A method for detecting inclusion content, area, and particle size distribution in quartz sand

Abstract

This invention discloses a method for detecting the content, area, and particle size distribution of inclusions in quartz sand, comprising the following steps: S1, acquiring an optical microscope image of the quartz sand; S2, acquiring the total area of ​​the quartz sand in the optical microscope image; S3, acquiring the area and particle size distribution of inclusions in the optical microscope image, and acquiring the total area of ​​inclusions; S4, dividing the total area of ​​inclusions by the total area of ​​the quartz sand to determine the inclusion content in the quartz sand; the inclusion content in quartz sand is defined as: inclusion content = total area of ​​inclusions / total area of ​​quartz sand. This invention solves the problems of low accuracy and slow speed in existing quartz sand inclusion detection methods.

Description

Technical Field

[0001] This invention relates to the field of quartz sand inclusion detection technology, specifically to a method for detecting the content, area, and particle size distribution of quartz sand inclusions. Background Technology

[0002] High-purity quartz sand is a basic raw material widely used in integrated circuits, photovoltaics, and civil engineering. It is mainly used to prepare high-purity quartz crucibles for Czochralski single-crystal silicon production and as auxiliary materials for various processes in semiconductor and photovoltaic manufacturing, such as quartz glass tubes, rods, and plates.

[0003] Quartz sand is made from siliceous minerals such as quartz, pegmatite, and vein quartz through crushing and a series of processing steps. Therefore, inclusions intrinsically present in siliceous minerals are also widely distributed in quartz sand. These inclusions significantly affect the bubble content in quartz glass products, thereby affecting properties such as transmittance, thermal stability, and mechanical strength, ultimately impacting the performance and stability of the finished product. Therefore, it is crucial to analyze the inclusion content, area, and particle size distribution in quartz sand and select quartz sand with lower inclusion content as raw material.

[0004] For example, Chinese invention patent application CN112858097A uses an optical microscope to observe quartz sand and measures the inclusion content by the ratio of the number of quartz sand particles containing inclusions to the total number of quartz sand particles in the optical microscope image. This method has low accuracy and requires manual calculation of the total number of quartz sand particles, resulting in a slow detection speed.

[0005] For example, Chinese invention patent application CN113340780A uses an optical microscope to capture bright-field background images of the sample and multiple sets of images to be tested, obtaining the cumulative probability density of different transmittances, and then calculating the fluid inclusion index. This method is complex, the definition of the inclusion index is not intuitive, and it also suffers from low detection accuracy and slow speed.

[0006] Therefore, developing a method to improve the detection accuracy and shorten the detection time of inclusions in quartz sand is crucial for the quality assessment of quartz sand and the selection and screening of quartz ore sources. Summary of the Invention

[0007] To address the aforementioned technical problems and shortcomings in the field, this invention provides a method for detecting the content, area, and particle size distribution of inclusions in quartz sand, which solves the problems of low accuracy and slow speed in existing quartz sand inclusion detection methods.

[0008] This invention defines inclusion content in quartz sand for the first time as: Inclusion content = Total area of ​​inclusions / Total area of ​​quartz sand. This definition is more scientific and accurate than existing methods that ignore inclusion size or only use the number of inclusions.

[0009] The specific technical solution is as follows:

[0010] [1] A method for detecting the content, area, and particle size distribution of inclusions in quartz sand, comprising the following steps:

[0011] S1, Obtain an optical microscope image of the quartz sand;

[0012] S2, obtain the total area of ​​quartz sand in the optical microscope image;

[0013] S3, obtain the area and particle size distribution of inclusions in the optical microscope image, and obtain the total area of ​​inclusions;

[0014] S4. Divide the total area of ​​inclusions by the total area of ​​quartz sand to determine the inclusion content in the quartz sand; the inclusion content in the quartz sand is defined as: inclusion content = total area of ​​inclusions / total area of ​​quartz sand.

[0015] When obtaining optical microscope images of quartz sand, a clearing agent can be used to increase the refractive index between the objective lens and the slide, allowing light to penetrate the quartz sand sample better. Furthermore, the clearing agent may include at least one of clove oil, kerosene, wintergreen oil, etc., preferably clove oil.

[0016] For any optical microscope image, the total area of ​​quartz sand can be obtained using a method including the following steps:

[0017] (1) Using machine learning and deep learning methods, during the training process, one or more optical microscope photos are used as the base dataset. Quartz sand particles are manually selected as prior knowledge for dataset preprocessing. A suitable machine learning and deep learning model is selected, and the preprocessed dataset is input into the machine learning and deep learning model for training. A quartz sand classifier is obtained to distinguish between quartz sand and the background of optical microscope photos.

[0018] (2) Input the optical microscope image of the quartz sand to be detected, load the trained quartz sand classifier, and distinguish the quartz sand from the background of the optical microscope image.

[0019] (3) Use image recognition software to measure the area of ​​the distinguished quartz sand and obtain the total area of ​​quartz sand.

[0020] For any optical microscope image, the total area of ​​the inclusion can be obtained using a method including the following steps:

[0021] 1) Using machine learning and deep learning methods, during the training process, one or more optical microscope images of quartz sand containing inclusions are used as the basic dataset. Quartz sand particles containing inclusions are manually selected as prior knowledge for dataset preprocessing. A suitable machine learning and deep learning model is selected, and the preprocessed dataset is input into the machine learning and deep learning model for training to obtain a target detection model for detecting quartz sand containing inclusions, which can significantly reduce the subsequent computational burden.

[0022] 2) Using machine learning and deep learning methods, during the training process, one or more optical microscope images of quartz sand containing inclusions are used as the base dataset. Inclusions are manually selected as prior knowledge for dataset preprocessing. A suitable machine learning and deep learning model is selected, and the preprocessed dataset is input into the machine learning and deep learning model for training to obtain an inclusion classifier, which is used to distinguish inclusions from the quartz sand backing.

[0023] 3) Input the optical microscope image of the quartz sand containing inclusions to be detected, load the trained target detection model and inclusion classifier, and distinguish the inclusions from the quartz sand backing.

[0024] 4) Use image recognition software to measure the area and particle size distribution of the distinguished inclusions and obtain the total area of ​​the inclusions.

[0025] In step 1), the YOLOv5 object detection algorithm can be used to detect quartz sand containing inclusions, which can significantly reduce computational overhead and improve the model's efficiency in complex scenes. Through image annotation, model training, and hyperparameter optimization, the YOLOv5 object detection algorithm can accurately identify sand regions containing black dots and provide input for subsequent U-Net segmentation.

[0026] In step 2), the U-Net deep learning network segmentation model can be used to segment the inclusions and the edges of the quartz sand, accurately obtaining the contours of the quartz sand and the inclusions, thereby more accurately identifying the inclusions.

[0027] [2] The application of the method described in [1] in measuring the inclusion area, grain size distribution, and inclusion content of minerals containing inclusions. Further, the minerals containing inclusions may include at least one of quartzite, vein quartz, natural quartz sand, quartz sandstone, etc.

[0028] Since the particle size distribution, composition, and content of inclusions vary among different minerals, classifiers and target detection models can be trained separately for inclusion content analysis of a specific mineral.

[0029] [3] A computer device includes a memory and a processor, the memory being used to store a computer program, and the processor being used to execute the computer program stored in the memory, wherein the computer program, when executed, causes the processor to perform the method described in [1].

[0030] [4] A computer-readable storage medium storing a program or instructions that, when executed by a computer device, cause the computer device to perform the method described in [1].

[0031] Compared with the prior art, the beneficial effects of this invention are as follows:

[0032] This invention can quickly obtain the total area of ​​quartz sand and the total area of ​​inclusions in an optical microscope image, and then divide the obtained total area of ​​inclusions by the total area of ​​quartz sand to determine the inclusion content in the quartz sand. This solves the problems of low accuracy and slow speed in the current detection of inclusion content in quartz sand, and provides a more efficient method for rapid detection and screening of high-quality quartz sand. Attached Figure Description

[0033] Figure 1 This is a flowchart of a method for detecting the content, area, and particle size distribution of inclusions in quartz sand according to the present invention;

[0034] Figure 2 The images shown are optical microscope photographs (a), quartz sand differentiation diagrams (b), and inclusion differentiation diagrams (c) of quartz sand according to Embodiment 1 of the present invention.

[0035] Figure 3 The images shown are optical microscope photographs (a), quartz sand differentiation diagrams (b), and inclusion differentiation diagrams (c) of quartz sand in Embodiment 2 of this invention.

[0036] Figure 4 The present invention discloses optical microscope images (a), quartz sand differentiation diagram (b), and inclusion differentiation diagram (c) of quartz sand as comparative example 1.

[0037] Figure 5 Comparative Example 2 is disclosed in the following optical microscope images (a), quartz sand differentiation diagram (b), and inclusion differentiation diagram (c).

[0038] Figure 6 The following are optical microscope photographs (a) and inclusion differentiation diagrams (b) of quartz sand, which are disclosed in Comparative Example 3 of this invention. Detailed Implementation

[0039] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention.

[0040] See Figure 1 A method for detecting the content, area, and particle size distribution of inclusions in quartz sand, comprising the following steps:

[0041] S1. Obtain an optical microscope image of the quartz sand. For example, sprinkle quartz sand particles onto a glass slide, then use a dropper to add a small amount of clove oil, observe the quartz sand particles at an appropriate magnification, and take an optical microscope image.

[0042] S2, Obtain the total area of ​​quartz sand in the optical microscope image, specifically including the following steps:

[0043] (1) Using machine learning and deep learning methods, during the training process, one or more optical microscope photos are used as the base dataset. Quartz sand particles are manually selected as prior knowledge for dataset preprocessing. A suitable machine learning and deep learning model is selected, and the preprocessed dataset is input into the machine learning and deep learning model for training. A quartz sand classifier is obtained to distinguish between quartz sand and the background of optical microscope photos.

[0044] (2) Input the optical microscope image of the quartz sand to be detected, load the trained quartz sand classifier, and distinguish the quartz sand from the background of the optical microscope image.

[0045] (3) Use image recognition software to measure the area of ​​the distinguished quartz sand and obtain the total area of ​​quartz sand.

[0046] S3, Obtain the area and particle size distribution of inclusions in the optical microscope image, and obtain the total area of ​​inclusions, specifically including the following steps:

[0047] 1) Using machine learning and deep learning methods, during training, one or more optical microscope images of quartz sand containing inclusions are used as the base dataset. Quartz sand particles containing inclusions are manually selected as prior knowledge for dataset preprocessing. A suitable machine learning and deep learning model is then selected, and the preprocessed dataset is input into the model for training, resulting in an object detection model for detecting quartz sand containing inclusions. For example, using the YOLOv5 object detection algorithm for detecting quartz sand containing inclusions can significantly reduce computational overhead and improve model efficiency in complex scenes. Through image annotation, model training, and hyperparameter optimization, the YOLOv5 object detection algorithm can accurately identify gravel regions containing black dots and provide input for subsequent U-Net segmentation.

[0048] 2) Using machine learning and deep learning methods, during training, one or more optical microscope images containing quartz sand inclusions are used as the base dataset. Inclusions are manually selected as prior knowledge for dataset preprocessing. A suitable machine learning and deep learning model is chosen, and the preprocessed dataset is input into the model for training, resulting in an inclusion classifier used to distinguish inclusions from the quartz sand background. For example, the U-Net deep learning network segmentation model is used to segment the edges of inclusions and quartz sand, accurately obtaining the contours of the quartz sand and inclusions, thus more accurately identifying inclusions. The output of the U-Net model is a two-dimensional probability map of the same size as the input image, predicting the probability that each pixel belongs to the target category (e.g., inclusion or quartz sand background). For each pixel i, the predicted probability p... i It can be represented as: p i =σ(z) i ), where z i This is the original output of U-Net. σ is the Sigmoid function, which maps the output to the range [0,1], representing the probability that the pixel belongs to the positive class (wrap). Then, the probability map is converted into a binary segmentation result by a threshold (such as 0.5).

[0049] 3) Input the optical microscope image of the quartz sand containing inclusions to be detected, load the trained target detection model and inclusion classifier, and distinguish the inclusions from the quartz sand backing.

[0050] 4) Use image recognition software to measure the area and particle size distribution of the distinguished inclusions and obtain the total area of ​​the inclusions.

[0051] S4. Divide the total area of ​​inclusions by the total area of ​​quartz sand to determine the inclusion content in the quartz sand; the inclusion content in the quartz sand is defined as: inclusion content = total area of ​​inclusions / total area of ​​quartz sand.

[0052] Based on the methods described above for detecting the content, area, and particle size distribution of inclusions in quartz sand, the following describes an embodiment of quartz sand content detection software: The quartz sand content detection software consists of three main modules: a classifier training module, a calculation module, and an output module. The classifier training module uses optical microscope images as the base dataset, manually selects regions of interest (quartz sand or inclusions) as prior knowledge for dataset preprocessing, selects appropriate machine learning and deep learning models, and inputs the preprocessed dataset into the machine learning and deep learning models for training, resulting in a quartz sand (or inclusion) classifier. The calculation module uses the classifier to separate the region of interest (quartz sand or inclusions) from the background in the optical microscope image to be detected, and then calculates the area, particle size distribution, and total area of ​​the region of interest. The output module performs data statistics and analysis, and then outputs the results required by the user.

[0053] Furthermore, the method of this invention for detecting the content, area, and particle size distribution of inclusions in quartz sand can be used to measure the area, particle size distribution, and content of inclusions in various minerals containing inclusions, including but not limited to: quartzite, vein quartz, natural quartz sand, and quartz sandstone. Since the particle size distribution, composition, and content of inclusions differ in each mineral, classifiers can be trained separately for inclusion content analysis of a specific mineral.

[0054] The following are specific embodiments and some comparative examples of using the above-described method for detecting the content, area, and particle size distribution of inclusions in quartz sand.

[0055] Example 1

[0056] The quartz sand used in this embodiment is sourced from a vein of quartz ore in India. One of the optical microscope images is shown below. Figure 2 As shown in a.

[0057] After obtaining multiple optical microscope images, these images were used as the base dataset for training, resulting in a classifier specifically designed for vein quartz ore and quartz sand from a certain region in India. Then, the data was input... Figure 2 The optical microscope image shown in image a is used to obtain a quartz sand image after classification by loading a quartz sand classifier. Figure 2 b).

[0058] Measurement using image recognition software Figure 2 The total area of ​​the quartz sand in b is 409715μm. 2 .

[0059] After obtaining multiple optical microscope images, a U-Net deep learning network segmentation model was trained using these images as the base dataset to obtain a classifier specifically for inclusions in vein quartz deposits from a certain region in India. Then, the data was input... Figure 2 The optical microscope image shown in image a is used to load an inclusion classifier to obtain an image of the classified inclusions. Figure 2 c).

[0060] Measurement using image recognition software Figure 2 The total area of ​​the inclusions in c is 1037 μm. 2 The total number of inclusions was 177, and the inclusion content was 0.25%.

[0061] In terms of area distribution, the inclusions are mainly distributed in the range of 0–20 μm. 2 Within the range: area 0–10 μm 2 The number of inclusions was 156; the area was 10–20 μm. 2 The number of inclusions was 12. The largest inclusion area was 108.9 μm. 2The minimum inclusion area is 0.05 μm. 2 .

[0062] Regarding particle size distribution, the inclusions were mainly distributed in the range of 0–6 μm: 94 inclusions were 0–2 μm in size; 44 inclusions were 2–4 μm in size; and 24 inclusions were 4–6 μm in size. The largest inclusion was 22.8 μm in size; and the smallest inclusion was 0.3 μm in size.

[0063] Example 2

[0064] The quartz sand used in this embodiment was sourced from granite pegmatite ore from a local Indian mine. One of the optical microscope images is shown below. Figure 3 As shown in a.

[0065] After obtaining multiple optical microscope images, these images were used as the base dataset for training, resulting in a classifier specifically designed for quartz sand from granite pegmatite ore in a certain region of India. Then, the data was input... Figure 3 The optical microscope image shown in image a is used to obtain a quartz sand image after classification by loading a quartz sand classifier. Figure 3 b).

[0066] Measurement using image recognition software Figure 3 The total area of ​​the quartz sand in b is 428126 μm. 2 .

[0067] After obtaining multiple optical microscope images, a U-Net deep learning network segmentation model was trained using these images as the base dataset to obtain a classifier specifically for granite pegmatite inclusions from a local Indian region. Then, the data was input... Figure 3 The optical microscope image shown in image a is used to load an inclusion classifier to obtain an image of the classified inclusions. Figure 3 c).

[0068] Measurement using image recognition software Figure 3 The total area of ​​the inclusions in c is 7475 μm. 2 The total number of inclusions was 1487, and the inclusion content was 1.75%.

[0069] In terms of area distribution, the inclusions are mainly distributed in the range of 0–40 μm. 2 Within the range: area 0–10 μm 2 The number of inclusions was 1306; the area was 10–20 μm. 2 The number of inclusions was 103; the area was 20–30 μm. 2 The number of inclusions was 28; the area was 30–40 μm. 2 The number of inclusions was 15. The largest inclusion area was 159.0 μm. 2The minimum inclusion area is 0.05 μm. 2 .

[0070] Regarding particle size distribution, the inclusions were mainly distributed in the range of 0–10 μm: 820 inclusions were 0–2 μm in size; 352 inclusions were 2–4 μm in size; 141 inclusions were 4–6 μm in size; 75 inclusions were 6–8 μm in size; and 43 inclusions were 8–10 μm in size. The largest inclusion was 35.4 μm in size; the smallest inclusion was 0.3 μm in size.

[0071] Comparative Example 1

[0072] The quartz sand used in this comparative example was sourced from a vein of quartz ore in India. One of the optical microscope images is shown below. Figure 4 As shown in a (to eliminate interference), Figure 4 a and Figure 2 (a uses the same photo).

[0073] by Figure 4 A single photograph is used as the base dataset to train a classifier specifically for quartz sand from a vein of quartz ore in a certain region of India. Then, the input... Figure 4 The optical microscope image shown in image a is used to obtain a quartz sand image after classification by loading a quartz sand classifier. Figure 4 b).

[0074] Measurement using image recognition software Figure 4 The total area of ​​the quartz sand in b is 342481 μm. 2 .

[0075] by Figure 4 Based on a single photograph as the base dataset, a U-Net deep learning network segmentation model was trained to obtain a classifier specifically for inclusions in vein quartz ore from a certain region in India. Then, the input... Figure 4 The optical microscope image shown in image a is used to load an inclusion classifier to obtain an image of the classified inclusions. Figure 4 c).

[0076] Measurement using image recognition software Figure 4 The total area of ​​the inclusions in c is 558 μm. 2 The total number of inclusions was 134, and the inclusion content was 0.16%.

[0077] In terms of area distribution, the inclusions are mainly distributed in the range of 0–20 μm. 2 Within the range: area 0–10 μm 2 The number of inclusions was 120; the area was 10–20 μm. 2 The number of inclusions was 10. The largest inclusion area was 50.6 μm.2 The minimum inclusion area is 0.05 μm. 2 .

[0078] Regarding particle size distribution, the inclusions were mainly distributed in the range of 0–6 μm: 73 inclusions were 0–2 μm in size; 33 inclusions were 2–4 μm in size; and 19 inclusions were 4–6 μm in size. The largest inclusion was 14.0 μm in size; and the smallest inclusion was 0.3 μm in size.

[0079] Comparative Example 2

[0080] The quartz sand used in this comparative example was sourced from granite pegmatite ore from a local Indian mine. One of the optical microscope photographs is shown below. Figure 5 As shown in a (to eliminate interference), Figure 5 a and Figure 3 (a uses the same photo).

[0081] by Figure 5 A single photograph is used as the base dataset to train a classifier specifically designed for quartz sand from granite pegmatite ore in a certain region of India. Then, the input... Figure 5 The optical microscope image shown in image a is used to obtain a quartz sand image after classification by loading a quartz sand classifier. Figure 5 b).

[0082] Measurement using image recognition software Figure 5 The total area of ​​the quartz sand in b is 334293μm. 2 .

[0083] by Figure 5 Based on a single photograph as the base dataset, a U-Net deep learning network segmentation model was trained to obtain a classifier specifically for granite pegmatite inclusions from a local Indian region. Then, the input... Figure 5 The optical microscope image shown in image a is used to load an inclusion classifier to obtain an image of the classified inclusions. Figure 5 c).

[0084] Measurement using image recognition software Figure 5 The total area of ​​the inclusions in c is 4353 μm. 2 The total number of inclusions was 795, and the inclusion content was 1.30%.

[0085] In terms of area distribution, the inclusions are mainly distributed in the range of 0–40 μm. 2 Within the range: area 0–10 μm 2 The number of inclusions was 696; the area was 10–20 μm. 2 The number of inclusions was 48; the area was 20–30 μm. 2The number of inclusions was 16; the area was 30–40 μm. 2 The number of inclusions was 10. The largest inclusion area was 143.3 μm. 2 The minimum inclusion area is 0.05 μm. 2 .

[0086] Regarding particle size distribution, the inclusions were mainly distributed in the range of 0–10 μm: 431 inclusions were 0–2 μm in size; 201 inclusions were 2–4 μm in size; 72 inclusions were 4–6 μm in size; 37 inclusions were 6–8 μm in size; and 16 inclusions were 8–10 μm in size. The largest inclusion was 34.5 μm in size; the smallest inclusion was 0.3 μm in size.

[0087] Examples 1 and 2 used multiple optical microscope images as the basic training set, and the total area of ​​the quartz sand was statistically calculated to be 409715 μm. 2 and 428126μm 2 The total area of ​​the inclusions was 1037 μm. 2 and 7475μm 2 The total number of inclusions was 177 and 1487, respectively; the inclusion content was 0.25% and 1.75%, respectively. It can be seen that the inclusion content of the granite pegmatite ore from a certain Indian source used in Example 2 is much greater than that of the vein quartz ore from a certain Indian source. Both are basically consistent in terms of inclusion area and grain size distribution, with small-area, small-grained inclusions dominating.

[0088] Compared to Comparative Examples 1 and 2, which use a single optical microscope image as the basic training set, the total area of ​​the quartz sand was statistically calculated to be 342481 μm. 2 and 334293μm 2 The total area of ​​the inclusions was 558 μm. 2 and 4353μm 2 The total number of inclusions was 134 and 795, respectively; the inclusion content was 0.16% and 1.30%, respectively. Due to the lack of sufficient optical micrographs as a training set, all indicators decreased compared to the examples. Manual inspection and confirmation revealed that many inclusions in the comparative examples were not identified or counted, reflecting incomplete statistical analysis. Manual inspection and confirmation showed that inclusions in the examples were largely identified and counted, highlighting the importance of using sufficient data as a basic training set.

[0089] Comparative Example 3

[0090] The quartz sand used in this comparative example was sourced from a vein of quartz ore in India. One of the optical microscope images is shown below. Figure 6 As shown in a (to eliminate interference), Figure 6 a and Figure 2 (a uses the same photo).

[0091] After obtaining multiple optical microscope images, these images were used as the base dataset. A classifier specifically designed for inclusions in vein quartz deposits from a specific region in India was trained using the same model as the quartz sand classifier, but without employing the U-Net deep learning network segmentation model. Then, the data was input... Figure 6 The optical microscope image shown in image a is used to load an inclusion classifier to obtain an image of the classified inclusions. Figure 6 b).

[0092] from Figure 6 As shown in b, the inclusion images trained using the same model as the quartz sand classifier, but without the U-Net deep learning network segmentation model, contain many quartz sand edges, resulting in very low recognition accuracy. This comparison illustrates the importance of using deep learning models for fine segmentation.

[0093] Furthermore, it should be understood that after reading the above description of the present invention, those skilled in the art can make various alterations or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims.

Claims

1. A method for detecting the content, area, and particle size distribution of inclusions in quartz sand, characterized in that, Including the following steps: S1, Obtain an optical microscope image of the quartz sand; S2, Obtain the total area of ​​quartz sand in the optical microscope image; for any optical microscope image, the total area of ​​quartz sand is obtained using a method including the following steps: (1) Using machine learning and deep learning methods, during the training process, one or more optical microscope photos are used as the base dataset. Quartz sand particles are manually selected as prior knowledge for dataset preprocessing. A suitable machine learning and deep learning model is selected, and the preprocessed dataset is input into the machine learning and deep learning model for training to obtain a quartz sand classifier to distinguish between quartz sand and the background of optical microscope photos. (2) Input the optical microscope image of the quartz sand to be detected, load the trained quartz sand classifier, and distinguish the quartz sand from the background of the optical microscope image; (3) Use image recognition software to measure the area of ​​the distinguished quartz sand to obtain the total area of ​​the quartz sand; S3, obtain the area and particle size distribution of inclusions in the optical microscope image, and obtain the total area of ​​inclusions; for any optical microscope image, obtain the total area of ​​inclusions using a method including the following steps: 1) Using machine learning and deep learning methods, during the training process, one or more optical microscope images of quartz sand containing inclusions are used as the basic dataset. Quartz sand particles containing inclusions are manually selected as prior knowledge for dataset preprocessing. A suitable machine learning and deep learning model is selected, and the preprocessed dataset is input into the machine learning and deep learning model for training to obtain a target detection model for detecting quartz sand containing inclusions. 2) Using machine learning and deep learning methods, during the training process, one or more optical microscope images of quartz sand containing inclusions are used as the base dataset. Inclusions are manually selected as prior knowledge for dataset preprocessing. A suitable machine learning and deep learning model is selected, and the preprocessed dataset is input into the machine learning and deep learning model for training to obtain an inclusion classifier, which is used to distinguish inclusions from the quartz sand backing. 3) Input the optical microscope image of the quartz sand containing inclusions to be detected, load the trained target detection model and inclusion classifier, and distinguish the inclusions from the quartz sand backing. 4) Use image recognition software to measure the area and particle size distribution of the well-distinguished inclusions, and obtain the total area of ​​the inclusions; S4. Divide the total area of ​​inclusions by the total area of ​​quartz sand to determine the inclusion content in the quartz sand; the inclusion content in the quartz sand is defined as: inclusion content = total area of ​​inclusions / total area of ​​quartz sand.

2. The method according to claim 1, characterized in that, A clearing agent is used when obtaining optical microscope images of quartz sand to increase the refractive index between the objective lens and the slide, so that light can better penetrate the quartz sand sample. The transparent agent includes at least one of clove oil, kerosene, and wintergreen oil.

3. The method according to claim 1, characterized in that, In step 1), the YOLOv5 target detection algorithm is used to detect quartz sand containing inclusions.

4. The method according to claim 1, characterized in that, In step 2), the U-Net deep learning network segmentation model is used to segment the inclusions and the edges of the quartz sand, accurately obtaining the contours of the quartz sand and the inclusions, thereby more accurately identifying the inclusions.

5. The application of the method according to any one of claims 1 to 4 in the measurement of inclusion area, grain size distribution and inclusion content of minerals containing inclusions.

6. The application according to claim 5, characterized in that, The mineral containing inclusions includes at least one of quartzite, vein quartz, natural quartz sand, and quartz sandstone.

7. A computer device comprising a memory and a processor, the memory for storing a computer program, and the processor for executing the computer program stored in the memory, characterized in that, When the computer program is executed, it causes the processor to perform the method according to any one of claims 1 to 4.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program or instructions that, when executed by a computer device, cause the computer device to perform the method described in any one of claims 1 to 4.

Images