A photoelastic test force chain information automatic extraction method

Abstract

The application discloses a kind of photoelastic test force chain information automatic extraction method, the same test state under the particle image is obtained by polarizing mirror and non-polarizing mirror respectively, and image preprocessing is carried out;For the particle image of non-polarizing mirror, based on Hough transform and Euclidean distance, the particle center coordinates, particle radius, interparticle contact point position and interparticle contact angle are extracted;For the particle image of polarizing mirror, based on the mean square value of color gradient, the contact force information of particle is extracted by mask operation and calibration curve;Finally, based on the composition criterion of force chain and its collimation characteristic, angle threshold is converted into distance threshold, and the force chain information of photoelastic test particle image is extracted.The force chain information automatic extraction method established by the application solves the problem of accurate identification of force chain branch point, distinguishes the strong and weak force chain type, identifies the macroscopic distribution state of force chain, force chain length, force chain quantity and force chain orientation and other force chain network structure information.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital image processing, in particular to a photoelastic test force chain information automatic extraction method. BACKGROUND

[0002] Particulate matter is a complex system formed by a large number of discrete solid particles interacting with each other. In rock mass engineering, engineering geological disasters such as landslides, tailings dam collapses, debris flows, tunnel collapses, and chute blockages caused by the flow of rock particle media seriously threaten the safety of construction personnel and equipment. Revealing the macroscopic mechanical behavior and microscopic mechanical mechanism of the rock particle medium system is the basis for disaster prediction and protection. The macroscopic and microscopic mechanical properties of the particle medium and the contact force transmission path are often characterized by the force chain parameters of the photoelastic test. The quantification and extraction of photoelastic force chain information is a key link in its research, and has important scientific and engineering significance.

[0003] Photoelastic test is a stress measurement method based on optics. As early as 1957, researchers first used photoelastic test technology to clearly observe the force transmission light band in the particle medium. The force transmission light band was defined as a force chain in 1995. In the following 20 years, many scholars have carried out a large number of researches on force chains using photoelastic test. For example, some literature shows that some scholars quantified the contact force between particles in the process of top coal mining, and described the formation mechanism of the mine pressure at the working face; some scholars studied the soil arching effect of particulate matter and discussed the transmission of force chains in the inner arch and outer arch; in addition, some people analyzed the role of particulate matter in plugging stratum cracks and discussed the evolution law of strong and weak force chains. However, the application number CN202210609208.8, entitled "Asphalt mixture load transmission behavior evaluation method based on force chain characteristic quantization", obtained the contact information between asphalt mixtures through numerical simulation data, and completed the evaluation of the force transmission behavior in the asphalt mixture from the perspective of force chains, but the method of the invention is not applicable to the extraction of force chain information in actual physical photoelastic test.

[0004] The above research results have important significance for promoting the development of particulate matter mechanics. However, through analysis, it is found that the existing researches on the extraction and analysis of force chain information in photoelastic test mainly focus on the macroscopic distribution characteristics of force chains, the quantification of contact forces, and the differentiation of strong and weak force chains. Moreover, the accurate identification of each force chain branch point, the quantitative extraction method of single force chain information, and the dynamic evolution law of force chains are still less studied. SUMMARY

[0005] The purpose of the present application is to provide a photoelastic test force chain information automatic extraction method to overcome the problem of accurate identification of force chain branch points in photoelastic test, realize the quantitative extraction of force chain information, and promote the development of particulate matter mechanics.

[0006] To solve the above problems, the technical scheme adopted by the present application is:

[0007] A photoelastic test force chain information automatic extraction method, comprising the following steps:

[0008] (1) Obtain the particle images under the same test state through the polarizer and the non-polarizer respectively, and pre-process the images;

[0009] (2) Extract the particle geometric information: perform Hough transform and Euclidean distance calculation on the pre-processed photoelastic image of the non-polarizer, and extract the particle center coordinates, particle radius, particle inter-contact point position and particle inter-contact angle;

[0010] (3) Extract the particle contact force information: based on the identified particle geometric information, perform mask operation on the image of the polarizer to obtain the color gradient mean square value of the particles; then, combine the calibration curve of the contact force and the color gradient mean square value to extract the contact force information of the particles;

[0011] (4) Extract the force chain network structure information: combine the particle geometric information, contact force information and force chain composition criterion, convert the angle threshold value into a distance threshold value, and then distinguish the strong and weak force chain types, extract the macroscopic distribution state of the force chain, and identify the force chain length, force chain quantity and force chain orientation information.

[0012] The particle image pre-processing of the non-polarizer is performed according to the following method: the particle image of the non-polarizer is binarized, eroded, dilated, edge detected and hole filled.

[0013] The particle image pre-processing of the polarizer is to filter the particle image of the polarizer.

[0014] The force chain composition criterion includes a strong and weak force chain criterion, a force chain length criterion and a distance threshold criterion, the strong and weak force chain criterion is: when the particle contact force F is greater than or equal to the average contact force of the particle system, it is determined that the particle is located on the strong force chain; otherwise, it is determined that the particle is located on the weak force chain or in a state not affected by external force.

[0015] The force chain length criterion is that the number of particles on a force chain is the length of the force chain, which must be greater than or equal to 3, that is, the number of contact points must be greater than or equal to 2.

[0016] The distance threshold criterion is that the distance d between two adjacent contact points must be greater than or equal to the distance threshold D c , and the calculation formula of the distance threshold is as follows:

[0017] D c =(2-0.0002358*θ c2.04 )*r

[0018] where D c is the distance threshold; θ c is the angle threshold, the smaller the value, the more straight the force chain; r is the particle radius.

[0019] The specific process of step (4) is as follows:

[0020] 4-1: input the identified particle geometric information and particle contact force information, and calculate the average contact force of the particle system

[0021]

[0022] 4-2: generate an empty matrix C all , C1, C2, C3, C4, C fc , for storing subsequent data;

[0023] 4-3: select the strong contact particle information of and store it in C all , and sort it by the contact point coordinate y value from large to small;

[0024] 4-4: determine whether the number of contact points in C all is greater than or equal to 2: if yes, select a starting contact point a, and search the contact points a i around a, and store the contact points a and the search results a i in C1; if no, end step (4);

[0025] 4-5: determine whether the number of contact points in C1 is greater than or equal to 2: if yes, copy C1 to C2; if no, delete the intersection of C1 from C all , empty C1, and jump to step 4-4;

[0026] 4-6: delete the contact point a i from C1 which forms 4 particles with the starting contact point a;

[0027] 4-7: delete the intersection of C1 and C4 from C1;

[0028] 4-8: select the contact point a i farthest from the starting contact point a from C1, and calculate the distance between the two contact points and assign it as d;

[0029] 4-9: determine whether the distance d is greater than or equal to the distance threshold D c : if yes, store the contact points a and a i in C3, and set the new starting contact point as a i, C2 is copied as C4, C1 and C2 are emptied, and jump to step 4-4; if no, the following steps are executed:

[0030] 4-9-1: judge whether the number of contact points in C3 is greater than or equal to 2: if yes, copy C3 as C fc , and go to step 4-9-2; if no, go to step 4-9-2 directly;

[0031] 4-9-2: delete the contact point a from C all , and delete the intersection with C3 from C all ;

[0032] 4-9-3: empty C1, C2, C3 and C4, and jump to step 4-4 to start the retrieval of a new force chain information.

[0033] The beneficial effects produced by the above technical solutions are that the light test force chain information automatic extraction method provided by the application can accurately extract the force chain network from the physical light test image, dynamically depict the force chain characteristics of the particulate matter, and has significant advantages in the extraction of the geometric information of the particles, the discrimination of the force chain branch points, the retrieval of the force chain, the distribution of the force chain and the representation of the azimuth angle. The research results provide a scientific basis for the research on the macroscopic mechanical behavior and the microscopic mechanical mechanism of the particulate matter. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 It is the overall flowchart of the application.

[0035] Figure 2 It is the force chain network structure information automatic extraction flowchart based on the chain formation criterion of the application.

[0036] Figure 3 It is the physical model diagram of the light test provided by the embodiment of the application.

[0037] Figure 4 It is the local enlarged diagram of the physical image of the light test provided by the embodiment of the application, wherein a is the light test image under the polarizer, and b is the light test image without the polarizer.

[0038] Figure 5 It is the pre-processing result diagram of the physical image of the light test provided by the embodiment of the application.

[0039] Figure 6 It is the geometric information extraction result diagram of the physical image of the light test provided by the embodiment of the application, wherein a is the particle recognition result, and b is the contact network recognition result.

[0040] Figure 7 It is the calibration curve diagram of the contact force F and the color gradient mean square value provided by the embodiment of the application.

[0041] Figure 8 This is a schematic diagram of contact point screening and force chain generation provided in an embodiment of the present invention.

[0042] Figure 9 A partial view of the strong chain information extraction results provided in an embodiment of the present invention.

[0043] Figure 10 The diagram illustrates the evolution of contact force information provided in this embodiment of the invention.

[0044] Figure 11 The image provided in this embodiment of the invention is a physical image of the photoelastic test and the result of force chain information extraction during the 10th ore discharge test. In this image, a is the overall physical image of the photoelastic test, b is the result of force chain network information extraction, c is the result of strong force chain azimuth angle extraction, and d is the result of weak force chain azimuth angle extraction. Detailed Implementation

[0045] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0046] This embodiment takes a physical photoelastic test under a large-scale ore-feeding, synchronous backfilling, pillarless ore-holding mining method as an example. The method of this invention was used to automatically extract the force chain information from the photoelastic test. The overall process is as follows: Figure 1 As shown, the specific steps are as follows:

[0047] Step 1: Acquire particle images under the same experimental conditions using both a polarizing filter and an unpolarizing filter, and preprocess the images. The two input particle images from the same photoelastic experiment, one with a polarizing filter and one without, are shown below. Figure 4 As shown, where Figure 4 Image a is a photoelastic experiment image under a polarizing microscope. Figure 4 b is an image of the photoelastic experiment without a polarizing mirror. In the... Figure 4 In the morphological preprocessing of b, the binarization threshold was 0.125, the area of ​​small blemishes deleted was 1300 pixels, and the edge detection operator used was the Canny operator. The preprocessing result is as follows: Figure 5 As shown.

[0048] Step 2: Extract particle geometry information: Perform Hough transform and Euclidean distance calculation on the preprocessed photoelastic image without a polarizing filter to extract the particle center coordinates, particle radius, contact point positions between particles, and contact angles between particles. The correction coefficient for distance discrimination is 1.05, and the contact angle range is -90° to 90°. The extraction results are shown below. Figure 6 As shown.

[0049] Step 3: Extracting Particle Contact Force Information: Based on the identified particle geometry, a masking operation is performed on the image with a polarizing filter to obtain the mean square value of the particle's color gradient. Then, combined with the calibration curve of contact force versus mean square value of color gradient, the particle's contact force information is extracted. The calibration curve of contact force versus mean square value of color gradient is shown below. Figure 7 As shown.

[0050] Step 4: Extract force chain network structure information: Combining the geometric information of particles, contact force information and force chain composition criteria, the angle threshold is converted into a distance threshold, thereby identifying strong and weak force chain types, extracting the macroscopic distribution state of force chains, and identifying the force chain length, number of force chains and force chain orientation information.

[0051] The criteria for the formation of a force chain include:

[0052] (1) Criterion for strong and weak force chains: When the particle contact force F is greater than or equal to the average contact force F in the particle system, the particle is determined to be located on a strong force chain; otherwise, the particle is determined to be located on a weak force chain or in a state where it is not acted upon by external forces.

[0053] (2) Criterion for force chain length: The number of particles on a force chain is the length of the force chain, and its value must be greater than or equal to 3, that is, the number of contact points must be greater than or equal to 2.

[0054] (3) Distance threshold criterion: The distance d between two adjacent contact points must be greater than or equal to the distance threshold D. c The formula for calculating the distance threshold is as follows:

[0055] D c =(2-0.0002358*θ) c 2.04 )*r

[0056] Among them, D c θ is the distance threshold; c is the angle threshold; the smaller the value, the straighter the force chain; r is the particle radius. In this embodiment, the distance threshold is 1.66r.

[0057] In step 4, the automatic extraction process of force chain network structure information based on chain formation criteria is as follows: Figure 2 As shown, the specific steps are as follows (using...). Figure 8 For example:

[0058] Step 4-1: Input the identified particle geometry and particle contact force information, and calculate the average contact force F of the particle system.

[0059] Step 4-2: Generate an empty matrix C all ,C1,C2,C3,C4,C fc This is used for storing subsequent data.

[0060] Step 4-3: Selecting out Strong contact particle information is stored in C all , and is ordered from large to small according to the contact point coordinate y value. If weak contact particle information is to be selected out, the discrimination condition is changed to , and the discrimination condition is changed to all . At this time, the contact point serial numbers in C

[0061] Step 4-4: Discriminating whether the number of contact points in C all is greater than or equal to 2. If yes, a starting contact point a is selected, and the contact points a i around a are searched, and the contact points a i and the search results a i are stored in C1. If no, the step 4 is ended. In this embodiment, the contact point search distance is 1.1r.

[0062] Since the calculation result of a is relatively simple, it is now assumed that the first search has been completed, and the starting contact point a is 3, and the relevant results are shown in Figure 8 b. At this time, the starting search point a is 3, and the peripheral contact points a all searched are 1, 2, 4, 5, 6, 7, 8, 9, and 10. The contact point serial numbers in C1 are 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10. The contact point serial numbers in C4 are 1, 2, 3, and 4.

[0063] Step 4-5: Discriminating whether the number of contact points in C1 is greater than or equal to 2. If yes, C1 is copied as C2. If no, the intersection with C1 is deleted from C i , C1 is emptied, and step 4-4 is jumped to. At this time, the contact point serial numbers in C2 are 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10.

[0064] Step 4-6: Deleting the contact points a i constituted by 4 particles with the starting contact point a from C1, that is, deleting 5 and 10. At this time, the contact point serial numbers in C1 are 1, 2, 3, 4, 6, 7, 8, and 9.

[0065] Step 4-7: Deleting the intersection with C4 from C1, that is, deleting 1, 2, 3, and 4. At this time, the contact point serial numbers in C1 are 6, 7, 8, and 9.

[0066] Step 4-8: Selecting the contact point a iand the distance between the two contact points is calculated and assigned as d. At this time, the initial contact point a is No. 3 and the farthest contact point a i is No. 7 and the distance d is 2r.

[0067] Step 4-9: Determine whether the distance d is greater than or equal to the distance threshold D c If yes, the contact points a and a i are stored in C3 and the new initial contact point is a i , C2 is copied as C4, C1 and C2 are emptied, and the process jumps to Step 4-4. In this embodiment, d = 2r > D c = 1.66r. At this time, the contact point serial numbers in C3 are No. 1, No. 3 and No. 7. The new initial contact point is No. 7. The contact point serial numbers in C4 are No. 1, No. 2, No. 3, No. 4, No. 5, No. 6, No. 7, No. 8, No. 9 and No. 10.

[0068] Since the judgment is consistent in each loop process, no more details are given here. The relevant loop search results are shown in Table 1 Figure 8 , Table 2 Figure 8 and Table 3 Figures 9 to 11 . The final identification results are shown in Table 4

[0069] Through this embodiment, it is shown that the light experiment force chain information automatic extraction method provided by the present application can accurately extract the force chain network from the physical light experiment image, dynamically depict the force chain characteristics of the particulate matter, and has significant advantages in the aspects of particulate geometric information extraction, force chain branch point discrimination, force chain search, force chain distribution and azimuth angle representation. The research results provide a scientific basis for the research on the macroscopic mechanical behavior and microscopic mechanical mechanism of the particulate matter.

[0070] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope defined by the claims of the present application.

Claims

1. A photoelastic test force chain information automatic extraction method, characterized by, The method comprises the following steps: (1) obtaining particle images under the same test state by using polarized and non-polarized mirrors respectively, and pre-processing the images; (2) extracting particle geometric information: performing Hough transform and Euclidean distance calculation on the pre-processed photoelastic image without polarization to extract particle center coordinates, particle radius, particle inter-contact point position and particle inter-contact angle; (3) extracting particle contact force information: based on the identified particle geometric information, performing mask operation on the image with polarization to obtain the color gradient mean square value of the particles; then combining the calibration curve of the contact force and the color gradient mean square value to extract the contact force information of the particles; (4) extracting force chain network structure information: combining the particle geometric information, contact force information and force chain composition criterion, converting the angle threshold value into a distance threshold value, and then identifying the strong and weak force chain types, extracting the macroscopic distribution state of the force chain, and identifying the force chain length, force chain number and force chain orientation information; The constitutive criterion of the force chain includes a strong and weak force chain criterion, a force chain length criterion, and a distance threshold criterion. The strong and weak force chain criterion is that when the particle contact force is greater than or equal to the average contact force in the particle system, it is determined that the particle is on a strong force chain; otherwise, it is determined that the particle is on a weak force chain or in a state not affected by external force. ​​ The specific process of step (4) is as follows: 4-1: input the identified particle geometry information and particle contact force information, and calculate the average contact force of the particle system ; 4-2: Generate empty matrix for storage of subsequent data; 4-3: Selecting out The strong contact particle information is stored in The contact point coordinates are sorted from large to small. 4-4: Discrimination Is the number of contact points greater than or equal to 1? If so, select a starting contact point. and to Surrounding contact points Perform a search to find the contact points. and search results Stored in If not, then end step (4); 4-5: Discrimination Is the number of contact points greater than or equal to 1? If so, then Make a copy If not, then from Delete with The intersection, clear Jump to step ; 4-6: from the initial contact point contact points constituting 4 particles ; 4-7: from the intersection of the intersection of​ 4-8: from the start contact point the farthest contact point and calculate the distance between the two contact points, assigned as ; 4-9: discriminant distance whether greater than or equal to distance threshold if yes, then contact point and stored in , another new starting contact point is , copy is , clear and , and jump to step ; if no, then perform the following steps: 4-9-1: Discrimination Is the number of contact points greater than or equal to 1? If so, then Copy as And proceed to step If not, proceed directly to step [number]. ; 4-9-2: from contact points , from intersection with ; 4-9-3: empty and jump to step and start a new search for force chain information.

2. The method according to claim 1, wherein The particle image pre-processing of the non-polarized mirror is performed according to the following method: the particle image of the non-polarized mirror is binarized, eroded, dilated, edge detected and hole filled.

3. The method of claim 1, wherein the method comprises: The particle image pre-processing of the polarized mirror is filtering the particle image of the polarized mirror.

4. The method of claim 1, wherein, The force chain length criterion is that the number of particles on a force chain is the force chain length, which must be greater than or equal to 3, that is, the number of contact points must be greater than or equal to 2.

5. The method of claim 1, wherein, The distance threshold criterion is that the distance between two adjacent contact points must be greater than or equal to a distance threshold The distance threshold is calculated as follows: ， wherein, is a distance threshold value; is an angle threshold value, the smaller the value, the straighter the force chain; is a particle radius.

Images