Needle sheet content detection method and system based on computer vision

Abstract

The invention relates to the technical field of material detection, in particular to a needle-sheet-shaped content detection system based on computer vision and a use method of the needle-sheet-shaped content detection system. Comprising the following steps: providing to-be-detected coarse aggregate, and paving the coarse aggregate on a platform surface; shooting the coarse aggregate on the platform surface to obtain image information; identifying the image information to obtain size information of all coarse aggregates in the image information; performing three-dimensional modeling based on the size information of all coarse aggregates in the obtained image information to obtain a three-dimensional model; and analyzing the three-dimensional model, marking the needle-sheet-shaped particles, and calculating the content of the needle-sheet-shaped particles. The device has the beneficial effects that the coarse aggregate is flatly laid on the platform surface, image information is obtained through shooting, three-dimensional modeling is carried out according to the size of the coarse aggregate in the image information, then the coarse aggregate is analyzed through the three-dimensional model, the needle-sheet-shaped particles are marked, and the content of the needle-sheet-shaped particles is calculated. The working efficiency is improved, errors generated during manual measurement are avoided, and the measurement precision is improved.

Description

Technical Field

[0001] This invention relates to the field of materials testing technology, and specifically to a method and system for detecting the content of needle-like and flake-like particles based on computer vision. Background Technology

[0002] The size of coarse aggregate particles is a key factor in the skeleton structure of asphalt mixtures, affecting the interaction between asphalt and aggregate particles. This directly impacts the workability, shear strength, tensile strength, fatigue strength, and stiffness of the asphalt mixture, significantly influencing its road performance. Since flaky and needle-shaped aggregate particles are easily crushed, they reduce various properties of the mixture. Therefore, measuring the size of coarse aggregate particles is commonly used to determine whether they are flaky or needle-shaped.

[0003] Currently, the commonly used methods for determining the content of needle-shaped and flaky particles in coarse aggregates are mainly the standard instrument method and the vernier caliper method. Both are manual measurements, which are greatly affected by subjective human factors and are prone to errors. There is a lack of faster, more convenient, and traceable testing methods for aggregate size data.

[0004] Therefore, this invention proposes a method and system for detecting needle-like and flake-like content based on computer vision. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and system for detecting the content of needle-like particles based on computer vision. This solves the problems that the existing standard instrument method and vernier caliper method are both manual measurements, which are prone to errors and have low work efficiency.

[0006] The technical solution to achieve the above objectives is:

[0007] This invention provides a computer vision-based method for detecting the content of needle-like particles, comprising the following steps:

[0008] Provide the coarse aggregate to be tested, and spread the coarse aggregate evenly on a platform surface;

[0009] The coarse aggregate on the platform surface was photographed to obtain image information;

[0010] The image information is identified to obtain the size information of all coarse aggregates in the image information;

[0011] Based on the size information of all coarse aggregates in the obtained image information, a three-dimensional model is created; the three-dimensional model is then analyzed to identify needle-like and flaky particles, and the content of these particles is calculated.

[0012] Furthermore, the specific steps for photographing the coarse aggregate are as follows:

[0013] Provides an intelligent image sensor matrix;

[0014] The intelligent image sensor matrix 7 is placed above the platform surface, and then moved horizontally from one side of the platform surface to the other side. During the movement, the intelligent image sensor matrix is ​​used to capture images of the coarse aggregate to obtain image information.

[0015] Furthermore, before taking pictures, the intelligent image sensor matrix is ​​assembled, and the assembly steps are as follows:

[0016] Provides a cross-shaped bracket;

[0017] Five smart image sensors are provided. One smart image sensor is vertically mounted in the middle of the bottom surface of the cross-shaped bracket, and the remaining four smart image sensors are respectively mounted at the four ends of the bottom surface of the cross-shaped bracket, with the lower ends of the four smart image sensors tilted inward.

[0018] Furthermore, the steps for performing three-dimensional modeling of the coarse aggregate based on the obtained image information are as follows:

[0019] The image information is preprocessed, and the stripes and order of coarse aggregate in the image information are arranged to generate measurement area data with each matching point of coarse aggregate;

[0020] Aerial triangulation matching automatically matches and transfers points to the generated survey area data.

[0021] Perform fully automated regional network adjustment, construct a free network for the survey area, and output an adjustment accuracy report;

[0022] DSM matching, which performs epipolar constraint matching based on data obtained from aerial triangulation matching;

[0023] Based on the data obtained from DSM matching, multiple pre-intersections of the matching points are performed to generate a 3D point cloud, and the point cloud is then edited.

[0024] The generated point cloud is displayed and edited in three dimensions, and the edited point cloud is used to generate contour lines, as well as a three-dimensional TIN landscape map and orthophoto map.

[0025] The generated 3D TIN landscape map is fitted to the orthophoto map to obtain a 3D model.

[0026] Furthermore, before generating the point cloud, the data obtained from DSM matching is initially subjected to coarse point removal.

[0027] Furthermore, when analyzing the three-dimensional model, if the maximum length or width of the coarse aggregate is three times the minimum thickness, then the coarse aggregate is determined to be needle-shaped or flaky particles.

[0028] The present invention also provides a computer vision-based needle-like particle content detection system, comprising:

[0029] The platform surface is used to place the coarse aggregate to be tested;

[0030] A movable intelligent image sensor matrix is ​​installed above the platform surface to capture images of the coarse aggregate on the platform surface;

[0031] The data processing module is electrically connected to the intelligent image sensor matrix and is used to receive image information captured by the intelligent image sensor matrix and perform three-dimensional modeling on the image information.

[0032] Furthermore, it also includes a base and a support platform located on top of the base, the top surface of which forms a platform surface.

[0033] Furthermore, it also includes a lead screw slide on one side of the platform surface and an auxiliary slide on the other side of the platform surface. The top of the lead screw slide and the auxiliary slide are provided with support frames, and the intelligent image sensor matrix is ​​located between the two support frames.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] By spreading coarse aggregate flat on a platform and capturing images, a 3D model is created based on the dimensions of the coarse aggregate in the images. This 3D model is then used to analyze the coarse aggregate, identifying and calculating the content of needle-like and flaky particles. This method improves work efficiency, avoids errors caused by manual measurement, and increases measurement accuracy.

[0036] When photographing coarse aggregates, an intelligent image sensor matrix is ​​used, consisting of five intelligent image sensors spaced apart and moving from one side of the platform to the other, thus accurately obtaining information about each coarse aggregate. The angle between the five intelligent image sensors minimizes image distortion and facilitates matching of graphic information. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of the platform surface and intelligent image sensor matrix of the computer vision-based needle-like content detection system of the present invention.

[0038] Figure 2 This is a flowchart of the computer vision-based needle-like particle content detection method of the present invention.

[0039] Illustration: 1. Base; 2. Adjusting block; 3. Support platform; 4. Auxiliary slide; 5. Screw slide; 6. Support frame; 7. Intelligent image sensor matrix; 8. Intelligent image sensor; 9. Coarse aggregate. Detailed Implementation

[0040] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0041] See Figure 1 This invention provides a computer vision-based method and system for detecting the content of needle-like and flake-like particles, which solves the problems of existing standard instrument methods and vernier caliper methods, which are all manual measurements, prone to errors during the measurement process, and have low work efficiency.

[0042] By spreading coarse aggregate flat on a platform and capturing images, a 3D model is created based on the dimensions of the coarse aggregate in the images. This 3D model is then used to analyze the coarse aggregate, identifying and calculating the content of needle-like and flaky particles. This method improves work efficiency, avoids errors caused by manual measurement, and increases measurement accuracy.

[0043] When photographing coarse aggregates, an intelligent image sensor matrix is ​​used, consisting of five intelligent image sensors spaced apart and moving from one side of the platform to the other, thus accurately obtaining information about each coarse aggregate. The angle between the five intelligent image sensors minimizes image distortion and facilitates matching of graphic information.

[0044] The following describes a computer vision-based method and system for detecting needle-like and flake-like content according to the present invention, with reference to the accompanying drawings.

[0045] See Figure 1 This diagram shows the platform surface and intelligent image sensor matrix of the computer vision-based needle-like particle content detection system of the present invention. (See also...) Figure 2 The flowchart of the computer vision-based needle-like particle content detection method of the present invention is shown below. Figure 1 and Figure 2 This invention describes a computer vision-based method and system for detecting needle-like and flake-like content.

[0046] like Figure 1 As shown, this invention provides a computer vision-based method for detecting the content of needle-like particles.

[0047] Includes the following steps:

[0048] Provide the coarse aggregate 9 to be tested, and spread the coarse aggregate 9 evenly on a platform surface;

[0049] The coarse aggregate 9 on the platform surface was photographed to obtain image information;

[0050] The image information is identified to obtain the size information of all coarse aggregates 9 in the image information;

[0051] Based on the size information of all coarse aggregates 9 in the obtained image information, a three-dimensional model is created to obtain a three-dimensional model; the three-dimensional model is analyzed to mark needle-like and flaky particles and calculate the content of needle-like and flaky particles.

[0052] In one specific embodiment, when the coarse aggregate 9 is laid flat on the platform surface, the side with the largest area of ​​the coarse aggregate 9 is facing down, and the distance between two adjacent coarse aggregates 9 is not less than 1 cm, and the distance between each coarse aggregate 9 and the edge of the platform surface is not less than 1 cm.

[0053] In one specific embodiment, the step of photographing the coarse aggregate 9 is as follows:

[0054] Provides a smart image sensor matrix 7;

[0055] The intelligent image sensor matrix 7 is placed above the platform surface and moved horizontally from one side of the platform surface to the other side. During the movement, the intelligent image sensor matrix 7 is used to capture images of the coarse aggregate 9 to obtain image information.

[0056] In one specific embodiment, the intelligent image sensor matrix 7 is assembled before shooting, and the assembly steps are as follows:

[0057] Provides a cross-shaped bracket;

[0058] Five intelligent image sensors 8 are provided. One of the intelligent image sensors 8 is vertically mounted in the middle of the bottom surface of the cross-shaped bracket, and the other four intelligent image sensors 8 are respectively mounted at the four ends of the bottom surface of the cross-shaped bracket, with the lower ends of the four intelligent image sensors 8 tilted inward.

[0059] Specifically, the angle between the central intelligent image sensor 8 and the other four intelligent image sensors 8 is set to 5–10 degrees. This photographic method originates from rotating multi-baseline photography, which has the following properties: between adjacent camera stations (intelligent image sensors 8), when the total baseline is short, automation is easier; when the total baseline is long, the intersection angle is large, resulting in higher accuracy. This method, by controlling the angle between corresponding pairs of intelligent image sensors 8 to 10° or less, minimizes image distortion during shooting, thus facilitating image information matching. Traditional rotating multi-baseline photography increases the shooting angle by rotating the camera, while the above method increases the shooting angle by mounting the intelligent image sensors 8 at an angle on a cross-shaped bracket. Furthermore, by moving the intelligent image sensor matrix 7 from one side of the platform to the other, replacing the traditional rotational movement with horizontal movement, more targets can be captured simultaneously, increasing work efficiency.

[0060] In one specific embodiment, the step of performing three-dimensional modeling of the coarse aggregate 9 based on the obtained image information is as follows:

[0061] The image information is preprocessed and arranged according to the stripes and order of the coarse aggregate 9 in the image information to generate measurement area data with each matching point of the coarse aggregate 9.

[0062] Aerial triangulation matching automatically matches and transfers points to the generated survey area data.

[0063] Perform fully automated regional network adjustment, construct a free network for the survey area, and output an adjustment accuracy report;

[0064] DSM matching, which performs epipolar constraint matching based on data obtained from aerial triangulation matching;

[0065] Based on the data obtained from DSM matching, multiple pre-intersections of the matching points are performed to generate a 3D point cloud, and the point cloud is then edited.

[0066] The generated point cloud is displayed and edited in three dimensions, and the edited point cloud is used to generate contour lines, as well as a three-dimensional TIN landscape map and orthophoto map.

[0067] The generated 3D TIN landscape map is fitted to the orthophoto map to obtain a 3D model.

[0068] In one specific implementation, before generating the point cloud, coarse errors are initially removed from the data obtained by DSM matching. The greater the overlap rate of the points, the easier it is to remove coarse errors. An overlap threshold can be set when generating the point cloud.

[0069] In one specific embodiment, when analyzing the three-dimensional model, if the maximum length or width of the coarse aggregate 9 is three times the minimum thickness, then the coarse aggregate 9 is determined to be needle-shaped or flaky particles.

[0070] Specifically, during aerial triangulation matching, the matching results are displayed. If automatic matching fails, matching can be attempted again after manually providing seed points. The purpose of stereo editing of the point cloud is to remove coarse errors, and two methods are provided for this: manual removal and automatic removal based on triangulation filtering. Before automatic removal, an irregular triangulation is generated. When generating 3D TIN landscape maps and orthophoto maps, the best projection image closest to the principal image point is selected for texture mapping based on the projected coordinates of the target object's points on each image. Furthermore, when generating 3D TIN landscape maps, the optimal projection plane is first selected based on the depth direction of the photograph.

[0071] Ideally, if the target is large or has a complex structure, and it takes several sets of image information to obtain all the geometric and texture information of the target, then the image information of the target is partitioned and processed. After processing, the control points are aligned to stitch the target into the same coordinate system, thereby obtaining a complete three-dimensional model.

[0072] Specifically, the orthophoto image is a top view of the coarse aggregate 9 on the platform surface. When needle-like or flaky particles are detected in the coarse aggregate 9, the needle-like or flaky particles in the orthophoto image are smeared with a bright color, and the orthophoto image after smearing is exported to facilitate subsequent calculation of the content of needle-like or flaky particles. When calculating the content of needle-like or flaky particles, based on the smearing marks on the exported orthophoto image, the corresponding needle-like or flaky particles on the platform surface are collected, and the needle-like or flaky particles and the remaining coarse aggregate 9 are weighed to calculate the content of needle-like or flaky particles.

[0073] A computer vision-based needle-like particle content detection system includes:

[0074] The platform surface is used to place the coarse aggregate 9 to be tested;

[0075] A smart image sensor matrix 7 is movably mounted above the platform surface for capturing images of the coarse aggregate 9 on the platform surface;

[0076] The data processing module is electrically connected to the intelligent image sensor matrix 7 and is used to receive image information captured by the intelligent image sensor matrix 7 and perform three-dimensional modeling on the image information.

[0077] In one specific embodiment, it also includes a base 1 and a support platform 3 disposed on the top of the base 1, wherein the top surface of the support platform 3 forms a platform surface.

[0078] In one specific embodiment, it also includes a lead screw slide 5 disposed on one side of the platform surface and an auxiliary slide 4 disposed on the other side of the platform surface. The top of the lead screw slide 5 and the auxiliary slide 4 are provided with a support frame 6, and the intelligent image sensor matrix 7 is disposed between the two support frames 6.

[0079] Specifically, the top surface of the support platform 3 has a matte, non-slip design, which prevents the coarse aggregate 9 from sliding on the support platform 3 and the support platform 3 from reflecting light, thereby reducing the adverse effects of the external environment on the accuracy of image acquisition.

[0080] In one specific embodiment, the system further includes at least four adjusting blocks 2 installed at the bottom of the base 1 for adjusting the levelness of the base 1. When the ground where the base 1 is placed is uneven, the height of the adjusting blocks 2 is adjusted to keep the support platform 3 level, so as not to affect the accuracy of image acquisition.

[0081] Specifically, the upper end of the adjusting block 2 is provided with a screw, and the bottom of the base 1 is provided with a screw hole that is threadedly connected to the screw. By rotating the adjusting block 2, the screw is rotated in the screw hole, thereby adjusting the height of the adjusting block 2 to adjust the level of the support platform 3.

[0082] In one specific embodiment, the intelligent image sensor matrix 7 includes five intelligent image sensors 8, with four corresponding intelligent image sensors 8 distributed around another intelligent image sensor 8, and the intelligent image sensor 8 located in the center is vertically distributed. The lower ends of the four surrounding intelligent image sensors 8 are all tilted towards the central intelligent image sensor 8, so that the intelligent image sensors 8 face the detection area of ​​the support platform 3.

[0083] Preferably, it also includes a cross-shaped bracket on the top of the support frame 6, wherein the four ends and the middle of the cross-shaped bracket each have an extension portion extending downward, and the lower end of the extension portion at each of the four ends has an inclined portion that tilts towards the center, and the intelligent image sensor 8 is respectively disposed at the lower end of the corresponding extension portion and the inclined portion.

[0084] The present invention has been described in detail above with reference to the accompanying drawings and embodiments. Those skilled in the art can make various modifications to the present invention based on the above description. Therefore, certain details in the embodiments should not be construed as limiting the present invention, and the scope of protection of the present invention shall be defined by the appended claims.

Claims

1. A method for detecting the content of needle-like particles based on computer vision, characterized in that: Includes the following steps: Provide the coarse aggregate (9) to be tested, and spread the coarse aggregate (9) on a platform surface; The coarse aggregate (9) on the platform surface was photographed to obtain image information; The image information is identified to obtain the size information of all coarse aggregates (9) in the image information; Based on the size information of all coarse aggregates (9) in the obtained image information, a three-dimensional model is obtained; The three-dimensional model was analyzed to identify needle-like particles and calculate their content.

2. The method for detecting needle-like particle content based on computer vision according to claim 1, characterized in that: The specific steps for photographing the coarse aggregate (9) are as follows: Provides an intelligent image sensor matrix (7); The intelligent image sensor matrix (7) is placed above the platform surface, and the intelligent image sensor matrix (7) is moved horizontally from one side of the platform surface to the other side. During the movement, the intelligent image sensor matrix (7) is used to take pictures of the coarse aggregate (9) to obtain image information.

3. The method for detecting needle-like particle content based on computer vision according to claim 2, characterized in that: Before shooting, the intelligent image sensor matrix (7) is assembled. The assembly steps are as follows: Provides a cross-shaped bracket; Five smart image sensors (8) are provided. One of the smart image sensors (8) is vertically installed in the middle of the bottom surface of the cross-shaped bracket, and the other four smart image sensors (8) are respectively installed at the four ends of the bottom surface of the cross-shaped bracket, and the lower ends of the four smart image sensors (8) are tilted inward.

4. The method for detecting needle-like and flake-like content based on computer vision according to claim 1, characterized in that: The steps for performing three-dimensional modeling of the coarse aggregate (9) based on the obtained image information are as follows: The image information is preprocessed and arranged according to the stripes and order of the coarse aggregate (9) in the image information to generate the measurement area data with each matching point of the coarse aggregate (9); Aerial triangulation matching automatically matches and transfers points to the generated survey area data. Perform fully automated regional network adjustment, construct a free network for the survey area, and output an adjustment accuracy report; DSM matching, which performs epipolar constraint matching based on data obtained from aerial triangulation matching; Based on the data obtained from DSM matching, multiple pre-intersections of the matching points are performed to generate a 3D point cloud, and the point cloud is then edited. The generated point cloud is displayed and edited in three dimensions, and the edited point cloud is used to generate contour lines, as well as a three-dimensional TIN landscape map and orthophoto map. The generated 3D TIN landscape map is fitted to the orthophoto map to obtain a 3D model.

5. The method for detecting needle-like particle content based on computer vision according to claim 4, characterized in that: Before generating point clouds, coarse errors are initially removed from the data obtained by DSM matching.

6. The method for detecting needle-like particle content based on computer vision according to claim 4, characterized in that: When analyzing the three-dimensional model, if the maximum length or width of the coarse aggregate (9) is three times the minimum thickness, then the coarse aggregate (9) is determined to be needle-shaped particles.

7. A computer vision-based needle-like particle content detection system, characterized in that: include: The platform surface is used to place the coarse aggregate to be tested (9); A movable intelligent image sensor matrix (7) is installed above the platform surface to capture images of the coarse aggregate (9) on the platform surface; The data processing module is electrically connected to the intelligent image sensor matrix (7) and is used to receive image information captured by the intelligent image sensor matrix (7) and perform three-dimensional modeling on the image information.

8. The computer vision-based needle-like particle content detection system according to claim 7, characterized in that: It also includes a base (1) and a support platform (3) located on top of the base (1), the top surface of which forms a platform surface.

9. The computer vision-based needle-like particle content detection system according to claim 7, characterized in that: It also includes a lead screw slide (5) located on one side of the platform surface and an auxiliary slide (4) located on the other side of the platform surface. The top of the lead screw slide (5) and the auxiliary slide (4) are provided with support frames (6), and the intelligent image sensor matrix (7) is located between the two support frames (6).

Images