Non-contact strain gauge identification and coordinate accurate positioning method

Abstract

This invention discloses a non-contact strain gauge identification and precise coordinate positioning method, comprising the following steps: acquiring strain gauge image data; converting the strain gauge image data to grayscale and acquiring filtered strain gauge contour image data; determining the target strain gauge contour data; determining the essential matrix of the image acquisition device based on the image rotation matrix and translation vector; identifying the same strain gauge in the target strain gauge contour data based on the essential matrix, and performing coordinate positioning of the same strain gauge based on binocular vision, transforming it to the coordinates under the image acquisition device to obtain the coordinate information of the strain gauge under the image acquisition device; establishing the transformation relationship between the coordinate information and the target coordinate system, and transforming the coordinate information to the target coordinate system to obtain the strain gauge identification result in the target coordinate system; repeating steps S1-S5 to obtain the strain gauge identification results for all structural tests. This invention can improve the accuracy of strain gauge position coordinates.

Description

Technical Field

[0001] This invention relates to the field of structural testing technology, and specifically to a non-contact strain gauge identification and precise coordinate positioning method. Background Technology

[0002] Strain gauges are the primary means of measuring strain. In structural testing, a large number of strain gauges are often deployed to monitor the structural condition. However, traditional structural testing typically involves manually attaching strain gauges, which can lead to placement errors and a lack of coordinate verification, resulting in compromised measurement accuracy. In regions with high strain gradients within the structure, these placement errors can cause a significant drop in measurement accuracy. Therefore, developing effective strain gauge coordinate positioning methods is crucial for improving experimental accuracy.

[0003] Currently, coordinate measurement of strain gauges is mainly done manually with the aid of tools such as rulers, primarily using contact measurement methods. Due to the complexity and large size of the structure, problems such as difficult operation and inaccurate positioning are unavoidable. Summary of the Invention

[0004] In view of the above-mentioned shortcomings in the prior art, the present invention provides a non-contact strain gauge identification and precise coordinate positioning method, which solves the problem of low accuracy of strain gauge position coordinate identification results in the prior art.

[0005] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows: a non-contact strain gauge identification and precise coordinate positioning method, comprising the following steps:

[0006] S1. Strain gauge image data is acquired using an automated, controllable device equipped with an image acquisition unit, and rotational transformations are performed during the acquisition process to obtain strain gauge image data under different poses.

[0007] S2. The strain gauge image data is converted to grayscale to obtain strain gauge grayscale image data. The strain gauge grayscale image data is extracted to obtain strain gauge contour image data. The strain gauge contour image data is then filtered to obtain filtered strain gauge contour image data.

[0008] S3. Determine the target strain gauge profile data based on the filtered strain gauge profile image data and the strain gauge image data corresponding to the filtered strain gauge profile image data.

[0009] S4. Based on the image rotation matrix and translation vector, determine the essential matrix of the image acquisition device. Based on the essential matrix, identify the same strain gauge in the target strain gauge contour data, and perform coordinate positioning of the same strain gauge based on binocular vision. Transform the coordinates to the coordinates under the image acquisition device to obtain the coordinate information of the strain gauge under the image acquisition device.

[0010] S5. Establish the transformation relationship between coordinate information and target coordinate system, and transform the coordinate information into target coordinate system to obtain strain gauge identification results in target coordinate system;

[0011] S6. Repeat steps S1-S5 to obtain the strain gauge identification results for all structural tests.

[0012] The beneficial effects of the above scheme are as follows: This invention utilizes the geometric and color features of strain gauges for identification. By processing the acquired images in grayscale and color, information on the outline and color of the strain gauges is obtained respectively, and strain gauge identification is performed. Based on the pose relationship between images, the matching relationship of the same strain gauge between different images is obtained. Based on the binocular principle, coordinate positioning is performed, and finally the coordinates are transformed into the corresponding coordinate system to complete the non-contact positioning of the coordinates, thereby improving the accuracy of the strain gauge position coordinates.

[0013] Further, step S2 involves filtering the strain gauge contour image data, including:

[0014] S21. Determine the polygon measurement threshold based on the aspect ratio of the minimum bounding rectangle of the contour, the ratio of the area of ​​the minimum bounding rectangle of the contour to the area of ​​the contour, and the number of sides of the polygon fitted to the contour.

[0015] S22. Filter the strain gauge contour image data according to the polygon measurement threshold.

[0016] The beneficial effects of the above-mentioned further scheme are: by using polygon measurement thresholds, noise or non-target objects that do not conform to physical characteristics can be effectively eliminated, retaining only possible strain gauge contours, which helps improve the accuracy of subsequent processing and reduce false positives and false negatives. Setting a reasonable polygon measurement threshold can filter out a large number of irrelevant contours in the early stages, reducing unnecessary further processing, thereby saving computational resources and time, and improving the overall system processing efficiency.

[0017] Further, step S3 specifically includes:

[0018] S31. Overlay the selected strain gauge contour image data onto the strain gauge image data corresponding to the selected strain gauge contour image data, and segment the strain gauge image data corresponding to the selected strain gauge contour image data to obtain the segmented strain gauge image data.

[0019] S32. Obtain the average value of the three RGB color channels in the segmented strain gauge image data as the main color information of the contour region;

[0020] S33. Based on the main color information of the contour region, filter the segmented strain gauge image data to determine the target strain gauge contour data.

[0021] The beneficial effects of the above-mentioned further approach are as follows: By precisely covering and segmenting the original strain gauge image data with the filtered contours, fine-grained isolation of the strain gauge region is achieved, effectively eliminating background interference and ensuring that subsequent analysis focuses on the true strain gauge region, thus improving the accuracy and relevance of the analysis. Extracting the average value of the RGB three-channel colors of the segmented strain gauge region as the main color information of the contour region helps to identify and quantify the color characteristics of the strain gauge. Judging the strain state based on color changes enhances the dimensionality and depth of data analysis, helps to exclude areas that do not meet the requirements due to color deviation, and ensures the reliability and target nature of the analysis results.

[0022] Further, in step S22, the strain gauge contour image data is filtered using the following formula:

[0023]

[0024] Here, retain means retaining strain gauge profile image data that meets the polygon measurement threshold, and delete means deleting strain gauge profile image data that does not meet the polygon measurement threshold.

[0025] Further, in step S33, the segmented strain gauge image data is filtered based on the main color information of the contour region, using the following formula:

[0026]

[0027] Among them, (c h ,c s ,c v (r) represents the color information of the region within the contour in the segmented strain gauge image data. h ,r s ,r v () indicates the main color information of the outline area.

[0028] Furthermore, in step S4, the coordinates of the same strain gauge are located based on binocular vision, and the essential matrix E is:

[0029] E = T × R = [T] × R

[0030] Where T represents the translation vector, derived from t x t y t z Composition: R represents the image rotation matrix, which is 3x3, T×R represents the vector product, [T] × Let represent the antisymmetric matrix of the translation vector T, and t x t y t zThese represent the magnitudes of translation in the x, y, and z directions, respectively.

[0031] Furthermore, in step S4, based on the essential matrix, the same strain gauge in the target strain gauge contour data is identified using the following formula:

[0032]

[0033] Among them, M nm This represents the matching matrix of all strain gauges in the two images. The matrix represents the transpose of the coordinates of key points in the first target strain gauge profile data, where N represents the number of key points in the first target strain gauge profile data, n represents the nth key point in the first target strain gauge profile data, and p r =(u r ,v r ), p lm (m=1,2,...,M) represents the coordinates of key points in the profile data of the second target strain gauge, M represents the number of key points in the profile data of the second target strain gauge, and m represents the m-th key point in the profile data of the second target strain gauge. l =(u l ,v l The target strain gauge profile data includes the first target strain gauge profile data and the second target strain gauge profile data. Where E represents the essential matrix.

[0034] Furthermore, in step S4, the formula used to convert to coordinates under the image acquisition device is:

[0035]

[0036] (z ci z cj )=(C T C) -1 C T T

[0037]

[0038] Where A represents the camera intrinsic parameter matrix, f x f y s represents the parameters in the camera intrinsic parameter matrix, (u0, v0) represents the pixel value of the keypoint, C represents the process matrix consisting of pixel coordinates, camera intrinsic parameters, and rotation matrix, and R represents the rotation matrix between different camera positions. i ,v i (u) represents the pixel coordinates of key points of the strain gauge in the first target strain gauge contour data. j ,v j(z) represents the pixel coordinates of key points of the strain gauge in the profile data of the second target strain gauge. ci z cj ) represents the depth value of the strain gauge in different camera coordinate systems, T represents the translation vector between cameras, and y cj x represents the horizontal distance from the origin of the camera coordinate system. cj This represents the vertical distance from the origin of the camera coordinate system.

[0039] The beneficial effects of the above-mentioned further solutions are as follows: By calculating the essential matrix, the geometric relationship between the two camera coordinate systems can be accurately established. Combined with the principle of binocular stereo vision, the precise positioning of strain gauges in three-dimensional space can be achieved, significantly improving the accuracy and stability of spatial coordinate measurements. It supports automated positioning and tracking of strain gauges, reducing the need for manual intervention, improving monitoring efficiency, and making it suitable for large-scale, long-term structural health monitoring projects.

[0040] Furthermore, automated controllable devices include collaborative robots and drones; image acquisition devices include a single camera and two cameras with known pose relationships.

[0041] The beneficial effects of the aforementioned further solutions are: drones equipped with cameras can perform remote, efficient, and large-scale monitoring, significantly improving safety and work efficiency. Collaborative robots can perform detailed and precise inspections indoors or in specific locations. Binocular cameras provide depth information through parallax calculation, enabling three-dimensional reconstruction of target objects and providing richer visual information. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating a non-contact strain gauge identification and precise coordinate positioning method.

[0043] Figure 2 This is a schematic diagram of a device for a non-contact strain gauge identification and precise coordinate positioning method.

[0044] Figure 3 This is a schematic diagram of another device for a non-contact strain gauge identification and precise coordinate positioning method.

[0045] Figure 4 A schematic diagram of a polygon for measuring the threshold.

[0046] Figure 5 This is a schematic diagram of the process for identifying non-contact strain gauges. Detailed Implementation

[0047] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0048] like Figure 1As shown, a non-contact strain gauge identification and precise coordinate positioning method includes the following steps:

[0049] S1. Strain gauge image data is acquired by an automated controllable device equipped with an image acquisition device, and rotation transformation is performed during the acquisition process to obtain strain gauge image data under different poses.

[0050] In this embodiment, the automated controllable devices include collaborative robots and drones; the image acquisition devices include a single camera and two cameras with known pose relationships.

[0051] For example, this embodiment is conducted during a structural test. Furthermore, the strain gauge can be a uniaxial strain gauge, a strain flower, or other strain gauge with polygonal features.

[0052] For example, such as Figure 2 As shown, Figure 2 This is a schematic diagram of a device for a non-contact strain gauge identification and precise coordinate positioning method. (The diagram shows the device used for this method.) Figure 2 (a) is a schematic diagram of a small square tubular structure. Figure 2 (b) is a schematic diagram of an automated, controllable device equipped with an image acquisition device, corresponding to a small square tubular structure. Figure 2 In (a), a uniaxial strain gauge (such as) can be fixed on the surface of a small square tube structure. Figure 2 (as shown in the small square in (a)), and can be arranged around the small square tube structure as follows: Figure 2 The apparatus shown in (b) is used, and Figure 2 (b) shows the camera and lens used to acquire strain gauge image data of a uniaxial strain gauge.

[0053] For example, Figure 3 This is a schematic diagram of another device for a non-contact strain gauge identification and precise coordinate positioning method. Figure 3 In this context, the automated controllable equipment is a six-degree-of-freedom collaborative robot. Within this robot, appropriate image acquisition devices can be installed to collect strain gauge image data from different poses within the thin-walled cylindrical shell structure. Furthermore, Figure 3 The diagram on the right side shows the device facing the cylindrical shell structure. Figure 3 The left-hand side shows a schematic diagram of a six-degree-of-freedom collaborative robot corresponding to a cylindrical shell structure. Figure 3 In this design, uniaxial strain gauges can be fixed to the surface facing the cylindrical shell structure, and a six-degree-of-freedom collaborative robot can be arranged around the cylindrical shell structure. The camera and lens within the six-degree-of-freedom collaborative robot are used to acquire strain gauge image data from the uniaxial strain gauges. Figure 3The system includes a six-degree-of-freedom collaborative robot in two poses: the collaborative robot in its initial position and the collaborative robot after changing its pose. The pose change can be completed by the robotic arm of the six-degree-of-freedom collaborative robot.

[0054] S2. The strain gauge image data is converted to grayscale to obtain strain gauge grayscale image data. The strain gauge grayscale image data is extracted to obtain strain gauge contour image data. The strain gauge contour image data is then filtered to obtain filtered strain gauge contour image data.

[0055] In step S2 of this embodiment, the strain gauge contour image data is filtered, including:

[0056] S21. Determine the polygon measurement threshold based on the aspect ratio of the minimum bounding rectangle of the contour, the ratio of the area of ​​the minimum bounding rectangle of the contour to the area of ​​the contour, and the number of sides of the polygon fitted to the contour.

[0057] S22. Filter the strain gauge contour image data according to the polygon measurement threshold.

[0058] In step S22 of this embodiment, the strain gauge contour image data is filtered using the following formula:

[0059]

[0060] Here, retain means retaining strain gauge profile image data that meets the polygon measurement threshold, and delete means deleting strain gauge profile image data that does not meet the polygon measurement threshold.

[0061] This embodiment filters out interfering contours and reduces computational load by screening strain gauge contour image data.

[0062] Figure 4 This is a schematic diagram illustrating the threshold measurement using polygons. Wherein, Figure 4 (a) is a schematic diagram that uses the aspect ratio of the smallest bounding rectangle of the outline as the threshold for polygon measurement. Figure 4 (b) is a schematic diagram of a polygon measurement threshold using the ratio of the area of ​​the smallest bounding rectangle to the area of ​​the outline.

[0063] Optionally, the filtering of strain gauge contour image data in step S2 can be omitted. That is, in step S2, the strain gauge image data can be converted to grayscale to obtain strain gauge grayscale image data, the strain gauge grayscale image data can be extracted to obtain strain gauge contour image data, and then step S3 can be executed.

[0064] S3. Determine the target strain gauge profile data based on the filtered strain gauge profile image data and the strain gauge image data corresponding to the filtered strain gauge profile image data.

[0065] In this embodiment, step S3 specifically includes:

[0066] S31. Overlay the selected strain gauge contour image data onto the strain gauge image data corresponding to the selected strain gauge contour image data, and segment the strain gauge image data corresponding to the selected strain gauge contour image data to obtain the segmented strain gauge image data.

[0067] S32. Obtain the average value of the three RGB color channels in the segmented strain gauge image data as the main color information of the contour region;

[0068] S33. Based on the main color information of the contour region, filter the segmented strain gauge image data to determine the target strain gauge contour data.

[0069] In step S33 of this embodiment, the segmented strain gauge image data is filtered based on the main color information of the contour region. The formula used is:

[0070]

[0071] Among them, (c h ,c s ,c v (r) represents the color information of the region within the contour in the segmented strain gauge image data. h ,r s ,r v () indicates the main color information of the outline area.

[0072] Optionally, in step S32, the k-means clustering method can be used to cluster the segmented strain gauge image data to obtain the main color information of the contour region; no specific restrictions are imposed here.

[0073] Optionally, in step S32, the image format of the segmented strain gauge image data can be RGB image, HSV image, or HSL image, without any specific restrictions.

[0074] S4. Based on the image rotation matrix and translation vector, determine the essential matrix of the image acquisition device. Based on the essential matrix, identify the same strain gauge in the target strain gauge contour data, and perform coordinate positioning of the same strain gauge based on binocular vision. Transform the coordinates to the coordinates under the image acquisition device to obtain the coordinate information of the strain gauge under the image acquisition device.

[0075] In step S4 of this embodiment, the same strain gauge is located based on binocular vision, and the essential matrix E is:

[0076] E = T × R = [T] × R

[0077] Where T represents the translation vector, derived from t x t y t z Composition: R represents the image rotation matrix, which is 3x3, T×R represents the vector product, [T] × Let represent the antisymmetric matrix of the translation vector T, and t x t y t z These represent the magnitudes of translation in the x, y, and z directions, respectively.

[0078] Based on the essential matrix, the formula used to identify the same strain gauge in the target strain gauge contour data is:

[0079]

[0080] Among them, M nm This represents the matching matrix of all strain gauges in the two images. The matrix represents the transpose of the coordinates of key points in the first target strain gauge profile data, where N represents the number of key points in the first target strain gauge profile data, n represents the nth key point in the first target strain gauge profile data, and p r =(u r ,v r ), p lm (m=1,2,...,M) represents the coordinates of key points in the profile data of the second target strain gauge, M represents the number of key points in the profile data of the second target strain gauge, and m represents the m-th key point in the profile data of the second target strain gauge. l =(u l ,v l The target strain gauge profile data includes the first target strain gauge profile data and the second target strain gauge profile data. E represents the essential matrix.

[0081] The formula used to convert to coordinates under the image acquisition device is:

[0082]

[0083] (z ci z cj )=(C T C) -1 C T T

[0084]

[0085] Where A represents the camera intrinsic parameter matrix, f x f ys represents the parameters in the camera intrinsic parameter matrix, (u0, v0) represents the pixel value of the keypoint, C represents the process matrix consisting of pixel coordinates, camera intrinsic parameters, and rotation matrix, and R represents the rotation matrix between different camera positions. i ,v i (u) represents the pixel coordinates of key points of the strain gauge in the first target strain gauge contour data. j ,v j (z) represents the pixel coordinates of key points of the strain gauge in the profile data of the second target strain gauge. ci z cj ) represents the depth value of the strain gauge in different camera coordinate systems, T represents the translation vector between cameras, and y cj x represents the horizontal distance from the origin of the camera coordinate system. cj This represents the vertical distance from the origin of the camera coordinate system.

[0086] Figure 5 This is a schematic diagram illustrating the process of non-contact strain gauge identification. Figure 5 (a) is a schematic diagram of strain gauge image data using HSV color images. Figure 5 (b) is a schematic diagram of the process of filtering strain gauge contour image data. Figure 5 (c) Schematic diagram of the process for determining the target strain gauge profile data.

[0087] Optionally, the first target strain gauge contour data can be any target strain gauge contour data, and the second target strain gauge contour data can also be any target strain gauge contour data. Furthermore, based on the essential matrix, identifying the same strain gauge in the target strain gauge contour data can be achieved using two images from the target strain gauge contour data, multiple images (more than 2) from the target strain gauge contour data, or multiple images (more than 2) from the target strain gauge contour data.

[0088] Optionally, when using multiple (more than 2) images from the target strain gauge contour data to identify the strain gauge, the images can be combined in pairs first, and then the least squares of the multiple positioning results can be calculated to obtain the final identification result.

[0089] S5. Establish the transformation relationship between coordinate information and target coordinate system, and transform the coordinate information into target coordinate system to obtain strain gauge identification results in target coordinate system.

[0090] Optionally, the target coordinate system can refer to the experimental structure coordinate system. The transformation relationship between the coordinate information and the target coordinate system can be obtained by measurement or calculated from the markers with known spatial relationships in the image.

[0091] S6. Repeat steps S1-S5 to obtain the strain gauge identification results for all structural tests.

[0092] In this embodiment, non-contact positioning of strain gauge coordinates is achieved based on image processing and visual positioning principles, effectively improving positioning accuracy to 1 mm. Compared with traditional contact positioning methods, this embodiment has a high degree of automation and is easy to operate, enabling rapid positioning of strain gauges for large and complex structures. It shortens the test preparation cycle and improves test accuracy while ensuring accuracy.

[0093] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of the invention.

Claims

1. A non-contact strain gauge identification and precise coordinate positioning method, characterized in that, The method includes: S1. Strain gauge image data is acquired using an automated, controllable device equipped with an image acquisition unit. During the acquisition process, rotation and translation transformations are performed to obtain strain gauge image data under different poses. S2. The strain gauge image data is converted to grayscale to obtain strain gauge grayscale image data. The strain gauge grayscale image data is extracted to obtain strain gauge contour image data. The strain gauge contour image data is then filtered to obtain filtered strain gauge contour image data. S3. Determine the target strain gauge contour data based on the filtered strain gauge contour image data and the strain gauge image data corresponding to the filtered strain gauge contour image data; S4. Based on the image rotation matrix and translation vector, determine the essential matrix of the image acquisition device. Based on the essential matrix, identify the same strain gauge in the target strain gauge contour data, and perform coordinate positioning of the same strain gauge based on binocular vision, transforming it to the coordinates under the image acquisition device to obtain the coordinate information of the strain gauge under the image acquisition device. S5. Establish the transformation relationship between the coordinate information and the target coordinate system, and transform the coordinate information into the target coordinate system to obtain the strain gauge identification result in the target coordinate system; S6. Repeat steps S1-S5 to obtain the strain gauge identification results for all structural tests.

2. The method according to claim 1, characterized in that, In step S2, the filtering of the strain gauge contour image data includes: S21. Determine the polygon measurement threshold based on the aspect ratio of the minimum bounding rectangle of the contour, the ratio of the area of ​​the minimum bounding rectangle of the contour to the area of ​​the contour, and the number of sides of the contour-fitted polygon. S22. Filter the strain gauge contour image data according to the polygon measurement threshold.

3. The method according to claim 1, characterized in that, Step S3 specifically includes: S31. Overlay the selected strain gauge contour image data onto the strain gauge image data corresponding to the selected strain gauge contour image data, and segment the strain gauge image data corresponding to the selected strain gauge contour image data to obtain segmented strain gauge image data. S32. Obtain the average value of the RGB color three channels in the segmented strain gauge image data as the main color information of the contour region; S33. Based on the main color information of the contour region, the segmented strain gauge image data is filtered to determine the target strain gauge contour data.

4. The method according to claim 2, characterized in that, In step S22, the formula used to filter the strain gauge contour image data is: in, This indicates that strain gauge profile image data that meets the polygon measurement threshold are retained. This indicates that strain gauge profile image data that does not meet the polygon measurement threshold will be deleted.

5. The method according to claim 3, characterized in that, In step S33, the segmented strain gauge image data is filtered based on the main color information of the contour region using the following formula: in, This represents the color information of the region within the contour in the segmented strain gauge image data. This indicates the main color information of the outline area.

6. The method according to claim 1, characterized in that, In step S4, the coordinate positioning of the same strain gauge based on binocular vision is essentially a matrix. for: in, Represents the translation vector, by composition, This represents the image rotation matrix, which is 3x3 in size. Represents the vector product. Represents the translation vector The antisymmetric matrix, and , These represent the magnitudes of translation in the x, y, and z directions, respectively.

7. The method according to claim 1, characterized in that, In step S4, the formula used to identify the same strain gauge in the target strain gauge contour data based on the essential matrix is: in, This represents the matching matrix of all strain gauges in the two images. The transpose matrix representing the coordinates of key points in the first target strain gauge profile data. This indicates the number of key points in the profile data of the first target strain gauge. This indicates the first target strain gauge profile data. One key point, , This represents the coordinates of key points in the profile data of the second target strain gauge. This indicates the number of key points in the profile data of the second target strain gauge. This indicates the second target strain gauge profile data. One key point, Furthermore, the target strain gauge profile data includes the first target strain gauge profile data and the second target strain gauge profile data. ,in, This represents the essential matrix.

8. The method according to claim 1, characterized in that, In step S4, the formula used for converting to coordinates under the image acquisition device is: in, Represents the camera intrinsic parameter matrix. This represents the parameters in the camera intrinsic parameter matrix. This represents the pixel value of a key point within a key point. This represents the process matrix consisting of pixel coordinates, camera intrinsic parameters, and rotation matrix. This represents the rotation matrix between different camera positions. This represents the pixel coordinates of key points of the strain gauge in the first target strain gauge profile data. This represents the pixel coordinates of key points of the strain gauge in the profile data of the second target strain gauge. This represents the depth value of the strain gauge in different camera coordinate systems. This represents the translation vector between cameras. This represents the horizontal distance from the origin of the camera coordinate system. This represents the vertical distance from the origin of the camera coordinate system.

9. The method according to claim 1, characterized in that, The automated controllable equipment includes collaborative robots and drones; the image acquisition equipment includes a single camera and two cameras with known pose relationships.

Images