A document recognition method, system and device based on large model

Abstract

The present invention belongs to the field of document recognition and provides a document recognition method, system, and device based on a large model. The method acquires document image data containing text information and preprocesses the acquired document image data. The preprocessed image is evaluated for image quality, language distribution, and layout complexity. The evaluation results are integrated to determine the processing path for the document image data. Based on the determined processing path, the corresponding optical character recognition model is called to extract the text information from the document image data. The extracted text data is subjected to in-depth semantic analysis using a pre-trained deep learning large model to extract key feature information. The text information or the extracted key feature information is subjected to rule compliance analysis, and a recognition analysis report is generated based on the analysis results. The present invention improves processing speed and accuracy, enhances adaptability to various document formats, and effectively addresses the limitations of the existing technology.

Description

Technical Field

[0001] The present invention belongs to the field of document recognition, and in particular relates to a document recognition method, system and device based on a large model. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Traditional document expense recognition relies primarily on manual review, a time-consuming and labor-intensive process that is also prone to errors due to human error. The advancement of artificial intelligence (AI) technology, particularly optical character recognition (OCR) and deep learning models, has created new opportunities for automated document recognition.

[0004] However, existing OCR systems are typically optimized for a single language and require switching models to process multilingual texts, resulting in complex processes and fragmented contextual semantics. They also lack the ability to jointly recognize mixed documents such as Chinese and English, especially in scenarios with similar characters (such as Latin letters and Chinese radicals) or complex typesetting (such as mixed bilingual typesetting within tables). The error rate is high.

[0005] On the other hand, existing OCR systems usually adopt fixed processing procedures and cannot dynamically adjust recognition strategies according to the input content. For example, complex models are used for high-definition invoices, while lightweight models are enabled for fuzzy receipts, resulting in waste of resources or decreased accuracy. Summary of the Invention

[0006] In order to solve the above problems, the present invention proposes a document recognition method, system and equipment based on a large model. The present invention can accurately extract and analyze key information in documents, ensure the efficiency and reliability of the financial data processing process, not only improve the processing speed and accuracy, but also enhance the system's adaptability to various document formats, effectively solving the limitations of the existing technology.

[0007] According to some embodiments, the present invention adopts the following technical solutions:

[0008] A document recognition method based on a large model includes the following steps:

[0009] Acquire document image data containing text information, and pre-process the acquired document image data;

[0010] Evaluate the image quality, language distribution, and layout complexity of the pre-processed images, and determine the processing path for the document image data based on the comprehensive evaluation results;

[0011] According to the determined processing path, a multilingual optical character recognition model or an optical character recognition model is called to extract text information from the document image data;

[0012] When the multilingual optical character recognition model is called, the text information in different languages ​​is mapped into a unified semantic space based on the extracted text information. The pre-trained deep learning model is used to perform in-depth semantic analysis on the extracted text data to extract key feature information.

[0013] Conduct rule compliance analysis on text information or extracted key feature information to determine whether it meets the predetermined rule requirements;

[0014] Based on the analysis results, an identification analysis report is generated.

[0015] As an optional implementation, the process of pre-processing the acquired document image data includes: performing several operations of cropping, rotation correction, denoising, grayscale conversion and binarization on the acquired document image data.

[0016] As an optional implementation, the process of evaluating the image quality, language distribution and layout complexity of the preprocessed image includes: using a pre-trained neural network model to evaluate the image quality, using a pre-trained text detection model to evaluate the distribution ratio and layout complexity of different languages, and normalizing the evaluated image quality and layout complexity.

[0017] As an optional implementation method, the process of comprehensively evaluating the results and determining the processing path of the document image data includes: using a path decision function to determine the optimal processing path The specific options selected are:

[0018] ;

[0019] in, is the normalized image quality score, ranging from , 1 is the best image quality, and are the distribution ratios of different languages, C is the normalized layout complexity, ranging from , 1 is the most complex, 、 、 、 are weight parameters, which are set based on historical data learning or experience. 、 is a preconfigured threshold;

[0020] Among them, the processing path After calling the multilingual optical character recognition model and the pre-trained deep learning model for analysis, rule compliance analysis is performed;

[0021] Only the optical character recognition model is called, without calling the pre-trained deep learning model, and the rule compliance analysis is directly performed on the recognized text information;

[0022] In order to call the optical character recognition model, the pre-configured key field information in the recognized text information is analyzed by calling the pre-trained deep learning model, and then the rule compliance analysis is performed.

[0023] As an optional implementation, when calling a multilingual optical character recognition model, the shared visual feature encoder of the multilingual optical character recognition model is used to synchronously locate the mixed-language text area, and the language adaptation decoder is used to perform character-level segmentation on the mixed-language text area to distinguish text and symbols in different languages, and automatically adjust the recognition weight of the corresponding language according to the context to achieve recognition of different languages.

[0024] As a further implementation method, the process of automatically adjusting the recognition weight of the corresponding language based on the context includes: if the context is semantically related and is the same first language, then the recognition weight of the first language at this time is increased to a set threshold; if the context is semantically unrelated and is not the same language, then the recognition weight of the predetermined language at this time is reduced to a minimum threshold.

[0025] As an optional implementation manner, the key feature information includes financial information such as amount, date and supplier name.

[0026] As an optional implementation method, the process of conducting rule compliance analysis includes: pre-building a rule engine, which contains relevant rules and standards corresponding to the target business needs, and using the rule engine to perform compliance checks based on the input text information or key feature information to ensure that all information content complies with the requirements of the corresponding relevant rules and standards.

[0027] A document recognition system based on a large model, comprising:

[0028] a preprocessing module configured to obtain document image data containing text information and preprocess the obtained document image data;

[0029] a processing path determination module configured to evaluate the image quality, language distribution, and layout complexity of the pre-processed image, and determine a processing path for the document image data based on the comprehensive evaluation results;

[0030] a model calling module configured to call a multilingual optical character recognition model or an optical character recognition model according to a determined processing path to extract text information from the document image data;

[0031] The semantic analysis module is configured to, when calling the multilingual optical character recognition model, map the text information in different languages ​​​​in the extracted text information into a unified semantic space based on the extracted text information, and use the pre-trained deep learning model to perform in-depth semantic analysis on the extracted text data to extract key feature information;

[0032] A rule compliance analysis module is configured to perform rule compliance analysis on the text information or the extracted key feature information to determine whether it complies with predetermined rule requirements;

[0033] The audit report generation module is configured to generate an identification analysis report based on the analysis results.

[0034] An electronic device includes a memory and a processor, and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, the steps in the above method are completed.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] The present invention incorporates multi-path decision-making before optical character recognition, comprehensively considers image quality, language distribution ratio, and layout complexity, and determines the processing path, thereby improving accuracy, processing speed, and efficiency. The invention is able to adapt to documents of different formats, enhance the generalization capability of the corresponding model through historical data sets, and achieve high-precision extraction of key information in documents, such as amount, date, supplier, etc., and perform compliance checks based on preset rules or machine learning algorithms to ensure the accuracy and comprehensiveness of document recognition.

[0037] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0039] Figure 1 The present invention is a flowchart of a document recognition method based on a large model according to an embodiment. DETAILED DESCRIPTION

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

[0041] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0042] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0043] In the absence of conflict, the embodiments and features in the embodiments of this application can be combined with each other.

[0044] Example 1

[0045] As mentioned in the background technology, existing solutions still have many shortcomings when processing documents with complex layouts and diverse formats, such as the inability to accurately locate text content, difficulty in understanding semantics in different business scenarios, and the inability to implement compliance judgments based on specific rules.

[0046] In order to solve the above shortcomings, the present invention provides a document recognition method based on a large model, such as Figure 1 As shown, the following steps are included:

[0047] Acquire document image data containing text information, and pre-process the acquired document image data;

[0048] Evaluate the image quality, language distribution, and layout complexity of the pre-processed images, and determine the processing path for the document image data based on the comprehensive evaluation results;

[0049] According to the determined processing path, a multilingual optical character recognition model or an optical character recognition model is called to extract text information from the document image data;

[0050] When the multilingual optical character recognition model is called, the text information in different languages ​​is mapped into a unified semantic space based on the extracted text information. The pre-trained deep learning model is used to perform in-depth semantic analysis on the extracted text data to extract key feature information.

[0051] Conduct rule compliance analysis on text information or extracted key feature information to determine whether it meets the predetermined rule requirements;

[0052] Based on the analysis results, an identification analysis report is generated.

[0053] In this embodiment, image data of a paper document is obtained by a scanner or a mobile device, and several operations including cropping, rotation correction, denoising, grayscale conversion and binarization are performed on the obtained document image data to optimize the subsequent OCR recognition effect.

[0054] Of course, in other embodiments, other pre-processing processes may also be added, such as geometric transformation, smoothing, etc. These can be implemented by selecting corresponding existing operation methods according to specific circumstances and will not be described in detail here.

[0055] In this embodiment, the process of evaluating the image quality, language distribution and layout complexity of the preprocessed image includes: using a pre-trained neural network model to evaluate the image quality, using a pre-trained text detection model to evaluate the distribution ratio and layout complexity of different languages, and normalizing the evaluated image quality and layout complexity.

[0056] Of course, in some embodiments, the neural network model can select a convolutional neural network model, a lightweight convolutional neural network model, a U-Net model, etc., which are not listed exhaustively here, as long as the neural network model can evaluate image quality, such as clarity, noise level, etc.

[0057] Similarly, in some embodiments, existing text detection models such as CTPN, EAST, etc. can be selected. We will not list them all here. As long as the distribution ratio of different languages ​​(taking Chinese and English as examples) and the layout complexity of the text can be determined, it will be sufficient.

[0058] In this embodiment, the evaluation results are comprehensively evaluated and a path decision function is used to determine the processing path of the document image data.

[0059] First, let me introduce the following three paths:

[0060] The first path, This is a high-precision path suitable for high-value documents such as contracts. This path uses multilingual optical character recognition models and deep learning models throughout the process to maximize accuracy.

[0061] The second path, This is a fast path suitable for simpler documents, such as receipts. It only requires calling ordinary optical character recognition models and rule templates. This path sacrifices some generalization to increase speed.

[0062] The third path, This is a hybrid path, suitable for other documents. It calls the ordinary optical character recognition model and only calls the deep learning large model for key fields to balance efficiency and accuracy.

[0063] Determine the optimal processing path using the path decision function The specific options selected are:

[0064] ;

[0065] in, is the normalized image quality score, ranging from , 1 is the best image quality, and are the distribution ratios of different languages, C is the normalized layout complexity, ranging from , 1 is the most complex, 、 、 、 are weight parameters, which are set based on historical data learning or experience. 、 is a preconfigured threshold.

[0066] In some embodiments, the path can also be dynamically switched based on intermediate results, such as the confidence level of optical character recognition results. For example, if the confidence level of the recognized amount in the fast path is less than 90%, it will automatically switch to the high-precision path for reprocessing.

[0067] In this embodiment, when calling a multilingual optical character recognition model, glyph fusion detection is first performed. The shared visual feature encoder of the multilingual optical character recognition model is used to synchronously locate mixed-language text areas. Taking Chinese and English as an example, the Chinese and English text areas are synchronously located to avoid missed detections caused by language differences in traditional solutions, such as ignoring English small-print labels.

[0068] Next, joint semantic segmentation is performed, using a language-adaptive decoder to perform character-level segmentation on mixed-language text regions, distinguishing between Chinese characters, English words, and symbols such as '¥' and '$'.

[0069] In this embodiment, the recognition weight of the corresponding language is automatically adjusted according to the context to achieve recognition of different languages. For example, if the context is semantically related and both are in Chinese, the recognition weight of Chinese at this time is increased to the set threshold; if the context is semantically unrelated and not in the same language, the recognition weight of the predetermined language at this time is reduced to the minimum threshold. This can avoid excessive error correction. For example, for an English company name in a Chinese invoice, using the solution of this embodiment, only low-weight error correction is required, which can improve the accuracy of key fields.

[0070] Before calling a large deep learning model for semantic analysis, the pre-trained multilingual embedding module (Multilingual Embedding) in the multilingual optical character recognition model can be used to map the extracted Chinese and English texts into a unified semantic space, solving the problem of weak correlation between bilingual fields in traditional solutions.

[0071] A pre-trained deep learning model is then used to perform in-depth semantic analysis on the extracted text data. This step helps understand the business logic and context behind the document content. Based on the results of the semantic analysis, key features are further extracted, such as the amount, date, supplier name, and other important financial information.

[0072] Deep learning large models can use existing large language models, such as BERT with Transformer architecture or other suitable large-scale language models.

[0073] In the above embodiment, the pre-training process of the model may also adopt an existing process, which will not be described in detail here.

[0074] Of course, some embodiments may also involve transfer learning and fine-tuning of large deep learning models. These are not introduced in this embodiment, and existing solutions can be used.

[0075] In this embodiment, a rule engine is pre-built, which contains relevant rules and standards corresponding to the target business needs. The rule engine is used to perform compliance checks based on the input text information or key feature information to ensure that all information content meets the requirements of the corresponding relevant rules and standards.

[0076] Finally, based on the above analysis results, a report containing detailed audit comments is automatically generated. Any abnormal items can be marked in the report and corresponding correction suggestions can be provided.

[0077] In some embodiments, an interactive user interface may also be provided to allow the user to view the audit results and, if necessary, manually adjust errors or confirm certain judgments.

[0078] In summary, this embodiment adds multi-path decision-making before optical character recognition, comprehensively considers image quality, language distribution ratio and layout complexity, and determines the processing path, thereby improving accuracy, processing speed and efficiency.

[0079] On the other hand, this embodiment can adapt to documents of different formats and perform targeted processing. It can enhance the generalization ability of the corresponding model through historical data sets, achieve high-precision extraction of key information in documents, such as amount, date, supplier, etc., and perform compliance checks based on preset rules or machine learning algorithms to ensure the accuracy and comprehensiveness of document recognition.

[0080] Example 2

[0081] A document recognition system based on a large model, comprising:

[0082] a preprocessing module configured to obtain document image data containing text information and preprocess the obtained document image data;

[0083] a processing path determination module configured to evaluate the image quality, language distribution, and layout complexity of the pre-processed image, and determine a processing path for the document image data based on the comprehensive evaluation results;

[0084] a model calling module configured to call a multilingual optical character recognition model or an optical character recognition model according to a determined processing path to extract text information from the document image data;

[0085] The semantic analysis module is configured to, when calling the multilingual optical character recognition model, map the text information in different languages ​​​​in the extracted text information into a unified semantic space based on the extracted text information, and use the pre-trained deep learning model to perform in-depth semantic analysis on the extracted text data to extract key feature information;

[0086] A rule compliance analysis module is configured to perform rule compliance analysis on the text information or the extracted key feature information to determine whether it complies with predetermined rule requirements;

[0087] The audit report generation module is configured to generate an identification analysis report based on the analysis results.

[0088] Of course, it is understandable that the above-mentioned units / modules can be separately or completely combined into one or several other units / modules, or one (or some) of the units can be further divided into multiple functionally smaller units to form a composition, which can achieve the same operation without affecting the realization of the technical effects of the embodiments of the present application.

[0089] The above modules of the system are divided based on logical functions. In actual applications, the functions of one module can also be implemented by multiple modules, or the functions of multiple modules can be implemented by one module.

[0090] Similarly, in other embodiments of the present application, the system may also include other units / modules. In actual applications, these functions may also be implemented with the assistance of other units, and may be implemented by collaboration of multiple units.

[0091] According to another embodiment of the present application, the system described in this embodiment can be constructed and the method of embodiment 1 can be implemented by running a computer program (including program code) capable of executing the steps involved in the corresponding method described in embodiment 1 on a general-purpose computing device such as a computer, which includes processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM). The computer program can be recorded on, for example, a computer-readable recording medium, and loaded into the above-mentioned computing device through the computer-readable recording medium and run therein.

[0092] Example 3

[0093] An electronic device includes a processor, a communication interface, and a computer-readable storage medium, wherein the processor, the communication interface, and the computer-readable storage medium can be connected via a bus or other means.

[0094] Among them, the communication interface is used to receive and send data, the computer-readable storage medium can be stored in the memory of the electronic device, the computer-readable storage medium is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer-readable storage medium.

[0095] A processor (or CPU (Central Processing Unit)) is the computing and control core of an electronic device. It is suitable for implementing one or more instructions, specifically for loading and executing one or more instructions to implement corresponding method processes or corresponding functions.

[0096] The processor is configured to perform the following process:

[0097] Acquire document image data containing text information, and pre-process the acquired document image data;

[0098] Evaluate the image quality, language distribution, and layout complexity of the pre-processed images, and determine the processing path for the document image data based on the comprehensive evaluation results;

[0099] According to the determined processing path, a multilingual optical character recognition model or an optical character recognition model is called to extract text information from the document image data;

[0100] When the multilingual optical character recognition model is called, the text information in different languages ​​is mapped into a unified semantic space based on the extracted text information. The pre-trained deep learning model is used to perform in-depth semantic analysis on the extracted text data to extract key feature information.

[0101] Conduct rule compliance analysis on text information or extracted key feature information to determine whether it meets the predetermined rule requirements;

[0102] Based on the analysis results, an identification analysis report is generated.

[0103] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of one or more computer-usable storage media (including but not limited to disk storage, CD - ROM , optical storage, etc.).

[0104] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0105] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0106] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0107] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made by those skilled in the art that fall within the spirit and principles of the present invention and do not require creative effort are intended to be within the scope of protection of the present invention.

Claims

1. A document recognition method based on a large model, characterized in that: The following steps are involved: Acquire document image data containing text information, and pre-process the acquired document image data; Evaluate the image quality, language distribution, and layout complexity of the pre-processed images, and determine the processing path for the document image data based on the comprehensive evaluation results; According to the determined processing path, a multilingual optical character recognition model or an optical character recognition model is called to extract text information from the document image data; When the multilingual optical character recognition model is called, the text information in different languages ​​is mapped into a unified semantic space based on the extracted text information. The pre-trained deep learning model is used to perform in-depth semantic analysis on the extracted text data to extract key feature information. Conduct rule compliance analysis on text information or extracted key feature information to determine whether it meets the predetermined rule requirements; Generate an identification analysis report based on the analysis results; Comprehensive evaluation results, the process of determining the processing path of the document image data includes: using the path decision function to determine the optimal processing path The specific options selected are: ； in, is the normalized image quality score, ranging from , 1 is the best image quality, and are the distribution ratios of different languages, C is the normalized layout complexity, ranging from , 1 is the most complex, 、 、 、 are weight parameters, which are set based on historical data learning or experience. 、 is a preconfigured threshold; Among them, the processing path To call the multilingual optical character recognition model, a deep learning model is called to extract key features; Only the optical character recognition model is called, without calling the pre-trained deep learning model, and the rule compliance analysis is directly performed on the recognized text information; In order to call the optical character recognition model, the pre-configured key field information in the recognized text information is analyzed by calling the pre-trained deep learning model, and then the rule compliance analysis is performed.

2. A document recognition method based on a large model as claimed in claim 1, characterized in that: The process of pre-processing the acquired document image data includes: performing several operations of cropping, rotation correction, denoising, graying and binarization on the acquired document image data.

3. A document recognition method based on a large model as claimed in claim 1, characterized in that: The process of evaluating the image quality, language distribution, and layout complexity of the preprocessed image includes: using a pre-trained neural network model to evaluate the image quality, using a pre-trained text detection model to evaluate the distribution ratio and layout complexity of different languages, and normalizing the evaluated image quality and layout complexity.

4. A document recognition method based on a large model as claimed in claim 1, characterized in that: When calling the multilingual optical character recognition model, the shared visual feature encoder of the multilingual optical character recognition model is used to synchronously locate the mixed-language text area, and the language adaptation decoder is used to perform character-level segmentation on the mixed-language text area, distinguish texts and symbols in different languages, and automatically adjust the recognition weight of the corresponding language according to the context to achieve recognition of different languages.

5. A document recognition method based on a large model as claimed in claim 4, characterized in that: The process of automatically adjusting the recognition weight of the corresponding language according to the context includes: if the context is semantically related and both are first languages, then increasing the recognition weight of the first language at that time to a set threshold; If the contexts are not semantically related and are not in the same language, the recognition weight of the predetermined language at this time is reduced to the minimum threshold.

6. A document recognition method based on a large model as claimed in claim 1, characterized in that: The key characteristic information includes financial information such as amount, date and supplier name.

7. A document recognition method based on a large model as claimed in claim 1, characterized in that: The process of conducting rule compliance analysis includes: pre-building a rule engine that contains relevant rules and standards corresponding to the target business needs, using the rule engine to perform compliance checks based on input text information or key feature information to ensure that all information content meets the requirements of the corresponding relevant rules and standards.

8. A document recognition system based on a large model, applying the document recognition method based on a large model according to claim 1, characterized in that: include: a preprocessing module configured to obtain document image data containing text information and preprocess the obtained document image data; a processing path determination module configured to evaluate the image quality, language distribution, and layout complexity of the pre-processed image, and determine a processing path for the document image data based on the comprehensive evaluation results; a model calling module configured to call a multilingual optical character recognition model or an optical character recognition model according to a determined processing path to extract text information from the document image data; The semantic analysis module is configured to, when calling the multilingual optical character recognition model, map the text information in different languages ​​​​in the extracted text information into a unified semantic space based on the extracted text information, and use the pre-trained deep learning model to perform in-depth semantic analysis on the extracted text data to extract key feature information; A rule compliance analysis module is configured to perform rule compliance analysis on the text information or the extracted key feature information to determine whether it complies with predetermined rule requirements; The audit report generation module is configured to generate an identification analysis report based on the analysis results.

9. An electronic device, characterized in that: The method comprises a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein the steps of the method according to any one of claims 1 to 7 are completed when the computer instructions are executed by the processor.

Images