Plasmodium category identification method and device based on thin blood membrane

Abstract

The invention provides a plasmodium category identification method and device based on a thin blood membrane. The method comprises the following steps: acquiring a thin blood membrane image to be identified; inputting the thin blood membrane image to be recognized into a pre-trained classification recognition model to obtain a plasmodium category recognition result output by the classification recognition model; wherein the classification and identification model is obtained by distilling a plurality of expert models by utilizing self-paced learning and course learning, the plurality of expert models are obtained by training different tail category data in unlabeled data, and the unlabeled data is obtained by utilizing an unsupervised upstream task. The problems that in the prior art, plasmodium recognition accuracy is poor, time consumed in the recognition process is long, data samples are few, and the number of sample data in the category is unbalanced, so that artificial intelligence recognition cannot be achieved are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical data processing, and in particular to a method and device for identifying malarial parasite types based on thin blood films. Background Art

[0002] Malaria is a parasitic disease prevalent in tropical and subtropical regions. It is caused by the bite of female Anopheles mosquitoes carrying malarial parasites. There are four main types of malarial parasites that infect humans: Plasmodium falciparum, Plasmodium vivax, Plasmodium malariae and Plasmodium ovale. The most common ones are Plasmodium falciparum and Plasmodium vivax, and Plasmodium falciparum is the most dangerous. Different types of malarial parasites can show obvious differences in thin blood film morphology, including parasitized red blood cells, small trophozoites (ring bodies), large trophozoites, immature schizonts, mature schizonts, male gametocytes, and female gametocytes. Therefore, the identification of the type of malarial parasites is of great significance for studying the course of malaria.

[0003] In the existing technology, the identification of malarial parasites can be carried out by traditional medical means and artificial intelligence means. Among them, the identification of malarial parasites by traditional medical means mainly includes the following four methods:

[0004] (1) Microscopic examination: Doctors use a microscope to observe peripheral blood smears to check for malarial parasites. This is a routine method for diagnosing malaria. Usually, the entire blood slide needs to be observed under a 100x objective lens. Generally, thousands of fields of view need to be checked, which takes a day. If malarial parasites are found, the result is positive, which can be used as a reliable basis for diagnosis. A negative result cannot negate the diagnosis, and multiple reexaminations are required.

[0005] (2) Cultivation: Cultivate the microorganisms in the sample under certain nutritional conditions and observe the growth characteristics of reproduction and amplification to determine whether parasites are present;

[0006] (3) Antigen or antibody testing: by injecting specific antibodies (or antigens) into the sample and observing whether a specific reaction occurs, it is possible to infer whether the microorganism contains parasites;

[0007] (4) Genetic testing: Use nucleic acid hybridization, gene chips, polymerase reaction and other technologies to detect the nucleic acid sequence of parasites to determine whether parasites are present.

[0008] Due to the small amount of data, there are relatively few artificial intelligence technologies for malarial parasite classification. There are currently four different solutions for medical image classification tasks with a small total data volume (small sample) and unbalanced data volume. Data imbalance refers to the serious imbalance in the proportion of data volume in each category of the data set used for network training during the classification process. We call the category with a large amount of data the head category, the category with a moderate amount of data the middle category, and the category with a small amount of data the tail category.

[0009] (1) Data augmentation: This type of method usually augments the data of the tail category to increase the amount of data in the tail category, thus transforming the problem into a balanced data classification problem.

[0010] (2) Resampling: This method refers to selecting the same number of instances in each category from the existing data set to form a balanced data set to complete the classification task;

[0011] (3) Reweighting: In the process of training the neural network, different weights are assigned to different categories of data, so that the network has a greater bias towards the tail category, thereby solving the problem of data imbalance;

[0012] (4) Multi-expert model: Through a method similar to ensemble learning, multiple neural networks are established to classify data in parallel. During the training process of each network, the data received is different, and the number of rounds of tail data input is more than that of the head category, thus solving the problem of data imbalance to a certain extent.

[0013] However, for existing AI methods, in solving the long-tail problem of small data sets, data augmentation methods can cause overfitting of the network; resampling and reweighting can weaken the accuracy of the head category. None of these methods can effectively solve the problem of data imbalance, and since the overall number of native data sets is small, none of these methods can effectively solve the problem of insufficient data samples.

[0014] In view of this, a method and device for identifying the category of Plasmodium based on a thin blood film are provided, in the hope of solving the problems existing in the prior art, such as poor accuracy in identifying Plasmodium, a long identification process, and the inability to achieve artificial intelligence identification due to a small number of data samples and an imbalance in the number of sample data categories. Summary of the invention

[0015] The present invention provides a method and device for identifying malarial parasite categories based on thin blood films, in the hope of solving the problems in the prior art of poor malarial parasite identification accuracy, long identification process, and inability to implement artificial intelligence identification due to small number of data samples and unbalanced number of sample data categories.

[0016] The present invention provides a method for identifying the type of malarial parasite based on a thin blood film, the method comprising:

[0017] acquiring an image of a thin blood film to be identified;

[0018] Inputting the thin blood film image to be identified into a pre-trained classification and recognition model to obtain a malarial parasite category recognition result output by the classification and recognition model;

[0019] The classification and recognition model is obtained by distilling multiple expert models using self-paced learning and curriculum learning. The multiple expert models are trained based on different tail category data in unlabeled data. The unlabeled data is obtained using unsupervised upstream tasks.

[0020] In some embodiments, the training process of the classification recognition model includes:

[0021] Use unsupervised upstream tasks to obtain a large amount of unlabeled data;

[0022] The unlabeled data is divided into multiple tail category data, and the initial expert network is classified and trained based on different tail category data to obtain multiple expert models;

[0023] A plurality of expert models are distilled based on a distillation network to obtain the classification and recognition model.

[0024] In some embodiments, a large amount of unlabeled data is obtained by using unsupervised upstream tasks, specifically including:

[0025] Acquire label-free red blood cell images;

[0026] constructing a proxy task based on the red blood cell image through reconstruction loss so that the input red blood cell image has a preset degree of occlusion;

[0027] The occluded part of the image is restored according to the context information through the generative model to obtain unlabeled data.

[0028] In some embodiments, the unlabeled data is divided into a plurality of tail category data, and the initial expert network is classified and trained based on the different tail category data to obtain a plurality of expert models, specifically including:

[0029] Use the network structure of the upstream task as the backbone network;

[0030] The total task is divided into n subtasks according to the distribution of the data volume of each tail category data, and each expert model corresponds to a subtask;

[0031] Different tail category data are used to train each initial expert network separately to obtain multiple expert models.

[0032] In some embodiments, when n is 3, the number of expert models is three;

[0033] Among them, the first expert model is responsible for predicting all categories and inputting all category data. The second expert is only responsible for the classification prediction of the middle category and tail category data, and the third expert is only responsible for predicting the tail category.

[0034] In some embodiments, multiple expert models are distilled based on a distillation network to obtain the classification recognition model, and distillation is achieved through self-paced learning and curriculum learning.

[0035] The present invention also provides a device for identifying malarial parasite types based on a thin blood film, the device comprising:

[0036] A data acquisition unit, used for acquiring an image of a thin blood film to be identified;

[0037] A result generating unit, used for inputting the thin blood film image to be identified into a pre-trained classification and recognition model to obtain a malarial parasite category recognition result output by the classification and recognition model;

[0038] The classification and recognition model is obtained by distilling multiple expert models using self-paced learning and curriculum learning. The multiple expert models are trained based on different tail category data in unlabeled data. The unlabeled data is obtained using unsupervised upstream tasks.

[0039] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described above when executing the program.

[0040] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the method described above when executed by a processor.

[0041] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method described above is implemented.

[0042] The method and device for identifying the category of malarial parasites based on thin blood film provided by the present invention obtain the image of thin blood film to be identified; input the image of thin blood film to be identified into the pre-trained classification and recognition model, and obtain the malarial parasite category recognition result output by the classification and recognition model; wherein the classification and recognition model is obtained by distilling multiple expert models using self-paced learning and curriculum learning, and the multiple expert models are trained based on different tail category data in unlabeled data, and the unlabeled data is obtained using unsupervised upstream tasks. Thus, the problem of poor accuracy in malarial parasite identification, long time-consuming identification process, and the inability to realize artificial intelligence identification caused by the small number of data samples and the imbalance of sample data in categories in the prior art is solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 This is one of the flow charts of the method for identifying the type of Plasmodium based on thin blood film provided by the present invention;

[0045] Figure 2 This is the second flow chart of the method for identifying the type of Plasmodium based on thin blood film provided by the present invention;

[0046] Figure 3 This is the third flow chart of the method for identifying the type of Plasmodium based on thin blood film provided by the present invention;

[0047] Figure 4 It is a structural block diagram of a malarial parasite classification identification device based on thin blood film provided by the present invention;

[0048] Figure 5 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] In actual usage scenarios, malarial parasites can cause disease in the ring, macrotrophozoite, schizont and gametocyte stages. Judging the proportion of different stages is of certain significance for predicting the direction of the disease. Therefore, this embodiment sets the recognition range to four major stages of Plasmodium falciparum, Plasmodium vivax, Plasmodium malariae, Plasmodium ovale, Plasmodium knowlesi, and the ring, macrotrophozoite, schizont, and gametocyte of different parasite species, for a total of 20 categories (5 parasite species × 4 stages).

[0051] In order to solve the problems existing in the prior art, the problem of small number of data samples and unbalanced number of sample data categories in the artificial intelligence recognition process should be overcome.

[0052] In a specific embodiment, the present invention provides a method for identifying the type of Plasmodium based on thin blood film, such as Figure 1 As shown, the method comprises the following steps:

[0053] S110: Acquire a thin blood film image to be identified;

[0054] S120: inputting the thin blood film image to be identified into a pre-trained classification and recognition model to obtain a malarial parasite category recognition result output by the classification and recognition model;

[0055] The classification and recognition model is obtained by distilling multiple expert models using self-paced learning and curriculum learning. The multiple expert models are trained based on different tail category data in unlabeled data. The unlabeled data is obtained using unsupervised upstream tasks.

[0056] In some embodiments, Figure 2 As shown, the training process of the classification recognition model includes the following steps:

[0057] S210: Obtain a large amount of unlabeled data using unsupervised upstream tasks; specifically, obtain an unlabeled red blood cell image, construct a proxy task based on the red blood cell image through reconstruction loss so that the input red blood cell image has a preset degree of occlusion; restore the occluded part of the image according to context information through a generative model to obtain unlabeled data.

[0058] S220: dividing the unlabeled data into a plurality of tail category data, and performing classification training on the initial expert network based on the different tail category data to obtain a plurality of expert models;

[0059] S230: Distilling multiple expert models based on a distillation network to obtain the classification recognition model.

[0060] like Figure 3 As shown, in order to solve the problem of small samples, this embodiment introduces a semi-supervised algorithm. Because the carrier of malarial parasites is red blood cells, and unlabeled red blood cell samples are easy to obtain, therefore, in the early stage of training the neural network, unlabeled red blood cells are used to train the network first. The upstream task is the pre-training stage, and the purpose of pre-training is mainly to efficiently use unlabeled data. The cost of acquiring these unlabeled data is very low, so the amount of data is sufficient. At this stage, since the input is unlabeled red blood cell data, in order to smoothly optimize the parameters for network back propagation, reconstruction loss is used to construct the proxy task, which is mainly to occlude the input image to a certain extent, and restore the occluded part of the image according to the context information through the generative model.

[0061] In addition, the downstream task is defined as a long-tail data classification problem. Multiple expert models are selected to integrate and learn different types of features, so that more experts can focus on the tail category data, thereby alleviating the problem of data imbalance. Unlike the conventional multi-expert model method, the teacher-student distillation technology is used for multiple trained expert models, so that the student network has the ability to process all categories of data with high accuracy. In addition, in the process of distillation learning, the curriculum learning technology is used to allow the student model to perform distillation learning from simple to difficult.

[0062] In some embodiments, the unlabeled data is divided into a plurality of tail category data, and the initial expert network is classified and trained based on the different tail category data to obtain a plurality of expert models, specifically including:

[0063] Use the network structure of the upstream task as the backbone network;

[0064] The total task is divided into n subtasks according to the distribution of the data volume of each tail category data, and each expert model corresponds to one subtask; when n is 3, there are three expert models, the first expert model is responsible for predicting all categories and inputting all category data, the second expert is only responsible for the classification prediction of the middle category and the tail category data, and the third expert is only responsible for predicting the tail category;

[0065] Different tail category data are used to train each initial expert network separately to obtain multiple expert models.

[0066] Specifically, when distributing the training of the expert model, the network structure of the upstream task is first used as the backbone network, which can improve the utilization rate of malaria data. Then, the training of n expert models is carried out. The expert model can be regarded as an integrated learning model, that is, using several neural networks with similar network structures to jointly complete a set of classification tasks, and each expert network is responsible for a subtask. Regarding the classification of 20 categories (5 parasite species * 4 periods) of malarial parasites, this total task is divided into n subtasks according to the distribution of the amount of data in each category. According to the ratio of data volume, the 20 categories were divided into three subcategories, namely (1) head category (Plague falciparum - ring stage; Plague vivax - macrotrophozoite stage), (2) middle category (Plague malariae - gametocyte stage; Plague falciparum - gametocyte stage; Plague malariae - macrotrophozoite stage; Plague ovale - macrotrophozoite stage; Plague ovale - gametocyte stage), and (3) tail category (Plague knowlesi - all stages; Plague malariae - ring stage; Plague ovale - ring stage; Plague vivax - ring stage; Plague vivax - gametocyte stage; all parasite species - schizont stage; Plague falciparum - macrotrophozoite stage).

[0067] In this embodiment, n is 3, that is, there are three expert models. The first expert model is responsible for predicting all categories and inputting all category data; the second expert model is only responsible for the classification prediction of the middle category and the tail category data. The third expert is only responsible for predicting the tail category. This data input ratio method achieves a rebalance of the data volume from another perspective.

[0068] In some embodiments, multiple expert models are distilled based on a distillation network to obtain the classification recognition model, and distillation is achieved through self-paced learning and curriculum learning. Specifically, the student network is trained by self-paced learning and curriculum learning, and the teacher model is obtained through the above reverse method. Since the task of each teacher model is relatively simple, it is necessary to merge multiple teacher models into a powerful student model, so that the student network can obtain good results based on multiple tasks.

[0069] Self-paced learning is first introduced in the training process. Self-paced learning includes two supervision methods: 1) cross entropy loss of the real label of the image; 2) KL divergence between the logit of the teacher network and the logit of the student network. In the early stage of training, the teacher network takes the dominant position, while the real label takes the dominant position in the later stage.

[0070] In addition, the same course learning method is used to construct difficult samples by collecting incorrect prediction samples of the teacher model. In the early stage of network training, simple samples are input to the network to learn the network parameters, and difficult samples are input for learning in the later stage, so as to complete the course tasks from easy to difficult.

[0071] In the above specific implementation, the method for identifying the category of malarial parasites based on thin blood film provided by the present invention is to obtain a thin blood film image to be identified; input the thin blood film image to be identified into a pre-trained classification and recognition model, and obtain the malarial parasite category recognition result output by the classification and recognition model; wherein the classification and recognition model is obtained by distilling multiple expert models using self-paced learning and curriculum learning, and the multiple expert models are trained based on different tail category data in unlabeled data, and the unlabeled data is obtained using unsupervised upstream tasks. This solves the problem of poor accuracy in malarial parasite identification, long time-consuming identification process, and inability to achieve artificial intelligence identification due to a small number of data samples in the prior art.

[0072] In addition to the above method, the present invention also provides a device for identifying malarial parasites based on thin blood film, such as Figure 4 As shown, the device comprises:

[0073] A data acquisition unit 410 is used to acquire a thin blood film image to be identified;

[0074] A result generating unit 420 is used to input the thin blood film image to be identified into a pre-trained classification and recognition model to obtain a malarial parasite category recognition result output by the classification and recognition model;

[0075] The classification and recognition model is obtained by distilling multiple expert models using self-paced learning and curriculum learning. The multiple expert models are trained based on different tail category data in unlabeled data. The unlabeled data is obtained using unsupervised upstream tasks.

[0076] In some embodiments, the training process of the classification recognition model includes:

[0077] Use unsupervised upstream tasks to obtain a large amount of unlabeled data;

[0078] The unlabeled data is divided into multiple tail category data, and the initial expert network is classified and trained based on different tail category data to obtain multiple expert models;

[0079] A plurality of expert models are distilled based on a distillation network to obtain the classification and recognition model.

[0080] In some embodiments, a large amount of unlabeled data is obtained by using unsupervised upstream tasks, specifically including:

[0081] Acquire label-free red blood cell images;

[0082] constructing a proxy task based on the red blood cell image through reconstruction loss so that the input red blood cell image has a preset degree of occlusion;

[0083] The occluded part of the image is restored according to the context information through the generative model to obtain unlabeled data.

[0084] In some embodiments, the unlabeled data is divided into a plurality of tail category data, and the initial expert network is classified and trained based on the different tail category data to obtain a plurality of expert models, specifically including:

[0085] Use the network structure of the upstream task as the backbone network;

[0086] The total task is divided into n subtasks according to the distribution of the data volume of each tail category data, and each expert model corresponds to a subtask;

[0087] Different tail category data are used to train each initial expert network separately to obtain multiple expert models.

[0088] In some embodiments, when n is 3, the number of expert models is three;

[0089] Among them, the first expert model is responsible for predicting all categories and inputting all category data. The second expert is only responsible for the classification prediction of the middle category and tail category data, and the third expert is only responsible for predicting the tail category.

[0090] In some embodiments, multiple expert models are distilled based on a distillation network to obtain the classification recognition model, and distillation is achieved through self-paced learning and curriculum learning.

[0091] In the above specific implementation, the device for identifying malarial parasites based on thin blood film provided by the present invention obtains a thin blood film image to be identified; inputs the thin blood film image to be identified into a pre-trained classification and recognition model, and obtains the malarial parasite classification recognition result output by the classification and recognition model; wherein the classification and recognition model is obtained by distilling multiple expert models using self-paced learning and curriculum learning, and the multiple expert models are trained based on different tail category data in unlabeled data, and the unlabeled data is obtained using unsupervised upstream tasks. This solves the problem of poor malarial parasite recognition accuracy, long recognition process, and inability to achieve artificial intelligence recognition due to a small number of data samples in the prior art.

[0092] Figure 5 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 5 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530 and a communication bus 540, wherein the processor 510, the communication interface 520 and the memory 530 communicate with each other via the communication bus 540. The processor 510 may call the logic instructions in the memory 530 to execute the above method.

[0093] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the part of the technical solution of the present invention that essentially contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0094] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform the above method.

[0095] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0096] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying malarial parasite types based on thin blood films, characterized in that: The method comprises: acquiring an image of a thin blood film to be identified; Inputting the thin blood film image to be identified into a pre-trained classification and recognition model to obtain a malarial parasite category recognition result output by the classification and recognition model; The classification and recognition model is obtained by distilling multiple expert models using self-paced learning and curriculum learning. The multiple expert models are trained based on different tail category data in unlabeled data. The unlabeled data is obtained using unsupervised upstream tasks.

2. The method for identifying malarial parasite types based on thin blood film according to claim 1, characterized in that: The training process of the classification recognition model includes: Use unsupervised upstream tasks to obtain a large amount of unlabeled data; The unlabeled data is divided into multiple tail category data, and the initial expert network is classified and trained based on different tail category data to obtain multiple expert models; A plurality of expert models are distilled based on a distillation network to obtain the classification and recognition model.

3. The method for identifying malarial parasite types based on thin blood film according to claim 2, characterized in that: Use unsupervised upstream tasks to obtain a large amount of unlabeled data, including: Acquire label-free red blood cell images; constructing a proxy task based on the red blood cell image through reconstruction loss so that the input red blood cell image has a preset degree of occlusion; The occluded part of the image is restored according to the context information through the generative model to obtain unlabeled data.

4. The method for identifying malarial parasite types based on thin blood film according to claim 2, characterized in that: The unlabeled data is divided into multiple tail category data, and the initial expert network is classified and trained based on different tail category data to obtain multiple expert models, including: Use the network structure of the upstream task as the backbone network; The total task is divided into n subtasks according to the distribution of the data volume of each tail category data, and each expert model corresponds to a subtask; Different tail category data are used to train each initial expert network separately to obtain multiple expert models.

5. The method for identifying malarial parasite types based on thin blood film according to claim 4, characterized in that: When n is 3, there are three expert models; Among them, the first expert model is responsible for predicting all categories and inputting all category data. The second expert is only responsible for the classification prediction of the middle category and tail category data, and the third expert is only responsible for predicting the tail category.

6. The method for identifying malarial parasite types based on thin blood film according to claim 2, characterized in that: Based on the distillation network, multiple expert models are distilled to obtain the classification recognition model, and distillation is achieved through self-paced learning and curriculum learning.

7. A device for identifying malarial parasites based on thin blood film, characterized in that: The device comprises: A data acquisition unit, used for acquiring an image of a thin blood film to be identified; A result generating unit, used for inputting the thin blood film image to be identified into a pre-trained classification and recognition model to obtain a malarial parasite category recognition result output by the classification and recognition model; The classification and recognition model is obtained by distilling multiple expert models using self-paced learning and curriculum learning. The multiple expert models are trained based on different tail category data in unlabeled data. The unlabeled data is obtained using unsupervised upstream tasks.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images