Lattice structure identification method and apparatus for conductive carbon black, and computer system

Abstract

The present application relates to the technical field of image processing, and relates to a lattice structure identification method and apparatus for a conductive carbon black, and a computer system. The method comprises: acquiring a sample electron-microscope image of a conductive carbon black, labeling in the sample electron-microscope image a crystal structure of the conductive carbon black, and establishing a sample data set on the basis of the labeled sample electron-microscope; inputting the sample data set into an initial mask R-CNN model for model training, so as to acquire a target mask R-CNN model; and inputting into the target mask R-CNN model an image to be subjected to detection, so as to acquire an identification result for a crystal structure of a conductive carbon black in the image to be subjected to detection. By means of the method, the efficiency and accuracy of lattice structure identification can be improved. In addition, the present application also relates to the field of computers.

Description

Conductive carbon black lattice structure identification method, device and computer equipment

[0001] This application claims priority to a Chinese patent application filed with the Patent Office of China on October 31, 2023, with application number 2023114327268 and application name “Method, device and computer equipment for identifying the lattice structure of conductive carbon black”, the entire contents of which are incorporated herein by reference. Technical Field

[0002] The present application relates to the field of image processing technology, and in particular to a method, device and computer equipment for identifying the lattice structure of conductive carbon black. Background Art

[0003] Conductive carbon black is a typical specialty carbon black with specialized electrical conductivity. Due to its excellent conductivity and high cost-performance ratio, it is widely used in a variety of fields, including electronic devices, mining pipes, cable shielding materials, aerospace, petrochemicals, and power storage. Furthermore, compared to highly conductive fillers such as graphene and carbon nanotubes, conductive carbon black offers excellent cost-performance, making it increasingly popular in the energy sector, for example, as a conductive agent in (power) lithium-ion batteries, lead-acid batteries, and supercapacitors.

[0004] Existing methods for observing conductive carbon black include high-resolution transmission electron microscopy (HRTEM), Raman spectroscopy, X-ray diffraction (XRD), infrared spectroscopy, and dibutyl phthalate (DBP) oil absorption. These methods can be used to characterize basic properties of conductive carbon black, such as crystal structure, shape, and size. High-resolution transmission electron microscopy has been widely used to analyze the microscopic physical and chemical structure of aromatic hydrocarbon flakes in carbon-based materials such as conductive carbon black.

[0005] However, the inventors realized that the existing technology for extracting the lattice structure of conductive carbon black based on high-resolution transmission electron microscope images still has the problem of high error.

[0006] Summary of the Invention

[0007] According to various embodiments disclosed in the present application, a conductive carbon black lattice structure identification method, apparatus, and computer equipment are provided that can improve the accuracy of lattice fringes extraction and analysis of carbon-based materials.

[0008] A method for identifying the lattice structure of conductive carbon black comprises:

[0009] Obtaining a sample electron microscope image of conductive carbon black, annotating the crystal structure of the conductive carbon black in the sample electron microscope image, and establishing a sample data set based on the annotated sample electron microscope image;

[0010] Input the sample dataset into the initial Mask R-CNN model for model training to obtain the target Mask R-CNN model; and

[0011] The image to be detected is input into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

[0012] A conductive carbon black lattice structure identification device, comprising:

[0013] A sample acquisition module is used to obtain a sample electron microscope image of conductive carbon black, annotate the crystal structure of the conductive carbon black in the sample electron microscope image, and establish a sample data set based on the annotated sample electron microscope image;

[0014] The model training module is used to input the sample dataset into the initial Mask R-CNN model for model training to obtain the target Mask R-CNN model; and

[0015] The recognition module is used to input the image to be detected into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

[0016] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are performed:

[0017] Obtaining a sample electron microscope image of conductive carbon black, annotating the crystal structure of the conductive carbon black in the sample electron microscope image, and establishing a sample data set based on the annotated sample electron microscope image;

[0018] Input the sample dataset into the initial Mask R-CNN model for model training to obtain the target Mask R-CNN model; and

[0019] The image to be detected is input into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

[0020] One or more computer-readable storage media storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the following steps:

[0021] Obtaining a sample electron microscope image of conductive carbon black, annotating the crystal structure of the conductive carbon black in the sample electron microscope image, and establishing a sample data set based on the annotated sample electron microscope image;

[0022] Input the sample dataset into the initial Mask R-CNN model for model training to obtain the target Mask R-CNN model; and

[0023] The image to be detected is input into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

[0024] A computer program product, comprising a computer program, which, when executed by a processor, implements the following steps:

[0025] Obtaining a sample electron microscope image of conductive carbon black, annotating the crystal structure of the conductive carbon black in the sample electron microscope image, and establishing a sample data set based on the annotated sample electron microscope image;

[0026] Input the sample dataset into the initial Mask R-CNN model for model training to obtain the target Mask R-CNN model; and

[0027] The image to be detected is input into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

[0028] The above-mentioned conductive carbon black lattice structure recognition method, device and computer equipment obtain a sample electron microscope image of conductive carbon black, and annotate the crystal structure of the conductive carbon black in the sample electron microscope image, and establish a sample data set based on the annotated sample electron microscope image; input the sample data set into the initial Mask R-CNN model for model training to obtain a target Mask R-CNN model; input the image to be detected into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected. This application uses the Mask R-CNN model to perform target segmentation and recognition on the electron microscope image of conductive carbon black, which improves the accuracy and efficiency compared to the traditional manual extraction method. Compared with the traditional automated processing method developed based on computer language, it effectively avoids the loss of a large amount of useful information and ensures accurate lattice feature recognition and extraction.

[0029] The details of one or more embodiments of the present application are set forth in the accompanying drawings and the description below. Other features and advantages of the present application will become apparent from the description, drawings, and claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. 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.

[0031] FIG1 is a diagram illustrating an application environment of a method for identifying a conductive carbon black lattice structure according to one or more embodiments;

[0032] FIG2 is a schematic flow chart of a method for identifying the lattice structure of conductive carbon black according to one or more embodiments;

[0033] FIG3 is a schematic diagram of cropping an original electron microscope image in one embodiment;

[0034] FIG4 is a schematic diagram of a process for obtaining an electron microscope image of a sample in one embodiment;

[0035] FIG5 is a schematic diagram of a data sample set in one embodiment;

[0036] FIG6 is a schematic diagram showing changes in accuracy, loss function, and learning rate indicators during model training in one embodiment;

[0037] FIG7 is a schematic diagram of a lattice structure labeled after identification using a target Mask R-CNN model in one embodiment;

[0038] FIG8 is a structural block diagram of a device for identifying the lattice structure of conductive carbon black according to one or more embodiments;

[0039] FIG9 is a diagram illustrating the internal structure of a computer device according to one or more embodiments. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0041] The conductive carbon black lattice structure identification method provided in the embodiment of the present application can be applied to the application environment shown in Figure 1. The terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The terminal 102 can be, but is not limited to, various personal computers, laptops, smart phones, tablets, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server 104 can be implemented as an independent server or a server cluster consisting of multiple servers.

[0042] In one exemplary embodiment, as shown in FIG2 , a method for identifying the lattice structure of conductive carbon black is provided, which is described by taking the method applied to the terminal 102 in FIG1 as an example, and includes the following steps 202 to 206 . In which:

[0043] Step 202 : obtaining a sample electron microscope image of the conductive carbon black, annotating the crystal structure of the conductive carbon black in the sample electron microscope image, and establishing a sample data set based on the annotated sample electron microscope image.

[0044] Among them, the electron microscope image of the sample is obtained by preparing a dispersion of conductive carbon black and ethanol in a certain ratio, and then performing high-resolution transmission electron microscope testing after sample preparation.

[0045] In one of the embodiments, the sample data set contains manually annotated annotation information, and the annotation process is performed using an open source Labelme plug-in, etc., to mark the areas that need to be identified and assign the annotations to create a sample data set. The Labelme plug-in provides a convenient and efficient way to mark the areas that need to be identified, thereby creating a high-quality sample data set for subsequent training. By using the Labelme plug-in, target areas of interest in the image can be accurately annotated. These target areas may be the position and shape of conductive carbon black particles in the sample electron microscope image, or other areas that need to be identified. The annotation process involves drawing bounding boxes, marking key points, or drawing contours on the image to accurately define the boundaries of the target area.

[0046] In step 206, the sample data set is input into the initial Mask R-CNN model for model training to obtain a target Mask R-CNN model.

[0047] Among them, the Mask R-CNN (Mask Region-based Convolutional Neural Network) model is a deep learning model for target detection and instance segmentation tasks. It is an extension of the Faster R-CNN (Faster Region-based Convolutional Neural Network) model, which can simultaneously predict the bounding box of the target and the semantic segmentation mask of each pixel. The Mask R-CNN model introduces an additional branch in the target detection process to generate an accurate mask for each instance. It first generates candidate regions in the input image, then classifies these candidate regions and predicts the precise coordinates of the bounding box. Next, it uses a fully convolutional network to generate a mask for each candidate region to achieve accurate instance segmentation. By integrating target detection and instance segmentation, the Mask R-CNN model is able to achieve pixel-level object segmentation and can simultaneously detect and segment multiple objects in an image.

[0048] This embodiment uses a sample dataset to train the initial Mask R-CNN model. The target Mask R-CNN model obtained through training can automatically learn and extract image features in the sample dataset, and locate and segment the target area in the image through the neural network structure, thereby achieving pixel-level prediction of the target area.

[0049] In step 208 , the image to be detected is input into the target Mask R-CNN model to obtain a recognition result of the conductive carbon black crystal structure in the image to be detected.

[0050] The model to be tested is input into the trained target Mask R-CNN model, and after model prediction, the recognition result of the crystal structure of each conductive carbon black particle is obtained.

[0051] Traditional methods for extracting the crystal structure of conductive carbon black based on high-resolution transmission electron microscope images are mainly divided into two methods. The first method is manual extraction, which is to manually draw and extract the parameters of the crystal structure using mapping software such as ArcGIS or Image J. The second method is an automated processing method developed based on computer language, such as VirtualFringe. However, the manual extraction method is time-consuming and has a high degree of error; the automated processing method is prone to losing a lot of information in the filtering and noise reduction process due to the different brightness and darkness of the image, and some noise may remain. As a result, although this method is fast, the processing results for different images may not be accurate enough and still require manual intervention.

[0052] To address these issues, this example applies the Mask R-CNN model to extract the crystal structure of conductive carbon black. This model possesses powerful segmentation and target recognition capabilities, accurately segmenting conductive carbon black particles from enhanced images and precisely identifying their position and shape. The application of deep learning technology enables the Mask R-CNN model to perform highly accurate segmentation and target recognition in complex images, enabling refined analysis of conductive carbon black particles.

[0053] This example uses the Mask R-CNN model to segment and identify targets in electron microscope images of conductive carbon black samples. Compared to traditional manual extraction methods, this method achieves improved accuracy and efficiency. Compared to traditional automated processing methods developed using computer languages, it effectively avoids the loss of significant useful information and ensures accurate lattice feature recognition and extraction. Therefore, this example achieves improved accuracy and reliability in conductive carbon black lattice extraction and recognition.

[0054] In one embodiment, obtaining a sample electron microscope image of the conductive carbon black includes: collecting an original electron microscope image of the conductive carbon black, performing image enhancement processing on the original electron microscope image, and obtaining a sample electron microscope image.

[0055] The original electron microscope image is an image directly obtained from a high-resolution transmission electron microscope test. After image enhancement processing, the characteristics of the lattice structure can be enhanced to provide a more effective high-quality sample data set for subsequent model training.

[0056] In one embodiment, image enhancement processing is performed on the original electron microscope image to obtain a sample electron microscope image, including: cropping the original electron microscope image, adjusting a first image parameter based on the cropped original electron microscope image to obtain a first image; grayscale processing is performed on the first image to obtain a second image; Fourier-inverse Fourier transform processing is performed on the second image to obtain a third image; and second image parameter adjustment is performed on the third image to obtain a sample electron microscope image; the first image parameter adjustment and the second image parameter adjustment are respectively used to adjust basic parameters of the original electron microscope image and basic parameters of the third image, and the basic parameters include at least one of contrast, saturation and threshold.

[0057] Specifically, the first step is to crop the original electron microscope image into several small images. The original electron microscope image is generally a large image of 4096*4096 pixels, and after cropping, a small image of 256*256 pixels is obtained. As shown in Figure 3, the left side is the original electron microscope image before cropping, and the right side is the original electron microscope image after cropping. Based on the cropped original electron microscope image, the first image parameter adjustment is performed, that is, the basic parameters of each small image are adjusted, such as adjusting the contrast, and slightly adjusting the saturation and threshold to enhance the contrast between the crystal structure and the background. Image contrast refers to the difference in brightness levels between the brightest white and the darkest black in the light and dark areas of the image, that is, the size of the grayscale contrast of an image. Saturation refers to the vividness of the colors in the image. The threshold refers to the segmentation threshold that divides the pixels in the image into two categories (usually black and white), and the pixels in the image with grayscale values ​​higher or lower than the threshold are classified into two different categories respectively. Generally speaking, in the first step, the original electron microscope image is converted into a unified format. The original electron microscope image is generally a tif format file. After cropping and adjusting the first image parameters, it is converted into a jpg format file to obtain the first image.

[0058] The second step is to convert the first image's data format to grayscale, meaning the pixel color parameters are converted to a range of 0 to 255, to obtain the second image. This grayscale conversion transforms the multi-channel color first image into a single-channel grayscale second image. This improves the contrast between the lattice area and the background, facilitating subsequent processing.

[0059] In the third step, the second image is processed using a Fourier-inverse Fourier transform (IFT) to obtain a third image. The IFT refers to two steps in image processing, and the Fourier transform is the process of converting an image from the time domain to the frequency domain. In image processing, applying a Fourier transform to an image allows it to be represented as a set of spectral components, each representing information at a different frequency. This allows for a better understanding of image information, such as edges, texture, and periodicity. Frequency-domain image processing, such as filtering, is then performed to remove or enhance components at specific frequencies to achieve image smoothing, sharpening, and noise reduction. Once processing in the frequency domain is complete, an IFT can be used to convert the image back to its original time domain representation. The IFT allows processing in the frequency domain to be performed and results to be obtained in the time domain, facilitating various operations and analyses on the image. In general, the role of the Fourier-inverse Fourier transform in image processing is to provide a conversion mechanism between the time domain and the frequency domain. This mechanism can better understand and process the frequency information in an image, and has a wide range of applications in image filtering, enhancement, compression, image analysis, and feature extraction. Therefore, the Fourier-inverse Fourier transform is used to further enhance the contrast between the crystal structure and the background on the basis of the second image, preserving the original structural information as much as possible and highlighting the crystal structure characteristics.

[0060] In the third step, the Fourier-inverse Fourier transform can be performed using the open source computer vision library OpenCV in Python.

[0061] The fourth step involves adjusting the parameters of the second image on the third image. For example, the third image is imported into Adobe Photoshop and the threshold is adjusted to obtain an electron microscope image of the sample. The purpose of this fourth step is to remove noise from the image and further enhance the crystal structure to facilitate subsequent manual image annotation.

[0062] The image enhancement process described above can be implemented partially or entirely through the compilation of Python scripts, or by combining Python scripts with manual processing. For example, a Python script can be used for image cropping and basic image parameters can be manually adjusted. Manual adjustment offers the advantage of greater flexibility and can maximize the contrast between the lattice structure and the background.

[0063] Figure 4 is a schematic diagram of image changes during image enhancement processing. Figure 4 a is the cropped original electron microscope image, Figure 4 b is the first image obtained by adjusting the first image parameter on the original electron microscope image, Figures 4 c and 4 d are the second images after grayscale processing, Figure 4 e is the third image after Fourier-inverse Fourier processing, and Figure 4 f is the sample electron microscope image obtained by adjusting the second image parameter on the third image.

[0064] In one embodiment, the sample dataset includes a training set and a validation set. Inputting the sample dataset into an initial Mask R-CNN model for model training to obtain a target Mask R-CNN model includes: inputting the training set and the validation set into the initial Mask R-CNN model, iteratively performing training and validation operations until the initial Mask R-CNN model converges, thereby obtaining a target Mask R-CNN model.

[0065] Among them, the sample electron microscope images in the sample data set are sample electron microscope images with annotation information. This embodiment performs role segmentation on the annotated sample electron microscope images and exports them to a specific format. Part of the sample data set is divided into a training set, and the other part is divided into a validation set. The ratio of the training set to the validation set can be 8:2. The format of the sample data set can adopt the format of the coco data set, and the subfolders respectively include sample electron microscope images as a training set, sample electron microscope images as a validation set, and annotation information. As shown in Figure 5, it is a schematic diagram of part of the sample data set. Figure 5 contains manually annotated annotation information, and the circled area is the area where the conductive carbon black lattice structure is located, and the remaining area is the background area.

[0066] The sample dataset is input into the initial Mask R-CNN model for training. The training process of the initial Mask R-CNN model is as follows:

[0067] 1. Model initialization: Generally, you can use a pre-trained model as the base network and initialize the weights of the pre-trained model. This base network is usually a model for image classification, such as ResNet (Residual Neural Network) and VGG (Visual Geometry Group).

[0068] 2. Feature extraction: Using the training set as input, the image is fed into the model through a forward pass and feature maps are extracted. These feature maps can be used for object classification, bounding box regression, and mask generation in subsequent steps.

[0069] 3. Object Classification and Bounding Box Regression: Using the extracted feature maps, candidate boxes are generated, and object classification and accurate bounding box regression are performed for each candidate box. This can be accomplished by treating the matching between the candidate box and the manually annotated ground-truth box as a multi-class classification problem.

[0070] 4. Mask Generation: After obtaining the results of object classification and bounding box regression, a fully convolutional network is used to generate a corresponding mask for each candidate box. Mask generation can be considered as a binary classification process for each pixel, dividing it into object and background.

[0071] 5. Loss calculation and backpropagation: Calculate the loss function for object classification, bounding box regression, and mask generation, and propagate the loss back to the model. Through this process, the model can adjust the model weights according to the gradient of the loss function to improve the performance of the model.

[0072] 6. Model evaluation: During the training process, the validation set can be used to evaluate the performance of the model, and the model can be adjusted and parameters optimized based on the evaluation results.

[0073] 7. Repeat iteration: Repeat steps 2 to 6 above until the model converges and obtains the target Mask R-CNN model.

[0074] In one embodiment, the training and validation operations include: training an initial Mask R-CNN model using a training set, and validating and optimizing the initial Mask R-CNN model after training the training set using a validation set.

[0075] The training set is used to train the initial Mask R-CNN model. This trained model is then used to segment and identify objects in the augmented images. The validation set is used to modify model parameters and repeat the training process to further optimize and adapt the model to the new dataset and specific recognition task. During this process, model parameters can be modified, the network structure adjusted, or the diversity of the training data can be increased to improve model performance and accuracy.

[0076] Figure 6 shows the results of the initial Mask R-CNN model training. In the figure, mAP stands for mean average precision, which measures model accuracy. Eval mAP represents the evaluation function for mean average precision. epoch represents the training cycle in which all electron microscopy images in the training set are trained once. Train Loss and loss both represent the loss function. lr stands for learning rate. Step represents the number of training iterations. As can be seen from the figure, accuracy increases with increasing epochs. As the number of steps increases, the loss function and learning rate decrease and stabilize, serving as a criterion for whether the model is fully trained.

[0077] In one embodiment, inputting the image to be detected into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected includes: inputting the image to be detected into the target Mask R-CNN model to obtain the recognition result including the position and shape characteristics of the conductive carbon black crystal structure.

[0078] The image to be inspected is acquired in the same manner as the original electron microscope image, and image enhancement processing can be performed as needed. After acquiring the image to be inspected, it is input into the target Mask R-CNN model, which generates a binary mask corresponding to the image to be inspected. The mask contains information about the target lattice structure shape and its spatial layout. Based on the mask, the recognition result is obtained, which includes the position and shape characteristics of the conductive carbon black crystal structure.

[0079] As shown in Figure 7, the left image is the image to be detected, and the right image is the lattice structure annotated by the target Mask R-CNN model. As can be seen from the figure, the target Mask R-CNN model can obtain the position and shape characteristics of the conductive carbon black crystal structure.

[0080] This application uses a high-resolution transmission electron microscope as an analysis tool to identify and extract lattice fringes of conductive carbon black. Compared with traditional methods, the image processing method of this application has higher flexibility and automation level, and the operation process is simple. In addition, the noise screening method is optimized, and the binary image distortion problem caused by grayscale, binarization and multiple enhancement of microcrystalline fringes features is processed. In order to ensure the reliability of model training, a sample data set is artificially produced to provide reliable samples required for training the model. At the same time, the Mask R-CNN model is used to effectively avoid the loss of a large amount of useful information, ensuring the rapid and accurate identification and extraction of high-resolution transmission electron microscope images. This method can meet the demand for efficient processing of conductive carbon black images. Through the technical means of this application, the lattice fringes of conductive carbon black can be accurately identified and extracted. The application of optimized image processing methods and training models enables us to quickly and accurately identify the position and shape characteristics of target lattice fringes in high-resolution transmission electron microscope images.

[0081] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0082] Based on the same inventive concept, embodiments of the present application also provide a conductive carbon black lattice structure identification device for implementing the aforementioned conductive carbon black lattice structure identification method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more of the following embodiments of the conductive carbon black lattice structure identification device can be found in the aforementioned limitations of the conductive carbon black lattice structure identification method and will not be further elaborated here.

[0083] In an exemplary embodiment, as shown in FIG8 , a conductive carbon black lattice structure identification device is provided, comprising: a sample acquisition module 802 , a model training module 804 and an identification module 806 , wherein:

[0084] The sample acquisition module 802 is used to acquire a sample electron microscope image of the conductive carbon black, annotate the crystal structure of the conductive carbon black in the sample electron microscope image, and establish a sample data set based on the annotated sample electron microscope image;

[0085] A model training module 804 is used to input the sample data set into the initial Mask R-CNN model for model training to obtain a target Mask R-CNN model; and

[0086] The recognition module 806 is used to input the image to be detected into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

[0087] In one exemplary embodiment, the sample acquisition module 802 is used to collect an original electron microscope image of the conductive carbon black, perform image enhancement processing on the original electron microscope image, and obtain a sample electron microscope image.

[0088] In one exemplary embodiment, the sample acquisition module 802 is used to crop the original electron microscope image, perform a first image parameter adjustment based on the cropped original electron microscope image to obtain a first image; perform grayscale processing on the first image to obtain a second image; perform Fourier-inverse Fourier transform processing on the second image to obtain a third image; perform a second image parameter adjustment on the third image to obtain a sample electron microscope image; and the first image parameter adjustment and the second image parameter adjustment are respectively used to adjust the basic parameters of the original electron microscope image and the basic parameters of the third image, and the basic parameters include at least one of contrast, saturation and threshold.

[0089] In one exemplary embodiment, the sample dataset includes a training set and a validation set. The model training module 804 is further configured to input the training set and the validation set into an initial Mask R-CNN model, iteratively performing training and validation operations until the initial Mask R-CNN model converges, thereby obtaining a target Mask R-CNN model.

[0090] The training and validation operations include: using the training set to train the initial Mask R-CNN model, and using the validation set to validate and optimize the initial Mask R-CNN model trained with the training set.

[0091] In one exemplary embodiment, the recognition module 806 is further configured to input the image to be detected into the target Mask R-CNN model to obtain recognition results including position and shape features of the conductive carbon black crystal structure.

[0092] Each module in the conductive carbon black lattice structure identification device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in hardware form, or stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0093] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in Figure 9. The computer device includes a processor, a memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile or volatile storage medium and an internal memory. The non-volatile or volatile storage medium stores an operating system, computer-readable instructions, and a database. The internal memory provides an environment for the operation of the operating system and computer-readable instructions in the non-volatile or volatile storage medium. The database of the computer device is used to store federated learning data. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer-readable instructions are executed by the processor, a method for identifying the lattice structure of conductive carbon black is implemented.

[0094] Those skilled in the art will understand that the structure shown in Figure 9 is merely a block diagram of a portion of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0095] A computer device includes a memory and one or more processors, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processors, the one or more processors perform the following steps:

[0096] Obtaining a sample electron microscope image of conductive carbon black, annotating the crystal structure of the conductive carbon black in the sample electron microscope image, and establishing a sample data set based on the annotated sample electron microscope image;

[0097] Input the sample dataset into the initial Mask R-CNN model for model training to obtain the target Mask R-CNN model; and

[0098] The image to be detected is input into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

[0099] One or more computer-readable storage media storing computer-readable instructions, which may be non-volatile or volatile. When the computer-readable instructions are executed by one or more processors, the one or more processors perform the following steps:

[0100] Obtaining a sample electron microscope image of conductive carbon black, annotating the crystal structure of the conductive carbon black in the sample electron microscope image, and establishing a sample data set based on the annotated sample electron microscope image;

[0101] Input the sample dataset into the initial Mask R-CNN model for model training to obtain the target Mask R-CNN model; and

[0102] The image to be detected is input into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

[0103] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, performs the following steps:

[0104] Obtaining a sample electron microscope image of conductive carbon black, annotating the crystal structure of the conductive carbon black in the sample electron microscope image, and establishing a sample data set based on the annotated sample electron microscope image;

[0105] Input the sample dataset into the initial Mask R-CNN model for model training to obtain the target Mask R-CNN model; and

[0106] The image to be detected is input into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

[0107] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0108] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through computer-readable instructions. The computer-readable instructions can be stored in a non-volatile computer-readable storage medium. When the computer-readable instructions are executed, they can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processor involved in the various embodiments provided herein may be, but are not limited to, a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, and the like.

[0109] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0110] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for identifying the lattice structure of conductive carbon black, comprising: Acquire a sample electron microscope image of conductive carbon black, annotate the crystal structure of the conductive carbon black in the sample electron microscope image, and establish a sample data set according to the annotated sample electron microscope image; Inputting the sample data set into the initial Mask R-CNN model for model training to obtain a target Mask R-CNN model; and The image to be detected is input into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

2. The method according to claim 1, wherein: The method of obtaining a sample electron microscope image of conductive carbon black comprises: An original electron microscope image of the conductive carbon black is collected, and image enhancement processing is performed on the original electron microscope image to obtain the sample electron microscope image.

3. The method according to claim 2, wherein: The performing image enhancement processing on the original electron microscope image to obtain the sample electron microscope image comprises: Cropping the original electron microscope image, and adjusting a first image parameter based on the cropped original electron microscope image to obtain a first image; Performing grayscale processing on the first image to obtain a second image; Performing Fourier-inverse Fourier transform processing on the second image to obtain a third image; Performing second image parameter adjustment on the third image to obtain the electron microscope image of the sample; and The first image parameter adjustment and the second image parameter adjustment are used to adjust the basic parameters of the original electron microscope image and the basic parameters of the third image respectively, and the basic parameters include at least one of contrast, saturation and threshold.

4. The method according to claim 1, wherein: The sample data set includes a training set and a validation set; Inputting the sample data set into the initial Mask R-CNN model for model training to obtain the target Mask R-CNN model includes: The training set and the validation set are input into the initial Mask R-CNN model, and training and validation operations are iteratively performed until the initial Mask R-CNN model converges to obtain the target Mask R-CNN model.

5. The method according to claim 4, wherein: The training and validation operations include: The initial Mask R-CNN model is trained using the training set, and the initial Mask R-CNN model trained with the training set is verified and optimized using the verification set.

6. The method according to claim 1, wherein: The step of inputting the image to be detected into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected includes: The image to be detected is input into the target Mask R-CNN model to obtain the recognition result including the position and shape characteristics of the conductive carbon black crystal structure.

7. A conductive carbon black lattice structure identification device, characterized in that: The device comprises: A sample acquisition module, used to acquire a sample electron microscope image of conductive carbon black, annotate the crystal structure of the conductive carbon black in the sample electron microscope image, and establish a sample data set according to the annotated sample electron microscope image; A model training module, used to input the sample data set into the initial Mask R-CNN model for model training to obtain a target Mask R-CNN model; and The recognition module is used to input the image to be detected into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

8. The device according to claim 7, wherein: The sample acquisition module is also used to collect the original electron microscope image of the conductive carbon black, perform image enhancement processing on the original electron microscope image, and obtain the sample electron microscope image.

9. The device according to claim 8, wherein: The sample acquisition module is also used to crop the original electron microscope image, adjust the first image parameters based on the cropped original electron microscope image, and acquire the first image; Performing grayscale processing on the first image to obtain a second image; Performing Fourier-inverse Fourier transform processing on the second image to obtain a third image; Adjusting the parameters of the second image on the third image to obtain an electron microscope image of the sample; The first image parameter adjustment and the second image parameter adjustment are used to adjust the basic parameters of the original electron microscope image and the basic parameters of the third image respectively, and the basic parameters include at least one of contrast, saturation and threshold.

10. The device according to claim 7, wherein: The sample data set includes a training set and a validation set; the model training module is also used to input the training set and the validation set into the initial Mask R-CNN model, iteratively perform training and validation operations until the initial Mask R-CNN model converges to obtain the target Mask R-CNN model.

11. The device according to claim 10, wherein: The model training module is also used to train the initial Mask R-CNN model using the training set, and to verify and optimize the initial Mask R-CNN model after training the training set using the verification set.

12. The device according to claim 7, wherein: The recognition module is also used to input the image to be detected into the target Mask R-CNN model to obtain recognition results including the position and shape characteristics of the conductive carbon black crystal structure.

13. A computer device comprising a memory and one or more processors, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the one or more processors, the one or more processors perform the following steps: Acquire a sample electron microscope image of conductive carbon black, annotate the crystal structure of the conductive carbon black in the sample electron microscope image, and establish a sample data set according to the annotated sample electron microscope image; Inputting the sample data set into the initial Mask R-CNN model for model training to obtain a target Mask R-CNN model; and The image to be detected is input into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

14. The method according to claim 13, wherein: The method of obtaining a sample electron microscope image of conductive carbon black comprises: An original electron microscope image of the conductive carbon black is collected, and image enhancement processing is performed on the original electron microscope image to obtain the sample electron microscope image.

15. The method according to claim 14, wherein: The performing image enhancement processing on the original electron microscope image to obtain the sample electron microscope image comprises: Cropping the original electron microscope image, and adjusting a first image parameter based on the cropped original electron microscope image to obtain a first image; Performing grayscale processing on the first image to obtain a second image; Performing Fourier-inverse Fourier transform processing on the second image to obtain a third image; Performing second image parameter adjustment on the third image to obtain the electron microscope image of the sample; and The first image parameter adjustment and the second image parameter adjustment are used to adjust the basic parameters of the original electron microscope image and the basic parameters of the third image respectively, and the basic parameters include at least one of contrast, saturation and threshold.

16. The method according to claim 13, wherein: The sample data set includes a training set and a validation set; Inputting the sample data set into the initial Mask R-CNN model for model training to obtain the target Mask R-CNN model includes: The training set and the validation set are input into the initial Mask R-CNN model, and training and validation operations are iteratively performed until the initial Mask R-CNN model converges to obtain the target Mask R-CNN model.

17. The method according to claim 16, wherein: The training and validation operations include: The initial Mask R-CNN model is trained using the training set, and the initial Mask R-CNN model trained with the training set is verified and optimized using the verification set.

18. The method according to claim 13, wherein: The step of inputting the image to be detected into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected includes: The image to be detected is input into the target Mask R-CNN model to obtain the recognition result including the position and shape characteristics of the conductive carbon black crystal structure.

19. One or more computer-readable storage media storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the following steps: Acquire a sample electron microscope image of conductive carbon black, annotate the crystal structure of the conductive carbon black in the sample electron microscope image, and establish a sample data set according to the annotated sample electron microscope image; Inputting the sample data set into the initial Mask R-CNN model for model training to obtain a target Mask R-CNN model; and The image to be detected is input into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

20. The method according to claim 19, wherein: The method of obtaining a sample electron microscope image of conductive carbon black comprises: An original electron microscope image of the conductive carbon black is collected, and image enhancement processing is performed on the original electron microscope image to obtain the sample electron microscope image.

21. The method according to claim 20, wherein: The performing image enhancement processing on the original electron microscope image to obtain the sample electron microscope image comprises: Cropping the original electron microscope image, and adjusting a first image parameter based on the cropped original electron microscope image to obtain a first image; Performing grayscale processing on the first image to obtain a second image; Performing Fourier-inverse Fourier transform processing on the second image to obtain a third image; Performing second image parameter adjustment on the third image to obtain the electron microscope image of the sample; and The first image parameter adjustment and the second image parameter adjustment are used to adjust the basic parameters of the original electron microscope image and the basic parameters of the third image respectively, and the basic parameters include at least one of contrast, saturation and threshold.

22. The method according to claim 19, wherein: The sample data set includes a training set and a validation set; Inputting the sample data set into the initial Mask R-CNN model for model training to obtain the target Mask R-CNN model includes: The training set and the validation set are input into the initial Mask R-CNN model, and training and validation operations are iteratively performed until the initial Mask R-CNN model converges to obtain the target Mask R-CNN model.

23. The method according to claim 22, wherein: The training and validation operations include: The initial Mask R-CNN model is trained using the training set, and the initial Mask R-CNN model trained with the training set is verified and optimized using the verification set.

24. The method according to claim 19, wherein: The step of inputting the image to be detected into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected includes: The image to be detected is input into the target Mask R-CNN model to obtain the recognition result including the position and shape characteristics of the conductive carbon black crystal structure.

25. A computer program product comprising a computer program, wherein: When the computer program is executed by a processor, the following steps are implemented: Acquire a sample electron microscope image of conductive carbon black, annotate the crystal structure of the conductive carbon black in the sample electron microscope image, and establish a sample data set according to the annotated sample electron microscope image; Inputting the sample data set into the initial Mask R-CNN model for model training to obtain a target Mask R-CNN model; and The image to be detected is input into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected.

26. The method according to claim 25, wherein: The method of obtaining a sample electron microscope image of conductive carbon black comprises: An original electron microscope image of the conductive carbon black is collected, and image enhancement processing is performed on the original electron microscope image to obtain the sample electron microscope image.

27. The method according to claim 26, wherein: The performing image enhancement processing on the original electron microscope image to obtain the sample electron microscope image comprises: Cropping the original electron microscope image, and adjusting a first image parameter based on the cropped original electron microscope image to obtain a first image; Performing grayscale processing on the first image to obtain a second image; Performing Fourier-inverse Fourier transform processing on the second image to obtain a third image; Performing second image parameter adjustment on the third image to obtain the electron microscope image of the sample; and The first image parameter adjustment and the second image parameter adjustment are used to adjust the basic parameters of the original electron microscope image and the basic parameters of the third image respectively, and the basic parameters include at least one of contrast, saturation and threshold.

28. The method according to claim 25, wherein: The sample data set includes a training set and a validation set; Inputting the sample data set into the initial Mask R-CNN model for model training to obtain the target Mask R-CNN model includes: The training set and the validation set are input into the initial Mask R-CNN model, and training and validation operations are iteratively performed until the initial Mask R-CNN model converges to obtain the target Mask R-CNN model.

29. The method according to claim 28, wherein: The training and validation operations include: The initial Mask R-CNN model is trained using the training set, and the initial Mask R-CNN model trained with the training set is verified and optimized using the verification set.

30. The method of claim 25, wherein: The step of inputting the image to be detected into the target Mask R-CNN model to obtain the recognition result of the conductive carbon black crystal structure in the image to be detected includes: The image to be detected is input into the target Mask R-CNN model to obtain the recognition result including the position and shape characteristics of the conductive carbon black crystal structure.

Images