Material composition determination method, device, terminal and readable storage medium

Abstract

The present application is about a method, device, terminal and readable storage medium for determining material components, and relates to the field of rock identification technology. The method includes: obtaining a material slice image; inputting the material slice image into a crystalline recognition binary classification model, and outputting a single crystal component set image and a non-single crystal component set image; receiving scan data; verifying the single crystal component recognition result based on the scan data to obtain a verification result; and determining the component of the material slice based on the single crystal component recognition result, the non-single crystal component recognition result and the verification result. In this process, only the position corresponding to the single crystal component set image needs to be scanned, which reduces the area of ​​the scanning area, thereby reducing the working time consumed by the scanning, and improving the efficiency of determining the component of the material slice sample.

Description

Technical Field

[0001] The present application relates to the technical field of material identification, and in particular to a method, device, terminal and readable storage medium for determining material components. Background Art

[0002] In the field of material identification, when dealing with solid materials composed of a mixture of components of different types and quantities, that is, solid multi-component mixed materials, scanning analysis methods are usually used to determine the chemical composition or molecular structure of its constituent components, and then accurately identify the target material.

[0003] In the related art, scanning analysis usually adopts a combination of microscopic observation and micro-area analysis. The instruments used for microscopic observation include but are not limited to microscopes, scanning electron microscopes, etc., and the instruments used for micro-area analysis include but are not limited to molecular spectrometers, X-ray fluorescence spectrometers, X-ray energy spectrometers, etc. After the solid multi-component mixed material is cut to obtain a flat material sheet with a polished surface and fixed size, the surface of the material sheet is completely scanned by a scanning instrument in combination with microscopic observation. In one example, the circular material sheet is fully scanned using an X-ray fluorescence spectrum combined with microscopic observation, and the scanning results are output, and the properties of the material are determined by analyzing the scanning results. In one example, the circular material sheet is fully scanned using an X-ray fluorescence spectrum, and the scanning results are output, and the properties of the material are determined by analyzing the scanning results.

[0004] However, the scanning method in the related art is time-consuming, resulting in low efficiency in the process of determining the components of a solid multi-component mixed material through a comprehensive scanning analysis method. Summary of the invention

[0005] The present application is about a material component determination method, device, terminal and readable storage medium, which can improve the efficiency of determining the components of solid multi-component mixed materials. The technical solution is as follows:

[0006] In one aspect, a method for determining a material component is provided, the method comprising:

[0007] Acquire a material slice image, where the material slice image is an image obtained by cutting a material sample, the material slice image includes at least two component regions, and the material sample is a solid multi-component mixed material sample;

[0008] Inputting the material slice image into the crystalline state recognition binary classification model, outputting a single crystalline component set image and a non-single crystalline component set image, wherein the single crystalline component set image includes at least one single crystalline component region and a single crystalline component recognition result corresponding to the single crystalline component region, and the non-single crystalline component set image includes at least one non-single crystalline component region and a non-single crystalline component recognition result corresponding to the non-single crystalline component region, and the crystalline state recognition binary classification model is a neural network model based on deep learning;

[0009] receiving scanning data, where the scanning data is data obtained by scanning a region corresponding to a single crystal component;

[0010] The single crystal component identification result is verified based on the scanning data to obtain a verification result;

[0011] The components of the material slice image are determined based on the single crystal component identification result, the non-single crystal component identification result and the verification result.

[0012] In another aspect, a device for determining a material component is provided, the device comprising:

[0013] An acquisition module, used for acquiring a material slice image, wherein the material slice image is an image obtained by cutting a material sample, and the material slice image includes at least two component areas;

[0014] An input module is used to input a material slice image into a crystalline state recognition binary classification model, and output a single crystalline component set image and a non-single crystalline component set image, wherein the single crystalline component set image includes at least one single crystalline component region and a single crystalline component recognition result corresponding to the single crystalline component region, and the non-single crystalline component set image includes at least one non-single crystalline component region and a non-single crystalline component recognition result corresponding to the non-single crystalline component region, and the crystalline state recognition binary classification model is a neural network model based on deep learning;

[0015] A receiving module, used for receiving scanning data, where the scanning data is data obtained by scanning the corresponding single crystal region;

[0016] A verification module, used to verify the single crystal component identification result based on the scanning data to obtain a verification result;

[0017] The determination module is used to determine the components of the material slice based on the single crystal component identification result, the non-single crystal component identification result and the verification result.

[0018] On the other hand, a computer device is provided, which includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and the processor can load and execute at least one instruction, at least one program, code set or instruction set to implement the method for determining the material components provided in the above-mentioned embodiment of the present application.

[0019] On the other hand, a computer-readable storage medium is provided, in which at least one instruction, at least one program, code set or instruction set is stored, and a processor can load and execute at least one instruction, at least one program, code set or instruction set to implement the method for determining the material components provided in the above-mentioned embodiment of the present application.

[0020] On the other hand, a computer program product or a computer program is provided, the computer program product or the computer program comprising computer program instructions, the computer program instructions being stored in a computer-readable storage medium. A processor reads the computer instructions from the computer-readable storage medium and executes the computer instructions, so that the computer device performs the material component determination method provided in the embodiments of the present application.

[0021] The beneficial effects of the technical solution provided by this application include at least:

[0022] Before determining the composition of the material slice by scanning, the material slice image is divided into a single crystal component set image and a non-single crystal component set image by using a crystal recognition binary classification model, and the single crystal component set image is selected for further component verification by scanning. In this process, only the position corresponding to the single crystal component set image needs to be scanned, which reduces the area of ​​the scanning area, thereby reducing the working time consumed by scanning, and improving the efficiency of determining the composition of the material slice sample. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 A flow chart showing a method for determining a material component provided by an exemplary embodiment of the present application is shown;

[0025] Figure 2 A schematic structural diagram of a crystalline state recognition binary classification model provided by an exemplary embodiment of the present application is shown;

[0026] Figure 3A schematic diagram of the structure of another crystalline state recognition binary classification model provided by an exemplary embodiment of the present application is shown;

[0027] Figure 4 A flow chart showing another method for determining material components provided by an exemplary embodiment of the present application is shown;

[0028] Figure 5 A schematic diagram showing a process of a method for determining material components provided by an exemplary embodiment of the present application is shown;

[0029] Figure 6 A schematic diagram of a process of determining a material component provided by an exemplary embodiment of the present application is shown;

[0030] Figure 7 A schematic diagram of a process of determining a material component provided by an exemplary embodiment of the present application;

[0031] Figure 8 A schematic structural diagram of a computer device for executing a method for determining material components provided by an exemplary embodiment of the present application is shown. DETAILED DESCRIPTION

[0032] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0033] First, the terms involved in each embodiment of the present application are explained:

[0034] Component refers to each ingredient in a mixture. In the present application, component refers to the component of a solid material. Since the solid material is a mixture, the mixture includes at least two mutually separated components. The solid material is also called a solid multi-component mixed material. In one example, the mixture is implemented as a metal wire, and the components of the metal wire include at least two of metal materials such as lead, cadmium, bismuth, and tin; in another example, the mixture is implemented as a rock, and the components of the rock include but are not limited to at least two of the rock components such as terrigenous debris components, non-terrigenous debris components, cement components, and matrix components.

[0035] Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. Artificial Intelligence attempts to understand the essence of intelligence and produce an intelligent machine that can respond in a similar way to human intelligence. The purpose of artificial intelligence is to enable machines to have the functions of perception, reasoning and decision-making.

[0036] Artificial intelligence technology is a comprehensive subject covering a wide range of fields. Basic artificial intelligence technologies include, but are not limited to, sensor technology, artificial intelligence chip technology, cloud computing technology, big data processing technology, and mechatronics technology. The artificial intelligence technology used in the embodiments of this application is machine learning technology, and the machine learning device is applied to computer devices.

[0037] Surface scanning, i.e. surface scanning analysis, is the mainstream method for accurately identifying material components using test instruments. The usual practice is to cut the material components into thin slices or make samples with a smooth surface, and then use surface scanning to accurately identify the components therein. The test instruments used for surface scanning include but are not limited to molecular spectra such as infrared spectra and Raman spectra, X-ray fluorescence (X-ray Fluorescence, XRF) spectra, X-ray energy spectrum analysis (Energy Dispersive X-ray Spectroscopy, EDS), etc., and in some embodiments of the present application, the surface scanning process can be combined with a microscope or a scanning electron microscope to combine microscopic observation with micro-area analysis.

[0038] In the related art, in the process of surface scanning of a material sheet, the surface of the material sheet is usually completely and comprehensively scanned and stored by point-by-point scanning. In one example, the surface of a circular material sheet sample with a radius of 25 mm is scanned by XRF. The scanning process can last up to 10 hours. After the scanning results are obtained, the components of the material sheet can be determined based on the scanning results. This process is inefficient in determining the material components.

[0039] The present application provides a method for determining material components, which can reduce the scanning area by scanning only a portion of the material sheet, thereby reducing the working time consumed by scanning and improving the efficiency of determining material components.

[0040] Figure 1 A flow chart of a method for determining a material component provided by an exemplary embodiment of the present application is shown, and the method is described by taking the method applied to a computer device as an example. The method includes:

[0041] Step 101: Acquire a material sheet image, wherein the material sheet image includes at least two component regions.

[0042] In the embodiment of the present application, the computer device is a terminal device with storage, communication and data processing functions. In one example, the computer device is implemented as a personal computer, the personal computer has a storage function, the material sheet image is stored in the personal computer, and the computer device obtains the material sheet image according to the storage path; in another example, the computer device is implemented as a personal computer, the personal computer establishes a communication connection with the image acquisition device, and the personal computer obtains the material sheet image from the image acquisition device through the communication connection.

[0043] In the embodiment of the present application, the material slice image is a material corresponding to a solid multi-component mixed material. In the embodiment of the present application, the main component of the material is a mixture, so it is called a solid multi-component mixed material. In this case, the type of material includes but is not limited to at least one of rock, ceramic, and building material. After cutting the material sample and performing polishing operations, flattening operations, etc. on the cut part, a material slice sample can be obtained.

[0044] In the embodiment of the present application, by observing the material through a microscope, it can be determined that the material includes at least two component regions. That is, there are at least two component regions in the material sheet image, and optionally, the components indicated by the at least two component regions are different, or, in the at least two component regions, there is a component region indicating the same component. In the embodiment of the present application, the positions of the component regions have been segmented on the material sheet image. In one example, the component regions are segmented on the material sheet image by a method of dividing by marked lines.

[0045] Step 102, inputting the material slice image into a crystalline state recognition binary classification model, and outputting a single crystalline component set image and a non-single crystalline component set image.

[0046] In an embodiment of the present application, the single-crystalline component set image includes at least one single-crystalline component region, and a single-crystalline component identification result corresponding to the single-crystalline component region, and the non-single-crystalline component set image includes at least one non-single-crystalline component region, and a non-single-crystalline component identification result corresponding to the non-single-crystalline component region.

[0047] In the embodiments of the present application, the crystalline recognition binary classification model, as well as the various "models" or "sub-models" that appear, are all neural network models based on deep learning under the application of artificial intelligence technology. The above model is trained and debugged by sample set recognition. Optionally, the above model also has the function of active learning.

[0048] Step 103, receiving scanning data, where the scanning data is data obtained by scanning the corresponding single crystal region.

[0049] In the embodiment of the present application, the scanning mode includes at least one of point scanning, line scanning, and surface scanning. In the present application, the scanning mode is surface scanning as an example for description.

[0050] In an embodiment of the present application, the computer device is also connected to a surface scanning device. The surface scanning device is implemented as a test instrument for accurately identifying material components. In an embodiment of the present application, the surface scanning device includes but is not limited to at least one of molecular spectra such as infrared spectra and Raman spectra, X-ray fluorescence (X-ray Fluorescence, XRF) spectra, and X-ray energy spectrum analysis (Energy Dispersive X-ray Spectroscopy, EDS). Optionally, the computer device is connected to the surface scanning device for communication, and the computer device sends a position corresponding to the single crystal component area to the surface scanning device. The surface scanning area performs a surface scan on the material slice based on the above position, and feeds back the generated surface scanning data to the computer device, and the surface scanning data is the scanning data.

[0051] Step 104 , verifying the single crystal component identification result based on the scanning data to obtain a verification result.

[0052] The scanning data can be used to determine the component type in each component region in the single-crystalline component set image from the dimension of chemical composition or the dimension of crystal structure, wherein each single-crystalline component region in the single-crystalline component set image corresponds to scanning data indicating its chemical composition, so the scanning data can be used to determine the single-crystalline component identification result in each single-crystalline component region in the single-crystalline component set.

[0053] Step 105 , determining the components of the material slice image based on the single crystal component identification result, the non-single crystal component identification result and the verification result.

[0054] In an embodiment of the present application, the single-crystalline component identification result can directly characterize the component corresponding to the single-crystalline component region; the non-single-crystalline component identification result can directly characterize the component corresponding to the non-single-crystalline component region. The verification result is the same as the single-crystalline component identification result, or it can be a verification result different from the single-crystalline component identification result. In one example, the single-crystalline component identification result indicates that the single-crystalline component corresponding to a single-crystalline component region is component A, and the probability of being component A is 50%. The verification result shows that the region corresponding to the single-crystalline component region is component B, and the probability of component B is 70%. The verification result is selected, and it is determined that the component corresponding to the single-crystalline component region is component B.

[0055] After determining the recognition result of each component area, the components corresponding to each component area in the material sheet image can be determined, and then the components of the material sample can be determined. In an embodiment of the present application, the component determination of the material sheet image will be output in the form of a material sheet component image. In one example, the material sheet component image includes a material sheet image, on which the division of component areas and the component content corresponding to each component area are superimposed. In other embodiments of the present application, the determination results of the components of the material sample will be output in the form of a list. The present application does not limit the presentation method of the components of the material sheet image.

[0056] In summary, the method provided in the embodiment of the present application, before determining the composition of the material slice by scanning, divides the material slice image into a single crystalline component set image and a non-single crystalline component set image by a crystalline recognition binary classification model, and selects the single crystalline component set image for further component verification by scanning. In this process, only the position corresponding to the single crystalline component set image needs to be scanned, which reduces the area of ​​the scanning area, thereby reducing the working time consumed by scanning, and improving the efficiency of determining the composition of the material slice sample.

[0057] In the embodiments of the present application, the non-single crystalline component includes a polycrystalline component and an amorphous component. The properties of the polycrystalline component and the amorphous component are different, so the detection methods for the polycrystalline component and the amorphous component are also different. The present application performs step-by-step identification of the crystalline component and the amorphous component, as well as the single crystalline component and the polycrystalline component under the crystalline component classification, by setting a sub-model in the crystalline recognition binary classification model. Figure 2 A schematic diagram of the structure of a crystalline state identification binary classification model provided by an exemplary embodiment of the present application is shown. The crystalline state identification binary classification model 200 includes a first classification model level 210 and a second classification model level 220, wherein the first classification model level 210 includes a crystalline state classification sub-model 211, and the crystalline state classification sub-model 211 is used to determine whether the component region is a component region corresponding to the crystalline component. The second classification level model 220 includes a single crystalline state classification sub-model 221. The single crystalline state classification sub-model 221 is used to determine whether the component region is a single crystalline component region.

[0058] Based on Figure 2 Based on the crystalline state identification binary classification model shown, in some embodiments of the present application, the corresponding component region is a single crystalline component region, a polycrystalline component region, and an amorphous component region. Please refer to Figure 3The crystalline state identification binary classification model 300 includes a first classification model level 310 and a second classification model level 320. The first classification level model includes a crystalline state classification sub-model 311, and the second classification level model 320 includes a single crystalline state classification sub-model 321. At the same time, the crystalline state identification binary classification model 300 also includes a single crystalline state identification sub-model 331, a polycrystalline state identification sub-model 332, and an amorphous state identification sub-model 333. The amorphous state identification sub-model 333 is connected to the crystalline state classification sub-model 311, and the single crystalline state identification sub-model 331 and the polycrystalline state identification sub-model 332 are connected to the single crystalline state classification sub-model 321.

[0059] based on Figure 3 The binary classification model for crystal identification shown in the figure and the specific method for determining the scanning area during the verification of the corresponding single crystal component area are shown in the figure. Figure 4 A flow chart of a method for determining material components provided by an exemplary embodiment of the present application is shown. Taking the method applied to a computer device as an example, the method includes:

[0060] Step 401, obtaining a material sheet image.

[0061] As mentioned above, the material sheet image includes at least two component areas, and at least two component areas include at least two different components. The material sheet image can be an image pre-stored in a computer device, or an image received from other computer devices through a communication network, or a processed image obtained from an image acquisition device through a data line connection. The embodiment of the present application does not limit the method for obtaining the material sheet image and the specific form of the image acquisition device. In one example, the image acquisition device is a polarizing microscope.

[0062] Step 402, input the material slice image into the crystalline classification sub-model, and output a crystalline component set image and an amorphous component set image.

[0063] In the embodiment of the present application, the structure of the crystalline recognition binary classification model is as follows: Figure 3 As shown, when the material slice image is input into the crystalline classification sub-model, it will first pass through the crystalline classification sub-model. The crystalline classification sub-model is used to distinguish whether the component indicated by the component area is a crystalline component or an amorphous component. After passing through the crystalline classification sub-model, a crystalline component set image and an amorphous component set image can be obtained. The crystalline component set image is used to indicate that the component area within the component set image belongs to a crystalline material, and the amorphous component set image is used to indicate that the component area within the component set image belongs to an amorphous material.

[0064] In the embodiment of the present application, the crystalline component set image and the amorphous component set image can be images obtained by re-joining after deleting the content of the non-similar component regions in the material slice image, or can be images generated by directly deleting the non-similar component regions in the material slice image and retaining the structure of the original image. The embodiment of the present application does not limit the specific generation method and expression form of the crystalline component set image and the amorphous component set image.

[0065] Step 403, outputting the crystalline component set image to the single crystalline classification sub-model, and outputting a single crystalline component set image and a polycrystalline component set image.

[0066] In the embodiment of the present application, the single-crystalline component set image is used to indicate that the component region within the component set image belongs to a single-crystalline substance, and the polycrystalline component set image is used to indicate that the component region within the component set image belongs to a polycrystalline substance.

[0067] As described in step 402 , both the single-crystalline component set image and the multi-crystalline component set image are component set images obtained by deleting and reorganizing the crystalline component set image.

[0068] Step 404, input the single crystal component set image into the single crystal identification sub-model, and output the single crystal identification result.

[0069] In the embodiment of the present application, the single crystal state identification result is the result obtained after identification and inspection of the single crystal state component set image.

[0070] In the embodiment of the present application, the single crystal recognition sub-model is also a neural network model based on machine learning, and as mentioned above, the "polycrystalline component set image recognition model" and "amorphous component set image recognition model" appearing in the present application are both neural network models based on machine learning. The above recognition model constructs parameters through the training process of the sample set, and after the parameter construction is completed and the accuracy of the model is guaranteed, it can be put into use.

[0071] Step 405 , input the polycrystalline component set image into the polycrystalline component set image recognition model, and output a polycrystalline recognition result.

[0072] Step 406, input the amorphous component set image into the amorphous component set image recognition model, and output an amorphous recognition result.

[0073] Step 407 , determining a single crystal component region based on the single crystal identification result, and determining a single crystal component set image.

[0074] Step 408 : determining a polycrystalline component region based on the polycrystalline recognition result, and determining a polycrystalline component set image.

[0075] Step 409 , determining an amorphous component region based on the amorphous recognition result, and determining an amorphous component set image.

[0076] Step 407, step 408 and step 409 correspond to step 404, step 405 and step 406 respectively. The polycrystalline state identification result is the result obtained by identifying and testing the polycrystalline component set image; the single crystalline state identification result is the result obtained by identifying and testing the single crystalline component set image; and the amorphous state identification result is the result obtained by identifying and testing the amorphous component set image.

[0077] Optionally, according to the connection relationship between the sub-models in the crystalline recognition binary classification model, the actual implementation order of steps 404 to 409 can be adjusted arbitrarily. In one example, steps 407 to 409 are synchronously executed after the synchronous execution of steps 404 to 406 is completed. This application does not limit the actual implementation order of the above steps.

[0078] Step 410 , determining a non-single-crystalline component region based on the polycrystalline recognition result and the amorphous recognition result, and determining a non-single-crystalline component set image.

[0079] The polycrystalline components and the amorphous components can be collectively referred to as "non-single crystalline" components. Therefore, in the embodiment of the present application, the non-single crystalline component collective image is determined based on the polycrystalline identification results and the amorphous identification results.

[0080] Step 411, performing clean area screening on the single crystal component set image to obtain a clean area screening result.

[0081] In the embodiment of the present application, in the clean area, the number of impurities per unit area is less than the impurity number threshold value pre-stored in the computer or temporarily set during the execution process. The clean area screening result will correspond to at least one clean area in the image of the single crystalline component set. Scanning the clean area can eliminate the influence of impurities or structural changes of single crystalline materials on the scanning results.

[0082] Step 412: Send the clean area screening result to the scanning device.

[0083] As mentioned above, the scanning method selected in the embodiment of the present application is surface scanning.

[0084] In the embodiment of the present application, the scanning device is a detection instrument connected to a computer device. In one example, the scanning device and the image acquisition device are integrated into the same device, and the present application does not limit the actual implementation of the scanning device. Optionally, the clastic rock thin section sample image is located on the scanning device, and after receiving the regional screening result sent by the computer device, the scanning device performs a surface scan based on the regional screening result, and sends the surface scanning data obtained by the surface scanning to the computer device.

[0085] Step 413, receiving scan data.

[0086] Step 413 is the process in which the computer device receives the scan data sent by the scanning device.

[0087] Step 414, determining a recognition result based on the scan data.

[0088] In the embodiment of the present application, the scanning data is used to indicate the identification result of the components in the single crystal component region.

[0089] Step 415 , in response to the scanning identification result being the same as the single-crystalline component identification result, determining that the verification result indicates that the verification is passed, and determining that the single-crystalline component identification result is unchanged.

[0090] In some embodiments of the present application, the scanning identification result is the same as the single-crystalline component identification result. In this case, it can be determined that the single-crystalline component identification result remains unchanged.

[0091] Step 416 , in response to the scanning identification result being different from the single crystal component identification result, determining that the verification result indicates that the verification has failed.

[0092] Step 417, determining the single crystal component identification result based on the type of scanning equipment and the scanning identification result.

[0093] In the embodiment of the present application, there is a situation where the scanning identification result is different from the single-crystalline component identification result, resulting in failure of verification. In this case, the single-crystalline component identification result will be determined in combination with the type of scanning equipment.

[0094] In one example, the scanning device includes a first scanning device and a second scanning device. The computer device stores a recognition result determination rule, when the scanning device is the first scanning device, and the surface scanning recognition result is different from the single-crystalline component recognition result, the surface scanning recognition result is used as the single-crystalline component recognition result; when the surface scanning device is the second scanning device, and the surface scanning recognition result is different from the single-crystalline component recognition result, the single-crystalline component recognition result is used as the single-crystalline component recognition result, that is, the single-crystalline component recognition result is determined to be unchanged.

[0095] Step 418, determining the components of the material slice image based on the single crystal component identification result, the non-single crystal component identification result and the verification result.

[0096] In the embodiment of the present application, after the above steps, the component type of each component area in the material sheet image will be determined. Optionally, at this time, if some areas are screened out due to screening reasons in the above steps, the screened out parts are re-identified to improve the final determination accuracy of the material components.

[0097] In an embodiment of the present application, the final determination result of the material components can be presented in at least one visual manner. In one example, the final determination result of the material components is implemented as a visual chart; in another example, the final determination result of the material components is implemented as a visual text description paragraph; in another example, the final determination result of the material components is implemented as a picture labeled with the names of each component. The present application does not limit the visualization method of the final determination result of the material components.

[0098] In summary, the method provided in the embodiment of the present application, before determining the composition of the material slice by scanning, divides the material slice image into a single crystalline component set image and a non-single crystalline component set image by a crystalline recognition binary classification model, and selects the single crystalline component set image for further component verification by scanning. In this process, only the position corresponding to the single crystalline component set image needs to be scanned, which reduces the area of ​​the scanning area, thereby reducing the working time consumed by scanning, and improving the efficiency of determining the composition of the material slice sample.

[0099] By connecting the classification module and the recognition module in the crystalline recognition binary classification model, the component type of the component region can be directly determined by the crystalline recognition binary classification model, further improving the efficiency of determining the components of the material thin film sample.

[0100] By screening clean areas in the single crystal component area, the influence of impurities or crystal morphology in the scanning area on the verification results is eliminated during the scanning process, and the efficiency of determining the components of the material thin film sample is improved.

[0101] In the present application, "material" may include but is not limited to multi-component mixed solid materials used in industry and agriculture, and may also be multi-component mixed solid materials existing in nature and can be directly applied in industry and agriculture. Figure 5 A schematic diagram of a method for determining material components provided by an exemplary embodiment of the present application is shown, and the method is applied to a computer device as an example for explanation. In the process, the material is a clastic rock material as an example for explanation, and the process includes:

[0102] Step 501, segmenting the microscopic image of the solid multi-component mixed material.

[0103] This process is the process of making a preliminary component area division on the microscopic images obtained by taking pictures of clastic rock thin sections.

[0104] Step 502, inputting the microscopic image of the solid multi-component mixed material into a crystalline binary classification module to determine whether the component region therein is crystalline.

[0105] This process is to identify the crystalline component area of ​​the microscopic image corresponding to the clastic rock thin section through the crystalline binary classification module in the crystalline identification binary classification model.

[0106] In the embodiment of the present application, each module in the crystalline recognition binary classification model, that is, the sub-model, adopts deep learning technology.

[0107] Optionally, in the crystalline binary classification module, X-ray diffraction technology can be used to assist in determining the structure and degree of crystallinity of the molecule.

[0108] When the component region is in a crystalline state, step 504 is executed; when the component region is in an amorphous state, step 503 is executed.

[0109] Step 503, further obtaining AI recognition results through the amorphous recognition module.

[0110] This process is to identify the amorphous component area through the amorphous identification sub-model and obtain the AI ​​identification result.

[0111] Step 504, determining whether the component region is in a single crystal state in the single crystal binary classification module.

[0112] This process is to distinguish whether the component region is single crystal or polycrystalline through the crystal classification sub-model.

[0113] When the component region is in a single crystal state, step 506 is executed; when the component region is in a polycrystalline state, step 505 is executed.

[0114] Step 505, further obtaining AI recognition results through the polycrystalline recognition module.

[0115] In the embodiment of the present application, when the component region is polycrystalline, a polycrystalline identification sub-model is provided.

[0116] Step 506, further obtaining AI recognition results through the single crystal state recognition module.

[0117] In the embodiment of the present application, when the component region is in a single crystal state, a single crystal state identification sub-model is provided.

[0118] It should be noted that for materials with clear component types such as clastic rocks, multiple classification sub-models and identification sub-models can be combined to independently identify each type of component. In this case, the structure of the crystalline identification binary classification model is modified accordingly.

[0119] Step 507: Perform clean area screening using a clean area screening module.

[0120] It should be noted that the clean area screening module corresponds to the clean area screening principle. The purpose of the clean area screening process is to avoid interference factors in the surface scanning process. The corresponding material thin section is clastic rock. The interference factors are mainly divided into the first type of interference factors and the second type of interference factors. The first type of interference factors indicate interference factors related to the properties of the mineral itself, including alteration, twinning, inclusions, cracks, etc. caused by the properties of the mineral itself. Among them, alteration refers to the process in which the composition of minerals changes and new minerals are generated due to the influence of the outside world. In one example, feldspar is easily altered into clay minerals; twinning refers to the regular intergrowth of two or more crystals of the same kind. The connection between the crystals may interfere with the test. In one example, polysynthetic twinning is common in plagioclase; inclusions refer to a closed system composed of one or more phases of matter in a mineral and having a phase boundary with the main crystal mineral. In one example, quartz has rutile inclusions; cracks refer to the fracture of minerals caused by external stress. In one example, quartz has cracks. The second type of interference factors indicate interference factors generated during the sample preparation process. Including but not limited to situations where the sample is contaminated with polishing powder, slicing glue, impurities, etc.

[0121] Step 508: Perform surface scanning based on the screening result of the clean area.

[0122] This process is to perform surface scanning on the clean area obtained by screening. Through surface scanning, a verification result corresponding to the AI ​​recognition result will be obtained.

[0123] Step 509 , in the data comparison module, the result obtained by scanning the clean area surface is compared with the AI ​​recognition result to obtain the final result of each component area.

[0124] In the embodiment of the present application, the comparison module performs a comparison for the single crystalline component area. The comparison standard is affected by the type of scanning equipment. In one example, the scanning equipment is a molecular spectroscopy analysis equipment, which has the advantage of having the characteristics of fingerprint recognition of minerals, and the disadvantage is that it is susceptible to interference from fluorescence and molecular spectroscopy sensitive minerals. If the surface scanning recognition result is feldspar and the second component recognition result is quartz, the final result is feldspar; if the surface scanning recognition result is fluorescence interference (altered clay mineral interference), the second component recognition result is feldspar, the final result is feldspar; if the surface scanning recognition result is rutile (inclusion, molecular spectroscopy sensitive mineral), the second component recognition result is quartz, the final result is quartz; if the surface scanning recognition result is silicon carbide (polishing agent), the second component recognition result is quartz, the final result is quartz.

[0125] Step 510, in the data comprehensive processing module, the surface scanning result and the AI ​​recognition result are comprehensively processed to obtain the final result.

[0126] This process is the process of outputting the component composition information corresponding to the clastic rock.

[0127] In summary, the method provided in the embodiment of the present application, before determining the composition of the material slice by scanning, divides the material slice image into a single crystalline component set image and a non-single crystalline component set image by a crystalline recognition binary classification model, and selects the single crystalline component set image for further component verification by scanning. In this process, only the position corresponding to the single crystalline component set image needs to be scanned, which reduces the area of ​​the scanning area, thereby reducing the working time consumed by scanning, and improving the efficiency of determining the composition of the material slice sample.

[0128] Figure 6 A schematic diagram of a method for determining a material component provided by an exemplary embodiment of the present application is shown, wherein the device comprises:

[0129] An acquisition module 601 is used to acquire a material slice image, where the material slice image is an image obtained by photographing the material slice, where the material slice is an image obtained by cutting a material sample, where the material slice image includes at least two component regions, and the material sample is a solid multi-component mixed material sample;

[0130] An input module 602 is used to input a material slice image into a crystalline state recognition binary classification model, and output a single crystalline component set image and a non-single crystalline component set image, wherein the single crystalline component set image includes at least one single crystalline component region and a single crystalline component recognition result corresponding to the single crystalline component region, and the non-single crystalline component set image includes at least one non-single crystalline component region and a non-single crystalline component recognition result corresponding to the non-single crystalline component region, and the crystalline state recognition binary classification model is a neural network model based on deep learning;

[0131] A receiving module 603 is used to receive scanning data, where the scanning data is data obtained by scanning the corresponding single crystal region;

[0132] A verification module 604 is used to verify the single crystal component identification result based on the scanning data to obtain a verification result;

[0133] The determination module 605 is used to determine the composition of the material slice based on the single crystal component identification result, the non-single crystal component identification result and the verification result.

[0134] In an optional embodiment, the non-single-crystalline component set image includes an amorphous component set image and a polycrystalline component set image, the non-single-crystalline component region includes an amorphous component region and a polycrystalline component region, the amorphous component region corresponds to an amorphous component identification result, and the polycrystalline component region corresponds to a polycrystalline component identification result;

[0135] The binary classification model for crystalline state identification includes a crystalline state classification sub-model and a single crystalline state classification sub-model.

[0136] In an optional embodiment, the crystalline state recognition binary classification model further includes a single crystalline state recognition sub-model, a polycrystalline state recognition sub-model, and an amorphous state recognition sub-model;

[0137] The input module 602 is also used to input the material slice image into the crystalline classification sub-model, and output a crystalline component set image and an amorphous component set image;

[0138] Outputting the crystalline component set image to the single crystal classification sub-model, and outputting a single crystal component set image and a polycrystalline component set image;

[0139] The single crystal component set image is input into the single crystal identification sub-model, and the single crystal identification result is output;

[0140] Inputting the polycrystalline component set image into the polycrystalline component set image recognition model, and outputting a polycrystalline recognition result;

[0141] Input the amorphous component set image into the amorphous recognition sub-model, and output the amorphous recognition result;

[0142] The determination module 605 is further used to determine the single crystal component region based on the single crystal recognition result, and determine the single crystal component set image;

[0143] The non-single-crystalline component region is determined based on the polycrystalline identification result and the amorphous identification result, and a non-single-crystalline component collective image is determined.

[0144] Input module 602 is also used to receive surface scan data before, please refer to Figure 7 The device further includes a screening module 606 for screening the single-crystalline component set image for a clean region to obtain a clean region screening result, wherein the number of impurities per unit area in the clean region is less than an impurity number threshold, and the clean region screening result indicates at least one clean region in the single-crystalline component set image;

[0145] The sending module 607 is used to send the clean area screening result to the scanning device, and the scanning device is used to scan the clean area according to the clean area screening result, generate and feed back the scanning data to the computer device.

[0146] In an optional embodiment, the determination module 605 is further configured to determine a scan recognition result based on the scan data;

[0147] In response to the scanning identification result being the same as the single-crystalline component identification result, determining that the verification result indicates that the verification is passed, and determining that the single-crystalline component identification result is unchanged.

[0148] In an optional embodiment, after determining the scan recognition result based on the scan data, the determination module 605 is further configured to determine that the verification result indicates that the verification fails in response to the scan recognition result being different from the single crystal component recognition result;

[0149] Based on the type of scanning equipment and the scanning identification result, the single crystal component identification result is determined.

[0150] In an optional embodiment, the scanning device is a device that performs a surface scanning operation, and the scanning data is surface scanning data.

[0151] In summary, the device provided in the embodiment of the present application, before determining the composition of the material slice by scanning, divides the material slice image into a single crystalline component set image and a non-single crystalline component set image by a crystalline recognition binary classification model, and selects the single crystalline component set image for further component verification by scanning. In this process, only the position corresponding to the single crystalline component set image needs to be scanned, which reduces the area of ​​the scanning area, thereby reducing the working time consumed by scanning, and improving the efficiency of determining the composition of the material slice sample.

[0152] It should be noted that the device for determining the material composition provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the equipment can be divided into different functional modules to complete all or part of the functions described above.

[0153] Figure 8 A schematic diagram of the structure of a computer device for executing a material component determination method provided by an exemplary embodiment of the present application is shown, and the computer device includes:

[0154] The processor 801 includes one or more processing cores. The processor 801 executes various functional applications and data processing by running software programs and modules.

[0155] The receiver 802 and the transmitter 803 can be implemented as a communication component, and the communication component can be a communication chip. Optionally, the communication component can be implemented to include a signal transmission function. That is, the transmitter 803 can be used to transmit a control signal to the image acquisition device and the scanning device, and the receiver 802 can be used to receive the corresponding feedback instruction.

[0156] The memory 804 is connected to the processor 801 via a bus 805 .

[0157] The memory 804 may be used to store at least one instruction, and the processor 801 may be used to execute the at least one instruction to implement each step in the above method embodiment.

[0158] An embodiment of the present application also provides a computer-readable storage medium, which stores at least one instruction, at least one program, code set or instruction set, which is loaded and executed by a processor to implement the above-mentioned material component determination method.

[0159] The present application also provides a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the material component determination method described in any of the above embodiments.

[0160] Optionally, the computer readable storage medium may include: a read-only memory (ROM), a random access memory (RAM), a solid state drive (SSD), or an optical disk. Among them, the random access memory may include a resistance random access memory (ReRAM) and a dynamic random access memory (DRAM). The serial numbers of the above embodiments of the present application are only for description and do not represent the advantages and disadvantages of the embodiments.

[0161] A person skilled in the art will appreciate that all or part of the steps to implement the above embodiments may be accomplished by hardware, or by a program to instruct the relevant hardware, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.

[0162] The above are only optional embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for determining material composition, characterized in that: The method is applied in a computer device, and the method comprises: Acquire a material slice image, wherein the material slice image is an image obtained by photographing the material slice, wherein the material slice is an image obtained by cutting a material sample, wherein the material slice image includes at least two component regions, and the material sample is a solid multi-component mixed material sample; Inputting the material slice image into a crystalline state recognition binary classification model, outputting a single crystalline component set image and a non-single crystalline component set image, wherein the single crystalline component set image includes at least one single crystalline component region and a single crystalline component recognition result corresponding to the single crystalline component region, and the non-single crystalline component set image includes at least one non-single crystalline component region and a non-single crystalline component recognition result corresponding to the non-single crystalline component region, and the crystalline state recognition binary classification model is a neural network model based on deep learning; receiving scanning data, wherein the scanning data is data obtained by scanning a region corresponding to the single crystal component; Verifying the single crystal component identification result based on the scanning data to obtain a verification result; Determining the composition of the material sheet based on the single crystalline component identification result, the non-single crystalline component identification result, and the verification result; The non-single-crystalline component set image includes an amorphous component set image and a polycrystalline component set image, the non-single-crystalline component region includes an amorphous component region and a polycrystalline component region, the amorphous component region corresponds to an amorphous component identification result, and the polycrystalline component region corresponds to a polycrystalline component identification result; The crystalline state identification binary classification model includes a crystalline state classification sub-model and a single crystalline state classification sub-model; The crystalline state recognition binary classification model also includes a single crystalline state recognition sub-model, a polycrystalline state recognition sub-model and an amorphous state recognition sub-model; The step of inputting the material slice image into a crystalline state recognition binary classification model and outputting a single crystalline component set image and a non-single crystalline component set image comprises: Inputting the material slice image into the crystalline classification sub-model, and outputting a crystalline component set image and an amorphous component set image; Inputting the crystalline component set image into the single crystalline classification sub-model, and outputting a single crystalline component set image and a polycrystalline component set image; Inputting the single crystal component set image into the single crystal identification sub-model, and outputting the single crystal identification result; Inputting the polycrystalline component set image into the polycrystalline component set image recognition model, and outputting the polycrystalline recognition result; Inputting the amorphous component set image into the amorphous identification sub-model, and outputting the amorphous identification result; Determining the single crystal component region based on the single crystal identification result, and determining the single crystal component set image; Determine the non-single-crystalline component region based on the polycrystalline recognition result and the amorphous recognition result, and determine the non-single-crystalline component set image; Wherein, verifying the single crystal component identification result based on the scanning data includes: Determine a scan recognition result based on the scan data; In response to the scanning identification result being the same as the single-crystalline component identification result, determining that the verification result indicates that the verification is passed, and determining that the single-crystalline component identification result is unchanged.

2. The method according to claim 1, characterized in that Before receiving the scan data, the method includes: Performing clean area screening on the single crystalline component set image to obtain a clean area screening result, wherein the number of impurities per unit area in the clean area is less than an impurity number threshold, and the clean area screening result indicates at least one clean area in the single crystalline component set image; The clean area screening result is sent to a scanning device, and the scanning device is used to scan the clean area according to the clean area screening result, and generate and feed back the scanning data to the computer device.

3. The method according to claim 1, characterized in that After determining the scan recognition result based on the scan data, the method further includes: In response to the scanning identification result being different from the single-crystalline component identification result, determining that the verification result indicates that the verification has failed; The single crystal component identification result is determined based on the type of scanning equipment and the scanning identification result.

4. The method according to claim 3, characterized in that The scanning device is a device that performs a surface scanning operation, and the scanning data is surface scanning data.

5. A device for determining material composition, using the method of claim 1, characterized in that: The device comprises: An acquisition module, used for acquiring a material slice image, wherein the material slice image is an image obtained by photographing the material slice, wherein the material slice is an image obtained by cutting a material sample, wherein the material slice image includes at least two component regions, and the material sample is a solid multi-component mixed material sample; An input module, used for inputting the material slice image into a crystalline state recognition binary classification model, and outputting a single crystalline component set image and a non-single crystalline component set image, wherein the single crystalline component set image includes at least one single crystalline component region and a single crystalline component recognition result corresponding to the single crystalline component region, and the non-single crystalline component set image includes at least one non-single crystalline component region and a non-single crystalline component recognition result corresponding to the non-single crystalline component region, and the crystalline state recognition binary classification model is a neural network model based on deep learning; A receiving module, used for receiving scanning data, wherein the scanning data is data obtained by scanning the corresponding single crystal component region; A verification module, used for verifying the single crystal component identification result based on the scanning data to obtain a verification result; A determination module is used to determine the composition of the material slice based on the single-crystalline component identification result, the non-single-crystalline component identification result and the verification result.

6. A computer device, characterized in that: The computer device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, at least one program, a code set or an instruction set is loaded and executed by the processor to implement the method for determining the material component as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that: The readable storage medium stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor to implement the method for determining the material component as described in any one of claims 1 to 4.

Images