and obtaining the second material classification result, and then identifies the overlapping warp based on the main view recognition result and the side view, and obtains the third material classification result, and combines the second material classification result and the third material classification result to obtain the recognition result of the microstructure of the target woven composite material. In this way, the first material classification result combines the top view and the structural tensor information, and can extract rich structural features to ensure the accuracy and reliability of the first material classification result. Further, combined with the information of the number of pixels, the first material classification result can be accurately optimized based on morphology, ensuring the recognition effect of the second material classification result. Furthermore, by combining the recognition result of the main view and the information of the side view, the microstructure of the woven composite material can be effectively located based on the overlap, and the third material classification result can be obtained, thereby ensuring the accuracy of the third material classification recognition result. Furthermore, the final recognition result is obtained through the second material classification result and the third material classification result, thereby improving the accuracy of the recognition of the microstructure of the woven composite material, and the recognition result can be obtained accurately and efficiently. In general, the method and device provided by the present invention utilize the image information of each view in the three views, comprehensively consider the influence of each angle view of the material on the recognition result, and improve the accuracy of structure recognition; at the same time, in the process of using each view, the information determined by other views is used as a constraint, that is, the image information of each view does not exist independently, but exists in a mutually constrained relationship, thereby narrowing the scope that needs to be considered when using one view to recognize the structure, reducing the amount of calculation in the recognition process, and improving the recognition efficiency, while avoiding the mutual exclusion of the recognition results of each view, thereby improving the accuracy and utilization rate of the recognition results.

[0102] Figure 6 This is a flow chart of Example 2 of the method for identifying the microscopic structure of a 2.5-dimensional braided composite material based on machine vision provided in this application. Please refer to Figure 6 , based on the above embodiment, the step of marking the warp yarn according to the sorting result comprises:

[0103] S601, obtaining the minimum distance between the longitudinal coordinates of the gravity centers of the warp yarns of the previous image and the next image of the currently processed image.

[0104] In a specific implementation, for example, in one embodiment, the vertical coordinate of the previous warp yarn a of the currently processed image is compared with the vertical coordinate of the next warp yarn b of the currently processed image to obtain the corresponding distance, and the distances of all the warp yarns in the two images are traversed, and the minimum value therein is determined as the minimum distance corresponding to the two images.

[0105] S602: When the minimum distance is greater than a preset distance, determine that the warp yarn corresponding to the ordinate of the center of gravity in the image currently being processed is a new warp yarn, and determine the serial number of the new warp yarn.

[0106] Specifically, the specific value of the preset distance is determined according to actual needs, and is not limited in this embodiment. For example, in one embodiment, the specific value of the preset distance can be set to 10.

[0107] In specific implementation, for example, in one embodiment, combined with the above example, the distance 12 corresponding to the vertical coordinate of warp yarn a in the currently processed image is the minimum distance. When the distance exceeds the preset distance, warp yarn a is determined as a new warp yarn and is numbered.

[0108] S603, after processing all unnumbered warp yarns, the number of warp yarns is obtained according to the number of numbers.

[0109] In specific implementation, in this step, for example, after obtaining the new warp yarn a, if no other new warp yarns are found after processing all images, the original total number of warp yarns 20 is increased by one, and the total number of warp yarns is 21.

[0110] The 2.5-dimensional woven composite material microstructure recognition method based on machine vision provided in this embodiment first obtains the minimum distance between the longitudinal coordinates of the center of gravity of the warp yarns of the previous and next images being processed, and then when the minimum distance is greater than the preset distance, determines that the warp yarn corresponding to the longitudinal coordinate of the center of gravity in the image being processed is a new warp yarn, determines the number of the new warp yarn, and finally processes all unnumbered warp yarns, and obtains the number of warp yarns according to the number of numbers. In this way, the new warp yarn is confirmed by the minimum distance of the longitudinal coordinates in the two images, and the new warp yarn is marked, so that the warp yarns of the target woven composite material can be accurately sorted, which is convenient for identifying the microstructure of the woven composite material.

[0111] Figure 7 This is a flowchart of Example 3 of the method for identifying the microscopic structure of a 2.5-dimensional woven composite material based on machine vision provided in this application. Please refer to Figure 7 , the step after obtaining the recognition result of the target braided composite material mesostructure further includes:

[0112] S701. Determine the number and position of the warp yarns according to the recognition result, and obtain a spatial motion function of the warp yarns, wherein the spatial motion function is used to characterize the relationship between the material of the warp yarns and the fluctuation degree.

[0113] Specifically, the center of gravity of each warp yarn is used as a sampling point, and a curve fitting method is used to fit the spatial motion function of the warp yarn. In specific implementation, the spatial motion function of the warp yarn can be obtained by using a traditional spatial motion function acquisition method, or by using a neural network spatial motion function acquisition method, which is not limited in this embodiment.

[0114] Furthermore, the spatial motion function of the warp yarn is determined based on the material and fluctuation of the warp yarn. In specific implementation, for example, in one embodiment, the spatial motion function of the warp yarn is defined in combination with the material and fluctuation analysis of the warp yarn.

[0115] S702: construct a finite element model of the material based on the number and position of the warp yarns and a marching cube algorithm.

[0116] In specific implementation, for example, in one embodiment, the number and position of the warp yarns may be preprocessed to reduce noise and contrast, etc., and then the model may be reconstructed based on the marching cubes algorithm.

[0117] Furthermore, the number and position of the warp yarns can be traversed based on the moving cube graph algorithm, and the relationship between the warp yarns and the isosurface can be determined to construct a finite element model of the material.

[0118] S703: Establishing the unit coordinate system and coordinate position of each unit of the finite element model based on the spatial motion function.

[0119] Specifically, for each unit in each finite element model, a coordinate system of each unit is defined based on the spatial motion function, and the coordinate positions of key points representing the warp yarn positions are marked in the coordinate system.

[0120] S704: Modify the finite element model based on the unit coordinate system and coordinate position.

[0121] Specifically, according to each unit coordinate system and coordinate position, the finite element model is aligned with the path direction of the warp yarn, ensuring that the directionality of each finite element model is consistent with the actual weaving structure of the warp yarn to accurately reflect the actual spatial position and distribution of the warp yarn. Figure 8 Schematic diagram of the finite element model provided for this application. Please refer to Figure 8 By constructing a good finite element model, the location and shape of different warp yarns, weft yarns and pores can be determined.

[0122] The 2.5-dimensional woven composite material microstructure recognition method based on machine vision provided in this embodiment first determines the number and position of the warp yarns according to the recognition results, and obtains the spatial motion function of the warp yarns. The spatial motion function is used to characterize the relationship between the material and the fluctuation degree of the warp yarns, and then builds a finite element model of the material based on the number, position and moving cube algorithm of the warp yarns, and then builds the unit coordinate system and coordinate position of each unit of the finite element model based on the spatial motion function, and finally corrects the finite element model based on the unit coordinate system and coordinate position. In this way, a finite element model corresponding to the microstructure of the 2.5-dimensional woven composite material can be established to reflect the accurate prediction of the mechanical behavior of the real microstructure inside the material, and further improve the efficiency of model construction.

[0123] Corresponding to the aforementioned embodiment of the method for identifying the microstructure of a 2.5-dimensional woven composite material based on machine vision, the present application also provides an embodiment of a device for identifying the microstructure of a 2.5-dimensional woven composite material based on machine vision.

[0124] The embodiment of the 2.5-dimensional woven composite material microstructure recognition device based on machine vision in this application can be applied to a 2.5-dimensional woven composite material microstructure recognition device based on machine vision. The device embodiment can be implemented through software, or through hardware or a combination of software and hardware. Taking software implementation as an example, as a device in a logical sense, it is formed by the processor of the 2.5-dimensional woven composite material microstructure recognition device based on machine vision reading the corresponding computer program instructions in the non-volatile memory into the internal memory for execution. From a hardware perspective, if Fig. 9 As shown, it is a hardware structure diagram of the device where the 2.5-dimensional braided composite material microstructure recognition device based on machine vision is located in this application, except Fig. 9 In addition to the processor, memory, network interface, and non-volatile memory shown, the machine vision-based 2.5-dimensional woven composite material microstructure recognition device in which the device in the embodiment is located may also include other hardware according to the actual function of the machine vision-based 2.5-dimensional woven composite material microstructure recognition device, which will not be elaborated here.

[0125] Fig.10 This is a schematic diagram of the structure of the first embodiment of the 2.5-dimensional braided composite material microstructure recognition device based on machine vision provided by this application. Fig.10 , the device provided in this embodiment includes a processing module 1010, an optimization module 1020 and an identification module 1030;

[0126] The processing module 1010 is used to identify the target woven composite material based on the top view and the structural tensor information identification method of the target woven composite material, and obtain a first material classification result, wherein the first material classification result at least includes warp yarns, weft yarns and pores;

[0127] The optimization module 1020 is used to determine the pixel ranges of the warp yarns and the weft yarns of the target woven composite material from the main view respectively based on the first material classification result of the top view, and determine the number of pixels in each independent area based on the pixel range, so as to optimize the first material classification result and obtain the second material classification result;

[0128] The identification module 1030 is used to obtain a third material classification result based on the front view recognition result and the side view recognition result of the overlapping warp yarns, and to obtain the recognition result of the target woven composite material mesostructure by combining the second material classification result and the third material classification result;

[0129] Wherein, a preset number of control points are selected on the warp yarns and the weft yarns based on the main view, and the second material classification result is optimized based on the correspondence between the control points in the main view and those in the side view.

[0130] The device of this embodiment can be used to perform Figure 1 The steps, specific implementation principles and implementation processes of the method embodiment shown are similar and will not be repeated here.

[0131] Optionally, the identification module 1030 is specifically configured to obtain gradient information and direction information of the target woven composite material based on the top view;

[0132] The identification module 1030 is further specifically configured to classify the target braided composite material based on the gradient information and the direction information to obtain the first material classification result.

[0133] Optionally, the identification module 1030 is further specifically configured to determine the pixel ranges of the warp yarns and weft yarns of the target woven composite material from the main view respectively based on the first material classification result of the top view, and determine the number of pixels in each independent area based on the pixel range, so as to obtain a second material classification result;

[0134] The recognition module 1030 is further specifically configured to determine, based on the warp yarns and weft yarns recognized in the top view, a pixel range corresponding to the warp yarns and weft yarns in the main view as a recognition range in the main view;

[0135] The identification module 1030 is further specifically configured to obtain a second material classification result based on the identification range.

[0136] Optionally, the recognition module 1030 is further specifically configured to divide the recognition range of the main view into at least one independent area, and determine the independent area as a warp yarn category or a weft yarn category according to the number of warp yarn pixels of the warp yarn and the number of weft yarn pixels of the weft yarn in each independent area;

[0137] The optimization module 1020 is specifically configured to correct the first material classification result based on the warp yarn category and the weft yarn category to determine the second material classification result.

[0138] Optionally, the processing module 1010 is specifically configured to mark control points on the warp yarns and the weft yarns based on the relative positions of the warp yarns and the weft yarns on the main view;

[0139] The processing module 1010 is further specifically configured to map the control points on the main view to the side view to obtain the control points in the side view;

[0140] The identification module 1030 is further specifically configured to identify edge points of warp yarns having a degree of overlap greater than a threshold based on the control points in the side view;

[0141] The processing module 1010 is further specifically configured to use the edge point as a new edge point to determine the warp edge.

[0142] Optionally, the processing module 1010 is further specifically configured to calculate the center of gravity of each bundle of warp yarns based on the warp yarn edges;

[0143] The processing module 1010 is further specifically used to sort the warp yarns based on the horizontal coordinate and the vertical coordinate of the center of gravity in the main view to obtain a sorting result;

[0144] The processing module 1010 is further specifically configured to mark the warp yarns according to the sorting result.

[0145] Optionally, the processing module 1010 is further specifically configured to obtain a minimum distance between the longitudinal coordinates of the center of gravity of the warp yarns of the previous image and the next image of the currently processed image;

[0146] The processing module 1010 is further specifically configured to determine that the warp yarn corresponding to the ordinate of the center of gravity in the currently processed image is a new warp yarn when the minimum distance is greater than a preset distance, and determine the serial number of the new warp yarn;

[0147] The processing module 1010 is further specifically configured to process all unnumbered warp yarns and obtain the number of warp yarns according to the number of numbers.

[0148] Optionally, the target woven composite material is scanned based on an XCT device to obtain a plurality of groups of three-view images, wherein the plurality of groups of three-view images are derived from scanning results of the target woven composite material of different thicknesses.

[0149] Optionally, the processing module 1010 is further specifically used to determine the number and position of the warp yarns according to the recognition result, and obtain the spatial motion function of the warp yarns, wherein the spatial motion function is used to characterize the relationship between the material of the warp yarns and the fluctuation degree;

[0150] The processing module 1010 is further specifically used to construct a finite element model of the material based on the number and position of the warp yarns and a marching cube algorithm;

[0151] The processing module 1010 is further specifically used to establish a unit coordinate system and coordinate position of each unit of the finite element model based on the spatial motion function;

[0152] The processing module 1010 is further specifically configured to modify the finite element model based on the unit coordinate system and coordinate position.

[0153] Please continue to refer to Fig. 9 The present application also provides a 2.5-dimensional woven composite material microstructure recognition device based on machine vision, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of any one of the methods provided in the first aspect of the present application are implemented.

[0154] The present application also provides a computer-readable storage medium having a computer program stored thereon, and when the program is executed by a processor, the steps of any one of the methods provided in the present application are implemented.

[0155] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0156] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present application scheme. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0157] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A 2.5-dimensional woven composite material microstructure recognition method based on machine vision, characterized in that: The method comprises: Identify the target woven composite material based on the top view and the structural tensor information identification method of the target woven composite material to obtain a first material classification result, wherein the first material classification result at least includes warp yarns, weft yarns and pores; Acquire gradient information and direction information of the target woven composite material based on the top view; Based on the gradient information and the direction information, classify the target braided composite material to obtain the first material classification result; Determine the pixel ranges of the warp yarns and weft yarns of the target woven composite material from the main view based on the first material classification result of the top view, respectively, and determine the number of pixels in each independent area based on the pixel range, thereby optimizing the first material classification result to obtain a second material classification result; Based on the front view recognition result and the warp yarns that overlap with the side view recognition result, a third material classification result is obtained, and the recognition result of the microstructure of the target woven composite material is obtained by combining the second material classification result and the third material classification result; Based on the relative positions of the warp yarns and the weft yarns on the main view, marking control points on the warp yarns and the weft yarns; Mapping the control points on the main view to the side view to obtain the control points in the side view; Identifying edge points of warp yarns having a degree of overlap greater than a threshold based on control points in the side view; Using the edge point as a new edge point to determine the warp edge; Calculating the center of gravity of each warp yarn based on the warp yarn edges; Sorting the warp yarns based on the abscissa and ordinate of the center of gravity in the main view to obtain a sorting result; marking the warp yarns according to the sorting result; Wherein, a preset number of control points are selected on the warp yarns and the weft yarns based on the main view, and the second material classification result is optimized based on the correspondence between the control points in the main view and those in the side view.

2. The method according to claim 1, characterized in that The obtaining of the second material classification result comprises: Determine pixel ranges of the warp yarns and weft yarns of the target woven composite material from the main view based on the first material classification result of the top view, respectively, and determine the number of pixels in each independent area based on the pixel ranges, thereby obtaining a second material classification result; Based on the warp and weft yarns identified in the top view, determining the pixel range of the corresponding warp and weft yarns from the main view as the identification range in the main view; A second material classification result is obtained based on the identification range.

3. The method according to claim 2, characterized in that The method for obtaining a second material classification result based on the identification range includes: Dividing the recognition range of the main view into at least one independent area, and determining the independent area as a warp yarn category or a weft yarn category according to the number of warp yarn pixels of the warp yarn and the number of weft yarn pixels of the weft yarn in each independent area; The first material classification result is corrected based on the warp yarn category and the weft yarn category, and the second material classification result is determined.

4. The method according to claim 1, characterized in that The marking of the warp yarns according to the sorting result comprises: Get the minimum distance between the vertical coordinates of the center of gravity of the warp yarns of the previous and next images currently processed; When the minimum distance is greater than a preset distance, determining that the warp yarn corresponding to the ordinate of the center of gravity in the image currently being processed is a new warp yarn, and determining the number of the new warp yarn; After processing all unnumbered warp threads, get the number of warp threads according to the number of threads.

5. The method according to claim 1, characterized in that: Before the method for identifying the target woven composite material based on the top view and structure tensor information of the target woven composite material identifies the target woven composite material, the method further includes: The target woven composite material is scanned based on an XCT device to obtain multiple groups of three-view images, wherein the multiple groups of three-view images are derived from scanning results of the target woven composite material of different thicknesses.

6. The method according to claim 1, characterized in that After obtaining the recognition result of the target braided composite material microstructure, the method further includes: Determine the number and position of the warp yarns according to the recognition result, and obtain the spatial motion function of the warp yarns, wherein the spatial motion function is used to characterize the relationship between the material of the warp yarns and the fluctuation degree; constructing a finite element model of the material based on the number and position of the warp yarns and a marching cubes algorithm; Establishing the unit coordinate system and coordinate position of each unit of the finite element model based on the spatial motion function; The finite element model is modified based on the element coordinate system and coordinate positions.

7. A device for distinguishing the microstructure of a woven composite material, characterized in that: The device comprises a processing module, an optimization module and an identification module; wherein, The processing module is used to identify the target woven composite material based on the top view of the target woven composite material and the structural tensor information identification method, and obtain a first material classification result, wherein the first material classification result at least includes warp yarns, weft yarns and pores; The processing module is specifically used to obtain the gradient information and direction information of the target woven composite material based on the top view; The processing module is specifically used to classify the target braided composite material based on the gradient information and the direction information to obtain the first material classification result; The optimization module is used to determine the pixel ranges of the warp yarns and weft yarns of the target woven composite material from the main view respectively based on the first material classification result of the top view, and determine the number of pixels in each independent area based on the pixel range, so as to optimize the first material classification result and obtain the second material classification result; The recognition module is used to obtain a third material classification result based on the front view recognition result and the side view recognition result of the overlapping warp yarns, and obtain the recognition result of the target woven composite material mesostructure by combining the second material classification result and the third material classification result; The identification module is specifically used to mark the control points on the warp yarn and the weft yarn based on the relative positions of the warp yarn and the weft yarn on the main view; The recognition module is specifically used to map the control points on the main view to the side view to obtain the control points in the side view; The identification module is specifically used to identify edge points of warp yarns with a coincidence greater than a threshold value based on the control points in the side view; The identification module is specifically used to use the edge point as a new edge point to determine the warp edge; The identification module is specifically used to calculate the center of gravity of each bundle of warp yarns based on the warp yarn edges; The identification module is specifically used to sort the warp yarns based on the horizontal coordinate and the vertical coordinate of the center of gravity in the main view to obtain a sorting result; The identification module is specifically used to mark the warp yarn according to the sorting result; Wherein, a preset number of control points are selected on the warp yarns and the weft yarns based on the main view, and the second material classification result is optimized based on the correspondence between the control points in the main view and those in the side view.

Images