A method for treating sludge transfer manifests

Abstract

The present application provides a method for processing a sludge transfer form, which belongs to the field of data processing technology. The method includes: obtaining multiple initial images, including multiple sludge transfer form images; determining multiple target images based on an initial image, wherein the target image is a partial image in the initial image, wherein each target image includes at least one of a signature image, a date image, and a seal image; performing text recognition on the target image, and determining multiple feature data based on the recognition result, wherein the feature data includes signature data, date data, and seal data; classifying and marking multiple initial images based on the multiple feature data, and classifying the sludge transfer form images corresponding to the initial images according to the classification marks. The present application also provides a sludge transfer form processing system and an electronic device.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a method for processing sludge transfer forms. Background Art

[0002] Sludge transfer forms are characterized by large volume and fixed digital features. For example, signatures, dates and other data need to be verified. At present, handwritten signatures are generally used in the process of counting sludge transfer forms. Automatic recognition of handwritten signatures and time information is of great significance for information collection and processing in the fields of finance, law, and medicine.

[0003] Significant progress has been made in the fields of information recognition technology and template matching technology at home and abroad. With the continuous application of deep learning algorithms, information recognition technology has improved recognition accuracy and has been widely used in digital transformation. Template matching technology has improved matching speed and efficiency with the support of hardware technology, and has developed towards intelligence. Information recognition and template matching technology can automatically identify and extract information. In the process of data recognition of sludge transfer forms, traditional manual input is prone to errors and is inefficient. Summary of the invention

[0004] The embodiments of the present application provide a sludge transfer form processing method and system.

[0005] In order to achieve the above objectives, this application adopts the following technical solutions:

[0006] In the first aspect, the embodiment of the present application proposes a sludge transfer form processing method, the method comprising:

[0007] Acquire a plurality of initial images, the plurality of initial images comprising a plurality of sludge transfer link images;

[0008] Determine multiple target images based on an initial image, where the target image is a partial image of the initial image, wherein each target image includes at least one of a signature image, a date image, and a seal image;

[0009] Performing text recognition on the target image, and determining a plurality of feature data according to the recognition result, the feature data including signature data, date data and seal data;

[0010] Based on the multiple feature data, the multiple initial images are classified and labeled, and the sludge transfer form images corresponding to the initial images are classified according to the classification labels.

[0011] In combination with the first aspect, in some embodiments, multiple initial images are classified and labeled based on multiple feature data, and sludge transfer form images corresponding to the initial images are classified according to the classification labels, including:

[0012] Determine all date data in an initial image, arrange multiple date data in an initial image according to date and time, and obtain the earliest date data and the latest date data among the multiple date data;

[0013] Whether the sludge transfer form corresponding to the initial image is compliant is determined based on the earliest date data and the latest date data, and the compliant initial image is marked as a compliant form, and the remaining initial images are marked as non-compliant forms.

[0014] In combination with the first aspect, in some implementations, text recognition is performed on the target image, and a plurality of feature data are determined according to the recognition result, the feature data including signature data, date data, and seal data, including:

[0015] Divide the initial image into multiple sub-images, the size of the sub-images is much smaller than the target image;

[0016] classifying the plurality of sub-images based on feature data of each sub-image, and determining a plurality of sub-image groups according to the classification results;

[0017] Separate all sub-images in each sub-image group into layers, and multiple sub-images form a target layer image;

[0018] Performing noise reduction processing on the target layer image, wherein the noise reduction process includes clearing discrete points in a plurality of target layer images;

[0019] Performing linear fitting on each target layer image so that multiple sub-images in the target layer image are in a linearly connected state;

[0020] Obtaining a preset font template, and fitting a plurality of preset linear font templates in the preset font template with the target layer image, wherein the fitting process includes scaling and twisting the preset linear font template;

[0021] According to the fitting result, all the text information in each target layer image is determined, and based on the text information in each target layer image, the text information in the initial image is determined.

[0022] In combination with the first aspect, in some implementations, multiple sub-images are classified based on feature data of each sub-image, and multiple sub-image groups are determined according to the classification results, including:

[0023] Matching is performed on each sub-image based on visual conditions, and characteristic data corresponding to the sub-image is determined according to the matching result, wherein the visual conditions include brightness, contrast and grayscale of each sub-image.

[0024] In combination with the first aspect, in some implementations, multiple sub-images are classified based on feature data of each sub-image, and multiple sub-image groups are determined according to the classification results, including:

[0025] A neural network model is obtained, feature data of each sub-image is imported into the neural network model, and a plurality of sub-image groups are determined according to an output result of the neural network model.

[0026] In combination with the first aspect, in some implementations, a neural network model is obtained, feature data of each sub-image is imported into the neural network model, and a plurality of sub-image groups are determined according to an output result of the neural network model, including:

[0027] Obtaining a training set, the training set including a training image and image feature data corresponding to the training image, wherein the image feature data includes color data, texture data, and shape data;

[0028] The training set is imported into the neural network training model and iterative calculation is performed. When the preset conditions are met, the training is stopped and the trained neural network training model is obtained.

[0029] In combination with the first aspect, in some implementations, a training set is imported into a neural network training model and iterative calculation is performed, training is stopped when a preset condition is met, and a trained neural network training model is obtained, including:

[0030] Determine the error value according to the loss function;

[0031] The gradient of the loss function to the neural network training model is determined according to the error value, and the training of the neural network training model is stopped when the gradient is less than the preset gradient.

[0032] In combination with the first aspect, in some implementations, determining the gradient of the loss function to the neural network training model according to the error value, and stopping the training of the neural network training model when the gradient is less than a preset gradient, includes:

[0033] Determine the gradient of the loss function for the neural network training model based on the back-propagation algorithm.

[0034] In the second aspect, the embodiment of the present application proposes a sludge transfer joint form processing system, which is configured as follows:

[0035] Acquire a plurality of initial images, the plurality of initial images comprising a plurality of sludge transfer link images;

[0036] Determine multiple target images based on an initial image, where the target image is a partial image of the initial image, wherein each target image includes at least one of a signature image, a date image, and a seal image;

[0037] Performing text recognition on the target image, and determining a plurality of feature data according to the recognition result, the feature data including signature data, date data and seal data;

[0038] Based on the multiple feature data, the multiple initial images are classified and labeled, and the sludge transfer form images corresponding to the initial images are classified according to the classification labels.

[0039] In conjunction with the second aspect, in some embodiments, the system is configured as follows:

[0040] Classifying and marking multiple initial images based on multiple feature data, and classifying sludge transfer form images corresponding to the initial images according to the classification marks, including:

[0041] Determine all date data in an initial image, arrange multiple date data in an initial image according to date and time, and obtain the earliest date data and the latest date data among the multiple date data;

[0042] Whether the sludge transfer form corresponding to the initial image is compliant is determined based on the earliest date data and the latest date data, and the compliant initial image is marked as a compliant form, and the remaining initial images are marked as non-compliant forms.

[0043] In conjunction with the second aspect, in some embodiments, the system is configured as follows:

[0044] Perform text recognition on the target image and determine multiple feature data based on the recognition results. The feature data includes signature data, date data, and seal data, including:

[0045] Divide the initial image into multiple sub-images, the size of the sub-images is much smaller than the target image;

[0046] classifying the plurality of sub-images based on feature data of each sub-image, and determining a plurality of sub-image groups according to the classification results;

[0047] Separate all sub-images in each sub-image group into layers, and multiple sub-images form a target layer image;

[0048] Performing noise reduction processing on the target layer image, wherein the noise reduction process includes clearing discrete points in a plurality of target layer images;

[0049] Performing linear fitting on each target layer image so that multiple sub-images in the target layer image are in a linearly connected state;

[0050] Obtaining a preset font template, and fitting a plurality of preset linear font templates in the preset font template with the target layer image, wherein the fitting process includes scaling and twisting the preset linear font template;

[0051] According to the fitting result, all the text information in each target layer image is determined, and based on the text information in each target layer image, the text information in the initial image is determined.

[0052] In conjunction with the second aspect, in some embodiments, the system is configured as follows:

[0053] Classifying the multiple sub-images based on the feature data of each sub-image, and determining multiple sub-image groups according to the classification results, including:

[0054] Matching is performed on each sub-image based on visual conditions, and characteristic data corresponding to the sub-image is determined according to the matching result, wherein the visual conditions include brightness, contrast and grayscale of each sub-image.

[0055] In conjunction with the second aspect, in some embodiments, the system is configured as follows:

[0056] Classifying the multiple sub-images based on the feature data of each sub-image, and determining multiple sub-image groups according to the classification results, including:

[0057] A neural network model is obtained, feature data of each sub-image is imported into the neural network model, and a plurality of sub-image groups are determined according to an output result of the neural network model.

[0058] In conjunction with the second aspect, in some embodiments, the system is configured as follows:

[0059] Obtain a neural network model, import feature data of each sub-image into the neural network model, and determine multiple sub-image groups according to the output results of the neural network model, including:

[0060] Obtaining a training set, the training set including a training image and image feature data corresponding to the training image, wherein the image feature data includes color data, texture data, and shape data;

[0061] The training set is imported into the neural network training model and iterative calculation is performed. When the preset conditions are met, the training is stopped and the trained neural network training model is obtained.

[0062] In conjunction with the second aspect, in some embodiments, the system is configured as follows:

[0063] Import the training set into the neural network training model and perform iterative calculations. When the preset conditions are met, stop the training and obtain the trained neural network training model, including:

[0064] Determine the error value according to the loss function;

[0065] The gradient of the loss function to the neural network training model is determined according to the error value, and the training of the neural network training model is stopped when the gradient is less than the preset gradient.

[0066] In conjunction with the second aspect, in some embodiments, the system is configured as follows:

[0067] Determine the gradient of the loss function to the neural network training model according to the error value, and stop training the neural network training model when the gradient is less than the preset gradient, including:

[0068] Determine the gradient of the loss function for the neural network training model based on the back-propagation algorithm.

[0069] A third aspect of an embodiment of the present invention provides an electronic device, the electronic device comprising:

[0070] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method proposed in the first aspect of the embodiment of the present invention.

[0071] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method provided in the first aspect of the embodiment of the present invention.

[0072] In summary, the above method and device have the following technical effects:

[0073] The present application proposes a method for processing sludge transfer forms, which first obtains multiple initial images, then determines multiple target images based on an initial image, where the target image is a partial image in the initial image, then performs text recognition on the target image, and determines multiple feature data based on the recognition result, where the feature data includes signature data, date data, and seal data, and finally classifies and labels the multiple initial images based on the multiple feature data, and classifies the sludge transfer form images corresponding to the initial images according to the classification labels. The present application proposes a method for processing sludge transfer forms, which improves the matching speed and efficiency of text information in sludge transfer forms, and improves the problem that traditional manual input is prone to errors and is inefficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A schematic flow chart of a sludge transfer joint form processing method proposed in an embodiment of the present application. DETAILED DESCRIPTION

[0075] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments 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.

[0076] See also Figure 1 The present application embodiment proposes a sludge transfer form processing method, comprising the following steps:

[0077] S101: Acquire multiple initial images, including multiple sludge transfer images

[0078] It is understandable that the sludge transfer form is photographed with a high-definition camera to obtain clear, high-quality images. This is the basis for subsequent text recognition and logical judgment. As an important document in the field of environmental protection, the sludge transfer form records in detail the information of each link such as the generation, transfer and disposal of sludge. Therefore, obtaining multiple initial images, that is, obtaining multiple sludge transfer form images, is crucial for subsequent image recognition, information extraction and data analysis.

[0079] In actual operation, obtaining multiple initial images usually involves multiple steps. First, it is necessary to determine the source and storage method of the sludge transfer form, such as whether it is a paper document or an electronic document. For paper documents, they can be digitized through devices such as scanners or mobile phones and converted into electronic images; for electronic documents, they can be directly extracted from the computer system.

[0080] After acquiring the initial image, a series of preprocessing operations are required to improve the accuracy and efficiency of subsequent processing. This includes image cropping, scaling, rotation and other adjustment operations, as well as image optimization operations such as noise removal and contrast enhancement. Through these preprocessing steps, it can be ensured that each initial image meets the processing requirements and provides a reliable data basis for subsequent analysis and recognition.

[0081] S102: Determine multiple target images based on an initial image, where the target image is a partial image in the initial image, wherein each target image includes at least one of a signature image, a date image, and a seal image.

[0082] It is understandable that in modern image processing technology, a key step is to determine multiple target images based on a given initial image. These target images are not complete copies of the original image, but are partial images carefully selected and extracted from the initial image. Each target image carries specific information, including at least one of the signature image, date image, and seal image.

[0083] Specifically, signature images usually refer to the part of an image that contains the signature of an individual or organization. They are usually unique and recognizable, and are an important basis for proving the authenticity and legitimacy of a document. Date images are the part of an image that records the time when a document was generated or processed. They are essential for tracking the circulation process and historical records of a document. Seal images are signs used by official agencies or organizations to verify the authenticity of a document. They usually have legal effect and are important evidence of the legitimacy of a document.

[0084] S103: Perform text recognition on the target image, and determine multiple feature data according to the recognition result, the feature data including signature data, date data and seal data.

[0085] It is understandable that for text recognition of a target image, not only can the text information in the image be extracted, but also multiple feature data can be further determined based on the information. These feature data include but are not limited to signature data, date data, and seal data, which are of great value for subsequent analysis, processing, and decision-making.

[0086] Specifically, the following steps may be taken to more accurately extract and recognize text information in the initial image.

[0087] First, the initial image can be segmented into multiple sub-images. The size of these sub-images is much smaller than the target image, which can better capture the local features and details in the image. During the segmentation process, image processing techniques such as sliding window method or image pyramid can be used to ensure that the sub-images are segmented evenly and fully cover the entire image.

[0088] Next, the sub-images can be classified based on the feature data of each sub-image. These feature data can include a variety of information such as color, texture, shape, etc., which helps to identify different elements in the image. The purpose of classification is to group sub-images with similar features into multiple sub-image groups. This helps to further analyze and process the information in the image.

[0089] As an implementation mode, each sub-image may be matched based on visual conditions, and feature data corresponding to the sub-image may be determined based on the matching result, wherein the visual conditions include brightness, contrast and grayscale of each sub-image.

[0090] In some other implementations, a neural network model may be obtained, feature data of each sub-image may be imported into the neural network model, and a plurality of sub-image groups may be determined based on output results of the neural network model.

[0091] As you can understand, a training set is a set of sample data with known answers, which is used to train a machine learning model. In the context of image processing, the training set usually includes a large number of training images, which can be of various types. The corresponding image feature data is the key information contained in these images, which can help the machine learning model understand and recognize the content in the image.

[0092] Image feature data includes many types, among which color data, texture data and shape data are the three most common feature types. Color data describes the color information of each pixel in the image. By extracting color features, the model can learn the color distribution and change rules of different objects in the image. Texture data describes the spatial relationship between pixels in the image, which can help the model recognize the texture pattern in the image, such as the thickness and direction of the texture. Shape data focuses on the outline and boundary information of objects in the image. By extracting shape features, the model can recognize the shape characteristics of different objects in the image.

[0093] After obtaining the training set, the next step is to import this data into the neural network training model for iterative calculation. During the training process, the model will continuously adjust the connection weights between neurons based on the image feature data in the training set to minimize the gap between the predicted value and the true value.

[0094] In order to meet the preset training stop conditions, some evaluation indicators and thresholds can be set, such as the value of the loss function, accuracy, etc. When the performance of the model on the training set reaches the preset threshold, it can be considered that the model has learned enough feature information. At this time, the training can be stopped and the trained neural network model can be obtained.

[0095] As an implementation method, the error value can be determined according to the loss function, and then the gradient of the loss function to the neural network training model can be determined according to the error value. When the gradient is less than a preset gradient, the training of the neural network training model is stopped.

[0096] In the neural network training process, the loss function plays a vital role. It is used to measure the difference between the model prediction result and the actual label, that is, the error value. Based on this error value, the gradient of the loss function to the neural network training model can be further determined in order to optimize the model. Specifically, the gradient of the loss function to the neural network training model can be determined based on the back propagation algorithm.

[0097] Specifically, the loss function calculates an error value based on the deviation between the model's predicted output and the true label. This error value can reflect the performance of the model in the current training state. Then, the gradient of the loss function to the model parameters is calculated based on this error value. The gradient represents the direction and speed of the loss function changing with the model parameters, which is the key information for model optimization.

[0098] During the training process, the model parameters are continuously updated based on the gradient information to reduce the error value and improve the performance of the model. However, when the gradient value gradually decreases and approaches the preset gradient, it means that the model has approached the optimal state, and continued training may not significantly improve the performance, and may even cause problems such as overfitting. Therefore, a preset gradient threshold is usually set, and when the gradient is less than the threshold, the training of the neural network training model is stopped.

[0099] For example, suppose that when processing an image classification task, the cross entropy loss function is used to measure the model's error. In the early stages of training, because the model has not yet fully learned the characteristics of the data, the loss function value is often high and the gradient is also large. As the training progresses, the model gradually learns the laws of the data, the loss function value gradually decreases, and the gradient also gradually decreases. When the gradient decreases below the preset threshold, it can be considered that the model has achieved good performance. At this time, stopping training can avoid wasting computing resources and time. By setting an appropriate preset gradient threshold to control the training process, you can effectively avoid overfitting problems and improve the generalization ability of the model.

[0100] Through the above process, a trained neural network model can be obtained, which has learned to extract key features from the image and can accurately identify and classify based on these features.

[0101] After determining the grouping in the sub-images, all sub-images in each sub-image group can be layer-separated. Layer separation is an image processing technology that can separate different elements in an image into different layers, thereby facilitating subsequent editing and processing. Through layer separation, multiple target layer images can be obtained, each layer image containing a portion of the information in the initial image. Then, the target layer image is subjected to denoising, wherein the denoising process includes clearing the discrete points in the multiple target layer images, and then performing linear fitting on each target layer image so that the multiple sub-images in the target layer image are in a linearly connected state.

[0102] After the noise reduction process is completed, each target layer image needs to be linearly fitted. Linear fitting is a mathematical method that can fit a set of data points into a straight line or curve. By performing linear fitting on the target layer image, multiple sub-images in the image can be linearly connected, thereby better preserving the structure and information in the image.

[0103] Next, you need to obtain the preset font template. The preset font template contains a variety of preset linear font templates, which can be scaled and twisted as needed to adapt to different application scenarios. Fit these preset linear font templates to the target layer image, and adjust the size, angle and other parameters of the template to match the text information in the target layer image.

[0104] Finally, based on the fitting results, all the text information in each target layer image can be determined. By comparing and analyzing the similarity between the preset linear font template and the target layer image, the text content in the image can be accurately identified. Based on the text information in each target layer image, the text information in the initial image can be further determined.

[0105] Through the above steps, text information can be effectively extracted and recognized from the initial image. This method combines multiple technologies such as image processing, classification, layer separation, noise reduction, and linear fitting to more accurately recognize and parse the text content in the image. In practical applications, this method can be widely used in image recognition, document processing, automatic translation and other fields, bringing convenience to our lives and work.

[0106] S104: Classify and label the multiple initial images based on the multiple feature data, and classify the sludge transfer form images corresponding to the initial images according to the classification labels.

[0107] Specifically, in the field of modern image processing and data analysis, accurate classification and labeling of images has become a crucial task. Especially when dealing with images with specific features and uses, such as sludge transfer form images, it is necessary to rely on efficient algorithms and technologies. Images often contain rich information, such as color, texture, shape, etc. By extracting and analyzing these feature data, a series of classification labels can be formed to distinguish different image categories.

[0108] Specifically, all date data in an initial image can be determined, multiple date data in an initial image can be arranged by date and time, and the earliest date data and the latest date data among the multiple date data can be obtained. Then, based on the earliest date data and the latest date data, it can be determined whether the sludge transfer form corresponding to the initial image is compliant, and the compliant initial image can be marked as a compliant form, and the remaining initial images can be marked as non-compliant forms.

[0109] It is understandable that multiple date data in an initial image are arranged by date and time. This can be achieved by parsing and comparing the extracted date data. The date data can be converted into a unified date format and sorted in chronological order. This step helps to better understand and analyze the relationship between date data, and facilitates subsequent analysis and judgment.

[0110] Based on the arrangement of date data, the earliest date data and the latest date data among multiple date data can be further obtained. These two data points are crucial for judging the compliance of the sludge transfer form. The earliest date data represents the start time of the form, while the latest date data represents the end time of the form. By comparing these two time points, the overall time span of the form can be understood, and then it can be judged whether it complies with relevant regulations.

[0111] After obtaining the earliest date data and the latest date data, the two data points can be used to determine whether the sludge transfer form corresponding to the initial image is compliant. The specific judgment criteria may vary depending on the region and industry, but usually include restrictions on the time span of the form, requirements for the completeness and accuracy of the date data, etc. If the time span of the form exceeds the specified range, or the date data is missing or incorrect, the form will be considered non-compliant.

[0112] Finally, the compliant initial images are marked as compliant slips for subsequent management and use. At the same time, non-compliant slips need to be further reviewed and processed to ensure that they are properly corrected and improved.

[0113] Through the above steps, the compliance of sludge transfer forms can be automatically judged and classified. This not only improves work efficiency and reduces the possibility of human errors, but also helps to improve the standardization and standardization of management.

[0114] In summary, by utilizing image processing technology and data analysis methods, the compliance of sludge transfer forms can be automatically judged and classified, providing strong support for environmental management and sustainable development.

[0115] The present application proposes a method for processing sludge transfer forms, which first obtains multiple initial images, then determines multiple target images based on an initial image, where the target image is a partial image in the initial image, then performs text recognition on the target image, and determines multiple feature data based on the recognition result, where the feature data includes signature data, date data, and seal data, and finally classifies and labels the multiple initial images based on the multiple feature data, and classifies the sludge transfer form images corresponding to the initial images according to the classification labels. The present application proposes a method for processing sludge transfer forms, which improves the matching speed and efficiency of text information in sludge transfer forms, and improves the problem that traditional manual input is prone to errors and is inefficient.

[0116] Based on the same inventive concept, the present application also proposes a sludge transfer joint form processing system, which is configured as follows:

[0117] Acquire a plurality of initial images, the plurality of initial images comprising a plurality of sludge transfer link images;

[0118] Determine multiple target images based on an initial image, where the target image is a partial image of the initial image, wherein each target image includes at least one of a signature image, a date image, and a seal image;

[0119] Performing text recognition on the target image, and determining a plurality of feature data according to the recognition result, the feature data including signature data, date data and seal data;

[0120] Based on the multiple feature data, the multiple initial images are classified and labeled, and the sludge transfer form images corresponding to the initial images are classified according to the classification labels.

[0121] In some embodiments, the system is configured to:

[0122] Classifying and marking multiple initial images based on multiple feature data, and classifying sludge transfer form images corresponding to the initial images according to the classification marks, including:

[0123] Determine all date data in an initial image, arrange multiple date data in an initial image according to date and time, and obtain the earliest date data and the latest date data among the multiple date data;

[0124] Whether the sludge transfer form corresponding to the initial image is compliant is determined based on the earliest date data and the latest date data, and the compliant initial image is marked as a compliant form, and the remaining initial images are marked as non-compliant forms.

[0125] In some embodiments, the system is configured to:

[0126] Perform text recognition on the target image and determine multiple feature data based on the recognition results. The feature data includes signature data, date data, and seal data, including:

[0127] Divide the initial image into multiple sub-images, the size of the sub-images is much smaller than the target image;

[0128] classifying the plurality of sub-images based on feature data of each sub-image, and determining a plurality of sub-image groups according to the classification results;

[0129] Separate all sub-images in each sub-image group into layers, and multiple sub-images form a target layer image;

[0130] Performing noise reduction processing on the target layer image, wherein the noise reduction process includes clearing discrete points in a plurality of target layer images;

[0131] Performing linear fitting on each target layer image so that multiple sub-images in the target layer image are in a linearly connected state;

[0132] Obtaining a preset font template, and fitting a plurality of preset linear font templates in the preset font template with the target layer image, wherein the fitting process includes scaling and twisting the preset linear font template;

[0133] According to the fitting result, all the text information in each target layer image is determined, and based on the text information in each target layer image, the text information in the initial image is determined.

[0134] In some embodiments, the system is configured to:

[0135] Classifying the multiple sub-images based on the feature data of each sub-image, and determining multiple sub-image groups according to the classification results, including:

[0136] Matching is performed on each sub-image based on visual conditions, and characteristic data corresponding to the sub-image is determined according to the matching result, wherein the visual conditions include brightness, contrast and grayscale of each sub-image.

[0137] In some embodiments, the system is configured to:

[0138] Classifying the multiple sub-images based on the feature data of each sub-image, and determining multiple sub-image groups according to the classification results, including:

[0139] A neural network model is obtained, feature data of each sub-image is imported into the neural network model, and a plurality of sub-image groups are determined according to an output result of the neural network model.

[0140] In some embodiments, the system is configured to:

[0141] Obtain a neural network model, import feature data of each sub-image into the neural network model, and determine multiple sub-image groups according to the output results of the neural network model, including:

[0142] Obtaining a training set, the training set including a training image and image feature data corresponding to the training image, wherein the image feature data includes color data, texture data, and shape data;

[0143] The training set is imported into the neural network training model and iterative calculation is performed. When the preset conditions are met, the training is stopped and the trained neural network training model is obtained.

[0144] In some embodiments, the system is configured to:

[0145] Import the training set into the neural network training model and perform iterative calculations. When the preset conditions are met, stop the training and obtain the trained neural network training model, including:

[0146] Determine the error value according to the loss function;

[0147] The gradient of the loss function to the neural network training model is determined according to the error value, and the training of the neural network training model is stopped when the gradient is less than the preset gradient.

[0148] In some embodiments, the system is configured to:

[0149] Determine the gradient of the loss function to the neural network training model according to the error value, and stop training the neural network training model when the gradient is less than the preset gradient, including:

[0150] Determine the gradient of the loss function for the neural network training model based on the back-propagation algorithm.

[0151] The present application proposes a sludge transfer form processing system, which first obtains multiple initial images, then determines multiple target images based on an initial image, where the target image is a partial image in the initial image, then performs text recognition on the target image, and determines multiple feature data based on the recognition result, where the feature data includes signature data, date data, and seal data, and finally classifies and labels the multiple initial images based on the multiple feature data, and classifies the sludge transfer form images corresponding to the initial images according to the classification labels. The present application proposes a sludge transfer form processing system, which improves the matching speed and efficiency of text information in the sludge transfer form, and improves the problem that traditional manual input is prone to errors and is inefficient.

[0152] Based on the same inventive concept, an embodiment of the present application further proposes an electronic device, the electronic device comprising:

[0153] At least one processor; and, a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the sludge transfer form processing method of an embodiment of the present application.

[0154] In addition, to achieve the above-mentioned purpose, an embodiment of the present application also proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the sludge transfer form processing method of the embodiment of the present application is implemented.

[0155] The following is a detailed introduction to the various components of electronic equipment:

[0156] The processor is the control center of the electronic device, which may be a processor or a general term for multiple processing elements. For example, the processor is one or more central processing units (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention, such as one or more microprocessors (DSP), or one or more field programmable gate arrays (FPGA).

[0157] Optionally, the processor can perform various functions of the electronic device by running or executing a software program stored in the memory, and calling data stored in the memory.

[0158] The memory is used to store the software program for executing the solution of the present invention, and the execution is controlled by the processor. The specific implementation method can refer to the above method embodiment and will not be repeated here.

[0159] Optionally, the memory may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory may be integrated with the processor, or may exist independently and be coupled to the processor through an interface circuit of the electronic device, which is not specifically limited in the embodiments of the present invention.

[0160] A transceiver is used to communicate with a network device or a terminal device.

[0161] Optionally, the transceiver may include a receiver and a transmitter, wherein the receiver is used to implement a receiving function, and the transmitter is used to implement a sending function.

[0162] Optionally, the transceiver may be integrated with the processor, or may exist independently and be coupled to the processor via an interface circuit of the router, which is not specifically limited in the embodiment of the present invention.

[0163] In addition, the technical effects of the electronic device can refer to the technical effects of the data transmission method in the above method embodiment, which will not be repeated here.

[0164] It should be understood that the processor in the embodiment of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0165] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0166] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When a computer instruction or computer program is loaded or executed on a computer, a process or function according to an embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0167] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0168] In the present invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can be represented by: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.

[0169] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0170] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

Claims

1. A sludge transfer form processing method, characterized in that: The method comprises: Acquire a plurality of initial images, wherein the plurality of initial images include a plurality of sludge transfer link images; Determine a plurality of target images based on one of the initial images, wherein the target images are partial images of the initial image, wherein each of the target images includes at least one of a signature image, a date image, and a seal image; Dividing the initial image into a plurality of sub-images, wherein the size of the sub-images is much smaller than the target image; classifying the plurality of sub-images based on feature data of each of the sub-images, and determining a plurality of sub-image groups according to the classification results; Separating all the sub-images in each sub-image group into layers, and forming a target layer image with a plurality of the sub-images; Performing noise reduction processing on the target layer image, wherein the noise reduction process includes removing a plurality of discrete points in the target layer image; Performing linear fitting on each of the target layer images so that the multiple sub-images in the target layer image are in a linearly connected state; Acquire a preset font template, and fit a plurality of preset linear font templates in the preset font template with the target layer image, wherein the fitting process includes scaling and twisting the preset linear font template; According to the fitting result, all text information in each of the target layer images is determined, and based on the text information in each of the target layer images, the text information in the initial image is determined; Performing text recognition on the target image, and determining a plurality of feature data according to the recognition result, wherein the feature data includes signature data, date data and seal data; Classify and label the plurality of initial images based on the plurality of feature data, and classify the sludge transfer form images corresponding to the initial images according to the classification labels; Determine all the date data in one of the initial images, arrange the multiple date data in one of the initial images according to date and time, and obtain the earliest date data and the latest date data among the multiple date data; Determine whether the sludge transfer form corresponding to the initial image is compliant based on the earliest date data and the latest date data, mark the compliant initial image as a compliant form, and mark the remaining initial images as non-compliant forms.

2. A sludge transfer joint form processing method according to claim 1, characterized in that: Classifying the plurality of sub-images based on the feature data of each of the sub-images, and determining a plurality of sub-image groups according to the classification results, including: Matching is performed on each of the sub-images based on visual conditions, and characteristic data corresponding to the sub-image is determined according to the matching result, wherein the visual conditions include brightness, contrast and grayscale of each of the sub-images.

3. A sludge transfer form processing method according to claim 2, characterized in that: Classifying the plurality of sub-images based on the feature data of each of the sub-images, and determining a plurality of sub-image groups according to the classification results, including: A neural network model is obtained, feature data of each sub-image is imported into the neural network model, and a plurality of sub-image groups are determined according to output results of the neural network model.

4. A sludge transfer joint form processing method according to claim 3, characterized in that: Acquiring a neural network model, importing feature data of each sub-image into the neural network model, and determining a plurality of sub-image groups according to an output result of the neural network model, including: Acquire a training set, the training set comprising a training image and image feature data corresponding to the training image, wherein the image feature data comprises color data, texture data and shape data; The training set is imported into the neural network training model and iterative calculation is performed. When a preset condition is met, the training is stopped and the trained neural network training model is obtained.

5. A sludge transfer joint form processing method according to claim 4, characterized in that: Importing the training set into the neural network training model and performing iterative calculations, stopping the training when a preset condition is met, and obtaining the trained neural network training model, including: Determine the error value according to the loss function; The gradient of the loss function to the neural network training model is determined according to the error value, and the training of the neural network training model is stopped when the gradient is less than a preset gradient.

6. A sludge transfer joint form processing method according to claim 5, characterized in that: Determining the gradient of the loss function to the neural network training model according to the error value, and stopping the training of the neural network training model when the gradient is less than a preset gradient, including: The gradient of the loss function to the neural network training model is determined based on a back-propagation algorithm.

7. A sludge transfer and processing system, characterized in that: For executing a method as claimed in claim 1, the system is configured to: Acquire a plurality of initial images, wherein the plurality of initial images include a plurality of sludge transfer link images; Determine a plurality of target images based on one of the initial images, wherein the target images are partial images of the initial image, wherein each of the target images includes at least one of a signature image, a date image, and a seal image; Performing text recognition on the target image, and determining a plurality of feature data according to the recognition result, wherein the feature data includes signature data, date data and seal data; The plurality of initial images are classified and labeled based on the plurality of feature data, and the sludge transfer form images corresponding to the initial images are classified according to the classification labels.

8. An electronic device, characterized in that: include: at least one processor; And, a memory communicatively connected to at least one of the processors; wherein the memory stores instructions executable by at least one of the processors, and the instructions are executed by at least one of the processors so that at least one of the processors can execute the method proposed in any one of claims 1 to 6.

Images