An Ancient Book Intelligent Recognition and Restoration Method and Device Based on Machine Learning

Abstract

The present invention provides a method and device for intelligent identification and restoration of ancient books based on machine learning, relating to the technical field of ancient book identification and restoration. By extracting the image features of each area to be identified in the paper image of the ancient book to be restored and inputting them into the intelligent identification model of ancient books for intelligent identification to obtain multiple identification results of each area to be identified, then, based on the multiple identification results, combining with the ancient book database for context semantic analysis and intelligent matching to determine the restoration scheme for the ancient book to be restored, and restoring the ancient book to be restored, the automatic identification of the ancient book to be restored and the automatic generation of the restoration scheme are realized, reducing the professional requirements for restorers, reducing the restoration time of restorers, and improving the identification and restoration efficiency of ancient books.

Description

Technical Field

[0001] The present invention relates to the technical field of ancient book recognition and restoration, and in particular to an intelligent recognition and restoration method and device for ancient books based on machine learning. Background Art

[0002] Ancient books are important resources for academic research and even more important carriers for cultural inheritance and development. However, over time, these precious ancient book documents generally face severe challenges such as aging and damage, and urgently need to be effectively protected and restored. The work of ancient book recognition and restoration has thus become a crucial task, which not only requires a high degree of professionalism but also implies respect for history and cultural inheritance.

[0003] In the traditional field of ancient book recognition and restoration, manual operation has always been dominant. This method relies on the profound professional background and rich practical experience of restorers. Restorers need to possess extensive knowledge in various aspects such as history, literature, art, and even papermaking technology in order to accurately identify the age, material, style, and even the cause of damage of ancient books. On this basis, they also need to determine the best restoration plan by consulting a large number of literature materials and conducting comparative analysis. This process is not only time-consuming and laborious but also greatly tests the individual ability and patience of restorers.

[0004] Over time, the problem of low efficiency in the traditional manual restoration method for ancient books has become increasingly prominent. Facing a large number of ancient book collections with different conditions, it is difficult for only a few experienced restorers to meet the growing restoration needs. Coupled with the characteristics of high precision and strong irreversibility in the ancient book restoration work itself, the entire restoration cycle is further lengthened. The low-efficiency manual restoration method for ancient books not only affects the effective utilization of ancient book resources but also poses a potential threat to cultural inheritance and development. Summary of the Invention

[0005] The present invention provides an intelligent recognition and restoration method and device for ancient books based on machine learning, which can achieve automatic recognition of ancient books, automatically generate restoration plans, reduce the professional requirements of restorers, reduce the restoration time of restorers, and improve the recognition and restoration efficiency of ancient books.

[0006] In a first aspect, the present invention provides a method for intelligent recognition and restoration of ancient books based on machine learning. The method includes: obtaining a paper image of the ancient book to be restored; performing image segmentation on the paper image of the ancient book to be restored to determine multiple regions to be recognized; extracting features from the multiple regions to be recognized to determine the image features of each region to be recognized; the image features include color features, texture features, shape features, and size features; based on the image features of each region to be recognized and an intelligent recognition model of ancient books, determining multiple recognition results for each region to be recognized; based on the multiple recognition results of each region to be recognized and an ancient book database, performing semantic analysis and intelligent matching analysis on the context information of the ancient book to be restored to determine a restoration plan for the ancient book to be restored; and restoring the ancient book to be restored based on the restoration plan of the ancient book to be restored.

[0007] In a possible implementation manner, performing image segmentation on the paper image of the ancient book to be restored to determine multiple regions to be recognized includes: performing edge detection on the paper image to determine multiple edge points in the paper image; based on the multiple edge points, determining multiple text regions in the paper image; performing image segmentation based on the multiple text regions to obtain multiple regions to be recognized; each region to be recognized includes one or more text regions and non-text regions adjacent to the text regions.

[0008] In a possible implementation manner, extracting features from the multiple regions to be recognized to determine the image features of each region to be recognized includes: calculating the color histogram of each region to be recognized, and based on the color histograms of each region to be recognized, determining the color features of each recognized region; determining the gray-level co-occurrence matrix of each region to be recognized, and based on the gray-level co-occurrence matrix of each region to be recognized, calculating the texture features of each region to be recognized, the texture features include contrast, homogeneity, and correlation; based on the edge points of each text region in each region to be recognized, determining the text contour of each text region; based on the text contours of each text region, determining the shape features of each region to be recognized; based on the text contours of each text region in each region to be recognized, calculating the size features of each text; based on the contour of the non-text region of each region to be recognized, determining the size features of the non-text region; based on the size features of each text and the size features of the non-text region, determining the size features of each region to be recognized; and performing feature fusion based on the color features, texture features, shape features, and size features of each region to be recognized to obtain the image features of each region to be recognized.

[0009] In a possible implementation manner, based on the image features of each region to be recognized and the ancient book intelligent recognition model, multiple recognition results of each region to be recognized are determined, including: based on the image features of each region to be recognized and the age recognition module of the ancient book intelligent recognition model, determining the age of the ancient book to be repaired; based on the age of the ancient book to be repaired, screening multiple glyph recognition modules of the ancient book intelligent recognition model to obtain a target glyph recognition module corresponding to the age of the ancient book to be repaired; based on the image features of each region to be recognized and the target glyph recognition module, performing glyph recognition to obtain multiple recognition results of each region to be recognized; wherein, the multiple recognition results are multiple recognition results with probabilities greater than a set value in the recognition results of the region to be recognized.

[0010] In a possible implementation manner, before determining multiple recognition results of each region to be recognized based on the image features of each region to be recognized and the ancient book intelligent recognition model, it further includes: obtaining the image features and text data of multiple ancient books with determined ages and glyphs; generating first training samples based on the image data and age of each ancient book; generating second training samples of each age based on the image features and text data of each ancient book in each age; performing neural network training based on the first training samples to obtain the age recognition module of the ancient book intelligent recognition model; performing neural network training based on the second training samples of each age to obtain the glyph recognition modules of each age.

[0011] In a possible implementation manner, based on the multiple recognition results of each region to be recognized and the ancient book database, semantic analysis and intelligent matching analysis are performed on the context information of the ancient book to be repaired to determine the repair plan of the ancient book to be repaired, including: based on the multiple recognition results of each region to be recognized and the ancient book database, determining the text semantics of each recognition result in each region to be recognized; performing context semantic association based on the text semantics of each recognition result in each region to be recognized to determine multiple semantic plans of the ancient book to be repaired; calculating evaluation indicators of each semantic plan based on the multiple semantic plans of the ancient book to be repaired, and the evaluation indicators include semantic coherence, historical accuracy, logical rationality, and text readability; determining the repair plan of the ancient book to be repaired based on the evaluation indicators of each semantic plan.

[0012] In a possible implementation manner, based on the multiple recognition results of each region to be recognized and the ancient book database, semantic analysis and intelligent matching analysis are performed on the context information of the ancient book to be repaired to determine the repair plan of the ancient book to be repaired, and it further includes: based on the repair plan, initially generating a repaired interface of the ancient book to be repaired and displaying the repaired interface to the repair personnel; receiving the annotation information of the repair personnel on the repaired interface; the annotation information includes text modification, semantic correction, and / or text completion of the repaired interface; based on the annotation information, re-performing semantic analysis and matching analysis to regenerate the repair plan of the ancient book to be repaired.

[0013] In a possible implementation, based on the restoration plan of the ancient book to be restored, the ancient book to be restored is restored, including: determining the background color and text color of the ancient book to be restored based on the age of the ancient book to be restored and the current color of the ancient book to be restored; determining the glyph data of the ancient book to be restored based on the restoration plan of the ancient book to be restored; performing digital rendering based on the glyph data, as well as the background color and text color, to obtain the digital interface of the ancient book to be restored; performing feature extraction based on the digital interface of the ancient book to be restored to obtain the image features of each region to be recognized in the digital interface; determining the restoration method and restoration content of each region to be recognized based on the image features of each region to be recognized in the paper-based image and the image features of each region to be recognized in the digital interface.

[0014] In a possible implementation, the method further includes: obtaining the post-restoration image and pre-restoration image of the manually restored ancient book; determining the restoration plan of the manually restored ancient book based on the pre-restoration image, the ancient book intelligent recognition model, and the ancient book database; comparing based on the restoration plan and the post-restoration image to determine the difference degree between the manual restoration and intelligent recognition in each region of the manually restored ancient book; generating restoration quality inspection information based on the difference degree between the manual restoration and intelligent recognition in each region, where the restoration quality inspection information includes multiple disputed regions where there are disputes in the manual restoration, and the dispute content of each disputed region; presenting the restoration quality inspection information to the restoration personnel.

[0015] In a second aspect, an embodiment of the present invention provides an ancient book intelligent recognition and restoration device based on machine learning. The device includes: a communication module and a processing module. The communication module is used to obtain the paper-based image of the ancient book to be restored. The processing module is used to perform image segmentation on the paper-based image of the ancient book to be restored to determine multiple regions to be recognized; perform feature extraction on the multiple regions to be recognized to determine the image features of each region to be recognized; the image features include color features, texture features, shape features, and size features; determine multiple recognition results of each region to be recognized based on the image features of each region to be recognized and the ancient book intelligent recognition model; perform semantic analysis and intelligent matching analysis on the context information of the ancient book to be restored based on the multiple recognition results of each region to be recognized and the ancient book database to determine the restoration plan of the ancient book to be restored; restore the ancient book to be restored based on the restoration plan of the ancient book to be restored.

[0016] In a third aspect, an embodiment of the present invention provides an electronic device. The electronic device includes a memory and a processor. The memory stores a computer program, and the processor is used to call and run the computer program stored in the memory to execute the steps of the method described in the first aspect and any possible implementation manner in the first aspect.

[0017] Fourthly, an embodiment of the present invention provides a computer-readable storage medium storing a computer program, characterized in that when the computer program is executed by a processor, the steps of the method described in the above first aspect and any possible implementation manner in the first aspect are implemented.

[0018] The present invention provides a method and device for intelligent identification and restoration of ancient books based on machine learning. By extracting the image features of each area to be identified in the paper image of the ancient book to be restored, inputting them into the ancient book intelligent identification model for intelligent identification, and obtaining multiple identification results for each area to be identified. Then, based on the multiple identification results, combined with the ancient book database for context semantic analysis and intelligent matching, the restoration plan for the ancient book to be restored is determined, and the ancient book to be restored is restored, realizing the automatic identification of the ancient book to be restored and the automatic generation of the restoration plan, reducing the professional requirements of restorers, reducing the restoration time of restorers, and improving the identification and restoration efficiency of ancient books. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0020] Figure 1 is a schematic flowchart of a method for intelligent identification and restoration of ancient books based on machine learning provided by an embodiment of the present invention;

[0021] Figure 2 is a schematic structural diagram of a device for intelligent identification and restoration of ancient books based on machine learning provided by an embodiment of the present invention;

[0022] Figure 3 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] In the following description, specific details such as specific system structures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present invention. However, those skilled in the art should clearly understand that the present invention can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present invention.

[0024] In the description of the present invention, unless otherwise specified, " / " means "or". For example, A / B may represent A or B. The "and / or" herein is merely a correlative relationship describing associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, "at least one" and "multiple" mean two or more. The words such as "first" and "second" do not limit the quantity and execution order, and the words such as "first" and "second" do not necessarily limit to be different.

[0025] In the embodiments of the present application, words such as "exemplary" or "for example" are used to give examples, illustrations or explanations. Any embodiment or design solution described as "exemplary" or "for example" in the embodiments of the present application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly speaking, using words such as "exemplary" or "for example" aims to present relevant concepts in a specific way for easy understanding.

[0026] In addition, the terms "comprising" and "having" mentioned in the description of the present application and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or modules is not limited to the listed steps or modules, but optionally further includes other unlisted steps or modules, or optionally further includes other steps or modules inherent to these processes, methods, products or devices.

[0027] To make the objectives, technical solutions and advantages of the present invention clearer, the following will be described through specific embodiments in conjunction with the drawings of the present invention.

[0028] As described in the background art, there is a problem of low efficiency in the current traditional manual restoration method for ancient books.

[0029] To solve the above technical problems, as Figure 1 shown, the embodiments of the present invention provide an intelligent recognition and restoration method for ancient books based on machine learning. The method includes steps S101 - S106.

[0030] S101. Obtain the paper-based image of the ancient book to be restored.

[0031] In some embodiments, the paper-based image of the ancient book to be restored is an image of an ancient book page obtained by a digital device. The paper-based image contains complete information such as text, illustrations, seals, etc. and physical damage features (such as wormholes, stains, creases, etc.).

[0032] Embodiments of the present invention can use a high-resolution scanner (≥600 dpi) or a multispectral imaging system (such as infrared / ultraviolet imaging) to capture ink marks or watermarks that are invisible under visible light. Use non-local means filtering (Non-Local Means) or a deep learning denoising model (such as DnCNN) to remove dust and scanning noise. Enhance the contrast through histogram equalization or the Retinex algorithm to highlight faded text. Use perspective transformation to correct page bending deformation.

[0033] S102. Perform image segmentation on the paper-based image of the ancient book to be restored, and determine multiple regions to be recognized.

[0034] As a possible implementation, step S102 can be specifically implemented as steps S1021 - S1023.

[0035] S1021. Perform edge detection on the paper-based image to determine multiple edge points in the paper-based image.

[0036] In some embodiments, edge detection is a technique in image processing used to identify points with significant brightness changes in an image, that is, edge points. These edge points usually mark the boundaries of different objects or regions in the image.

[0037] Exemplarily, embodiments of the present invention can use an edge detection algorithm, such as the Canny edge detector, to process the paper-based image. The Canny algorithm calculates the magnitude and direction of the image gradient, and then applies non-maximum suppression and double-threshold techniques to detect edges. Adjust the algorithm parameters (such as the standard deviation of the Gaussian filter, the high threshold, and the low threshold) to adapt to the specific characteristics of the ancient book image, such as blur caused by paper aging, stains, and the diversity of handwritten text. Output the edge detection result to obtain a series of edge points, which roughly outline the boundaries of text, patterns, and the background in the image.

[0038] S1022. Based on the multiple edge points, determine multiple text regions in the paper-based image.

[0039] In some embodiments, the text region is the part of the image that contains text. In the ancient book image, the text region usually appears as a series of connected or adjacent edge points, forming a regular or irregular arrangement.

[0040] Exemplarily, embodiments of the present invention can apply morphological operations (such as dilation and erosion) to connect adjacent edge points to form a more coherent outline of the text region. Use contour detection algorithms (such as line detection based on the Hough transform) to identify line structures in the image, which often correspond to the edges of text lines. Filter out possible text regions according to the geometric features of the contours (such as length, width, direction) and context information (such as the spacing between adjacent contours). Manual adjustment or verification of the automatically recognized text regions may be required to ensure accuracy.

[0041] S1023. Perform image segmentation based on multiple text regions to obtain multiple regions to be recognized.

[0042] In some embodiments, each region to be recognized includes one or more text regions and non-text regions adjacent to the text regions.

[0043] In some embodiments, the region to be recognized is a region that is recognized as needing to be analyzed or processed separately during the image segmentation process. In ancient book images, the regions to be recognized may include one or more text regions and non-text regions (such as patterns, decorations, stains, etc.) adjacent to these text regions.

[0044] Exemplarily, embodiments of the present invention can define an extended boundary that includes these regions based on the recognized text regions to include possible non-text context information (such as punctuation marks, decorative lines, etc.). Use region-based image segmentation algorithms (such as region growing, watershed algorithm) or graph-based segmentation algorithms (such as min-cut / max-flow algorithm) to further subdivide the image within the extended boundary to obtain finer regions to be recognized. Consider factors such as the arrangement of text regions (such as horizontal, vertical, inclined), font size, line spacing, etc., and adjust the parameters of the segmentation algorithm to ensure the accuracy and integrity of the regions to be recognized. Output the segmentation result to obtain multiple regions to be recognized, each region including one or more text regions and non-text regions adjacent to these text regions.

[0045] S103. Extract features from multiple regions to be recognized to determine the image features of each region to be recognized.

[0046] In some embodiments, the image features include color features, texture features, shape features, and size features.

[0047] As a possible implementation, step S103 can be specifically implemented as steps S1031 - S1035.

[0048] S1031. Calculate the color histogram of each region to be recognized, and determine the color features of each recognized region based on the color histograms of the regions to be recognized.

[0049] In some embodiments, color histogram: a statistical chart representing the frequency or quantity of different colors appearing in an image. Color feature: an attribute extracted from the color histogram for describing the color distribution of an image.

[0050] Exemplarily, embodiments of the present invention can convert each region to be recognized from the RGB color space to a space more suitable for color feature extraction, such as the HSV or Lab space. In the selected color space, calculate the color histogram of each region to be recognized. The entire color space can be divided into several color intervals (bins), and then the number of pixels in each interval is counted. Normalize the color histogram to eliminate the influence of different region sizes on color features. Extract the main statistics (such as mean, variance, median, etc.) of the color histogram as color features.

[0051] S1032. Determine the gray-level co-occurrence matrix of each region to be recognized, and calculate the texture features of each region to be recognized based on the gray-level co-occurrence matrix of each region to be recognized. The texture features include contrast, homogeneity, and correlation.

[0052] In some embodiments, gray-level co-occurrence matrix (GLCM): a matrix describing the spatial relationship between gray levels in an image. Texture feature: an attribute extracted from the gray-level co-occurrence matrix for describing the texture of an image, including contrast, homogeneity, correlation, etc.

[0053] Exemplarily, embodiments of the present invention can convert each region to be recognized into a grayscale image. Calculate the gray-level co-occurrence matrix of each grayscale image, considering different directions (such as 0 degrees, 45 degrees, 90 degrees, 135 degrees) and distance parameters. Extract texture features such as contrast, homogeneity, and correlation from the gray-level co-occurrence matrix. Contrast reflects the speed of local gray-level changes in the image, homogeneity reflects the degree of consistency of local gray levels in the image, and correlation reflects the linear correlation between gray levels.

[0054] S1033. Determine the text contour of each text region based on the edge points of each text region in each region to be recognized; determine the shape features of each region to be recognized based on the text contours of each text region.

[0055] In some embodiments, text contour: the boundary line of a text region, used to describe the shape of text. Shape feature: an attribute describing the shape of an object in an image.

[0056] Exemplarily, embodiments of the present invention can perform edge detection on text regions in each region to be recognized to obtain edge points of the text. A contour tracking algorithm (such as tracking the contour after Zhang-Suen thinning algorithm) is used to determine the contour of the text region. Based on the contour points, shape features are calculated, such as the perimeter, area, convex hull, bounding rectangle, etc. of the contour. These features can be used to describe the shape and size of the text region.

[0057] S1034. Calculate the size features of each character based on the text contours of the text regions in each region to be recognized; determine the size features of the non-text regions based on the contours of the non-text regions in each region to be recognized; determine the size features of each region to be recognized based on the size features of each character and the size features of the non-text regions.

[0058] In some embodiments, the size feature: an attribute describing the size of an image or an object in the image.

[0059] Exemplarily, embodiments of the present invention can calculate size features such as the average height, width, and aspect ratio of the text for the text regions in each region to be recognized based on the text contours. For non-text regions, size features such as area and perimeter are also calculated based on their contours. By integrating the size features of the text regions and non-text regions, the size features of the entire region to be recognized are determined. This may include the maximum / minimum size, average size, size distribution, etc. of the region.

[0060] S1035. Perform feature fusion based on the color features, texture features, shape features, and size features of each region to be recognized to obtain the image features of each region to be recognized.

[0061] Exemplarily, embodiments of the present invention can fuse the above-mentioned extracted color features, texture features, shape features, and size features. This can be achieved through simple concatenation, that is, connecting all feature vectors into a long vector. Before feature fusion, each feature may need to be normalized to ensure their comparability in the numerical range. The fused feature vector will be used as the image feature of each region to be recognized for subsequent recognition and analysis by machine learning models.

[0062] S104. Determine multiple recognition results of each region to be recognized based on the image features of each region to be recognized and the ancient book intelligent recognition model.

[0063] In some embodiments, the ancient book intelligent recognition model includes a year recognition module and a glyph recognition module. The year recognition module is used to recognize the year of the ancient book. Each year corresponds to a glyph recognition module. The glyph recognition module is used to recognize the estimated text of the corresponding year.

[0064] In some embodiments, the recognition results include a list of text candidates for the area to be recognized, the pronunciation of the text, the meaning of the words, and other content.

[0065] As a possible implementation, step S104 can be specifically implemented as steps S1041 - S1043.

[0066] S1041. Based on the image features of each area to be recognized and the age recognition module of the ancient book intelligent recognition model, determine the age of the ancient book to be repaired.

[0067] Exemplarily, in the embodiments of the present invention, the image features of each area to be recognized can be input into the age recognition module of the ancient book intelligent recognition model. The age recognition module may be trained using classification algorithms (such as support vector machines, random forests, convolutional neural networks, etc.) to identify the characteristic differences of ancient books from different ages. According to the output of the age recognition module and the comprehensive judgment of multiple areas to be recognized, determine the approximate age of the ancient book to be repaired. It may be necessary to consider majority voting, weighted average, or other fusion strategies to improve the accuracy of the judgment. Output the age information of the ancient book to be repaired for subsequent screening by the glyph recognition module.

[0068] S1042. Based on the age of the ancient book to be repaired, screen multiple glyph recognition modules of the ancient book intelligent recognition model to obtain a target glyph recognition module corresponding to the age of the ancient book to be repaired.

[0069] Exemplarily, in the embodiments of the present invention, according to the age information of the ancient book to be repaired, the module that matches or is closest to the target age can be screened from multiple glyph recognition modules of the ancient book intelligent recognition model. The glyph recognition module may be specifically trained for ancient books of different ages or styles to adapt to the changes in the text features of different historical periods. Output the screened target glyph recognition module for subsequent glyph recognition tasks.

[0070] S1043. Based on the image features of each area to be recognized and the target glyph recognition module, perform glyph recognition to obtain multiple recognition results for each area to be recognized.

[0071] Among them, the multiple recognition results are the multiple recognition results with probabilities greater than the set value in the recognition results of the area to be recognized.

[0072] Exemplarily, embodiments of the present invention may input the image features of each region to be recognized into the target glyph recognition module. The target glyph recognition module may be trained using character recognition algorithms (such as optical character recognition OCR, deep learning models, etc.) to accurately recognize the characters in ancient books. According to the output of the target glyph recognition module, multiple recognition results and their corresponding probabilities for each region to be recognized are obtained. Using a set value as the screening criterion, only the recognition results with probabilities greater than the set value are retained. This helps to reduce misrecognition and improve the accuracy of recognition. The screened recognition results are output, and these results can be used for subsequent tasks such as ancient book restoration, content understanding, or digital preservation.

[0073] S105. Based on the multiple recognition results of each region to be recognized and the ancient book database, perform semantic analysis and intelligent matching analysis on the context information of the ancient book to be restored, and determine the restoration plan for the ancient book to be restored.

[0074] In some embodiments, the ancient book database is a database storing a large amount of ancient book information (such as characters, patterns, layouts, historical backgrounds, etc.). By understanding the meaning and context relationship of the text or image content, information similar to or matching the ancient book to be restored is searched in the database.

[0075] As a possible implementation, step S105 may be specifically implemented as steps S1051 - S1054.

[0076] S1051. Based on the multiple recognition results of each region to be recognized and the ancient book database, determine the literal semantics of each recognition result in each region to be recognized.

[0077] Exemplarily, embodiments of the present invention may use OCR (optical character recognition) technology or deep learning models (such as convolutional neural network CNN) to perform character recognition on the ancient book image to obtain preliminary recognition results. Then, in combination with the vocabulary and semantic information in the ancient book database, semantic parsing is performed on each recognition result to determine its corresponding literal semantics.

[0078] S1052. Based on the literal semantics of each recognition result in each region to be recognized, perform context semantic association to determine multiple semantic schemes for the ancient book to be restored.

[0079] Exemplarily, in embodiments of the present invention, after determining the literal semantics of each recognition result, natural language processing technologies (such as semantic role labeling, dependency syntax analysis, etc.) may be used to analyze the semantic relationships between the recognition results and establish context semantic associations. Based on these associations, multiple possible semantic schemes are generated, and each scheme represents a reasonable interpretation of the ancient book content.

[0080] S1053. Based on the multiple semantic schemes of the ancient book to be restored, calculate the evaluation indexes of each semantic scheme.

[0081] In some embodiments, the evaluation metrics include semantic coherence, historical accuracy, logical rationality, and text readability.

[0082] Among them, semantic coherence: Evaluate the logical relationship and coherence between the words within the semantic scheme to ensure smooth and non-contradictory content. Embodiments of the present invention can grade semantic coherence. For example: Excellent: The logical relationship between the contents is tight, the connection is natural, and there are no abrupt places. Good: The logical relationship between the contents is basically clear, but there may be some unnatural connections. General: The logical relationship between the contents is not clear enough, and there are many unnatural or abrupt connections. Poor: The logical relationship between the contents is chaotic and cannot form a coherent narrative.

[0083] Exemplarily, embodiments of the present invention can use text coherence evaluation tools in the field of natural language processing, such as methods based on vector space models, language models, or graph neural networks, to score the coherence of the semantic scheme. These tools evaluate the coherence of the content by analyzing features such as vocabulary, syntax, and semantics in the text.

[0084] Alternatively, embodiments of the present invention can also periodically invite a certain number of experts or people with relevant background knowledge to read and understand the semantic scheme, and then evaluate and score it according to aspects such as the logical relationship between the contents and whether the connection is natural.

[0085] Historical accuracy: Combine the historical background information in the ancient book database to evaluate whether the semantic scheme conforms to the historical background and historical facts of the ancient book. Embodiments of the present invention can grade historical accuracy. For example: Completely accurate: The content is completely consistent with the historical materials without errors. Basically accurate: The content generally conforms to the historical materials, but there may be individual details with errors. There are errors: There are obvious errors between the content and the historical materials, but it does not affect the overall understanding. Seriously in error: The content is seriously inconsistent with the historical materials and misleads the reader's understanding.

[0086] Exemplarily, embodiments of the present invention can compare the content in the semantic scheme with historical materials, relevant literature, or expert interpretations in the ancient book database to evaluate its historical accuracy. Alternatively, embodiments of the present invention can also invite historians or ancient book experts to review the semantic scheme and judge the accuracy of the content based on their professional knowledge and experience.

[0087] Logical rationality: Analyze the logical structure in the semantic scheme to ensure that the content conforms to logical rules and has no unreasonable points. Embodiments of the present invention can classify logical rationality. For example: Completely reasonable: The content has clear logic and no contradictions. Basically reasonable: The content is generally logical, but there may be individual logical problems in details. Having logical problems: The content has obvious logical contradictions or unreasonable points. Logical chaos: The content is severely logically chaotic and cannot form a reasonable narrative.

[0088] Exemplarily, embodiments of the present invention can perform logical reasoning analysis on the content in the semantic scheme to check whether there are logical contradictions or illogical places. Alternatively, embodiments of the present invention can also invite logic experts or personnel with logical reasoning abilities to evaluate the semantic scheme to judge its logical rationality.

[0089] Text readability: Evaluate whether the text expression in the semantic scheme is clear, easy to understand, and conforms to reading habits. Embodiments of the present invention can classify text readability. For example: Very easy to read: The vocabulary is simple and easy to understand, the sentence structure is clear, and it can be understood without effort. Relatively easy to read: The vocabulary and sentence structure are relatively simple, but some attention may be required to fully understand. Having a certain degree of difficulty: The vocabulary and sentence structure are relatively complex, and certain reading ability and attention are required to understand. Difficult to read: The vocabulary is rare and the sentence structure is complex, making it very difficult to understand.

[0090] Exemplarily, embodiments of the present invention can use text readability evaluation tools to score the semantic scheme. These tools usually evaluate the readability of the text based on features such as vocabulary difficulty, sentence length, and grammar complexity. Alternatively, embodiments of the present invention can also invite a certain number of readers to conduct reading tests on the semantic scheme, and then collect their feedback to evaluate its readability.

[0091] S1054. Based on the evaluation indexes of each semantic scheme, determine the restoration scheme for the ancient books to be restored.

[0092] Exemplarily, embodiments of the present invention can, according to the scores of the evaluation indexes of each semantic scheme, comprehensively consider the weights and importance of each index, and select the semantic scheme with the highest score as the final restoration scheme. If the scores of multiple schemes are similar, expert opinions and ancient book restoration experience can be further combined for decision-making.

[0093] Optionally, after step S1054, embodiments of the present invention regenerate the restoration scheme by interacting with the restorers, such as steps S1055 - S1057.

[0094] S1055. Based on the restoration scheme, initially generate the restored interface of the ancient books to be restored and display the restored interface to the restorers.

[0095] In some embodiments, the repaired interface is a visual effect of the repaired ancient book simulated through technical means (such as image processing, text editing, etc.) based on the preliminarily determined repair plan.

[0096] Exemplarily, embodiments of the present invention can combine information such as characters and typesetting in the recognition result with the styles in the ancient book database based on the preliminarily determined repair plan by using image processing software or a text editor to generate the repaired ancient book interface. The generated repaired interface is displayed to the repair personnel through an electronic screen or in a printed manner for their review and annotation.

[0097] S1056. Receive the annotation information of the repair personnel for the repaired interface.

[0098] In some embodiments, the annotation information includes text modification, semantic correction, and / or text completion for the repaired interface.

[0099] Exemplarily, in embodiments of the present invention, when the repair personnel view the repaired interface, they can directly perform operations such as text modification, semantic correction, or text completion on the interface by using annotation tools (such as an electronic pen, a mouse, etc.). The system automatically records the annotation information of the repair personnel and saves it in an editable file format for subsequent processing.

[0100] S1057. Based on the annotation information, perform semantic analysis and matching analysis again, and regenerate the repair plan for the ancient book to be repaired.

[0101] Exemplarily, embodiments of the present invention can input the annotation information of the repair personnel into a machine learning algorithm for processing as new input data. The algorithm re-analyzes the content of the repaired ancient book according to the annotation information, including semantic understanding, context association, logical verification, etc. At the same time, in combination with the information in the ancient book database, it performs matching analysis on the content of the repaired ancient book to verify its historical accuracy, logical rationality, and text readability. According to the results of the re-analysis, the repair plan is corrected and optimized to generate a more accurate repair plan.

[0102] S106. Repair the ancient book to be repaired based on the repair plan for the ancient book to be repaired.

[0103] As a possible implementation manner, step S106 can be specifically implemented as steps S1061 - S1065.

[0104] S1061. Determine the background color and text color of the ancient book to be repaired based on the age of the ancient book to be repaired and the current color of the ancient book to be repaired.

[0105] Exemplarily, embodiments of the present invention can preliminarily determine the possible ranges of the background color and the text color according to the age information of the ancient book to be repaired and referring to the historical color data in the ancient book database. Then, in combination with the current color information of the ancient book to be repaired, the background color and the text color are further accurately determined through a color matching algorithm.

[0106] S1062. Determine the glyph data of the ancient book to be repaired based on the repair plan of the ancient book to be repaired.

[0107] Exemplarily, embodiments of the present invention can retrieve the corresponding glyph data from the ancient book database according to the age and style of the ancient book to be repaired. If there is no directly matching glyph data in the database, machine learning algorithms can be used to perform style transfer or generation on the characters in the ancient book to obtain glyph data that conforms to the style of the ancient book.

[0108] S1063. Perform digital rendering based on the glyph data, as well as the background color and the text color, to obtain the digital interface of the ancient book to be repaired.

[0109] Exemplarily, embodiments of the present invention can use digital rendering technology to combine the determined background color, text color, and glyph data to generate the digital interface of the ancient book to be repaired. During the rendering process, it is necessary to ensure that the contrast between the text and the background is appropriate for subsequent recognition and repair.

[0110] S1064. Perform feature extraction based on the digital interface of the ancient book to be repaired to obtain the image features of each region to be recognized in the digital interface.

[0111] Exemplarily, embodiments of the present invention can use feature extraction algorithms to extract the image features of each region to be recognized from the digital interface. These features can include color features, texture features, shape features, etc. The purpose of feature extraction is to match with the features in the paper-based image subsequently to determine the repair method and repair content.

[0112] S1065. Determine the repair method and repair content of each region to be recognized based on the image features of each region to be recognized in the paper-based image and the image features of each region to be recognized in the digital interface.

[0113] Exemplarily, embodiments of the present invention can match the image features of each region to be recognized in the paper-based image with the corresponding image features in the digital interface. According to the matching result, analyze the specific conditions of the region to be repaired, such as blurred text, missing text, contamination, etc. According to the analysis result, determine the repair method of each region to be recognized, such as digital repair, manual repair, etc., and clarify the specific repair content, such as replacing blurred text, filling in missing text, etc.

[0114] The present invention provides an intelligent recognition and restoration method for ancient books based on machine learning. By extracting the image features of each area to be recognized in the paper-based image of the ancient book to be restored, and inputting them into the intelligent recognition model of ancient books for intelligent recognition, multiple recognition results of each area to be recognized are obtained. Then, based on the multiple recognition results, combined with the ancient book database for context semantic analysis and intelligent matching, the restoration plan for the ancient book to be restored is determined, and the ancient book to be restored is restored, realizing the automatic recognition of the ancient book to be restored and the automatic generation of the restoration plan, reducing the professional requirements of restorers, reducing the restoration time of restorers, and improving the recognition and restoration efficiency of ancient books.

[0115] Optionally, for the intelligent recognition and restoration method for ancient books based on machine learning provided by the embodiments of the present invention, before step S104, steps S201 - S205 are further included.

[0116] S201. Obtain the image features and text data of multiple ancient books with determined ages and glyphs.

[0117] Exemplarily, the embodiments of the present invention can collect ancient book samples with known ages and glyphs, and these samples should cover multiple historical periods to ensure the comprehensiveness of training. High-resolution scanning or photographing is performed on each ancient book sample to obtain its image data. Image processing techniques, such as edge detection and texture analysis, are used to extract the features of the ancient book image. Through OCR (Optical Character Recognition) technology or manual transcription, the text data in the ancient book is obtained.

[0118] S202. Generate the first training sample based on the image data and age of each ancient book.

[0119] In some embodiments, the first training sample is a training dataset composed of the image data of multiple ancient books and the corresponding age information, and is used to train the age recognition module of the intelligent recognition model of ancient books.

[0120] Exemplarily, the embodiments of the present invention can combine the image data of each ancient book and its corresponding age information into a training sample. Ensure that the age distribution of ancient books in the training sample is uniform to avoid the model's preference for a specific age.

[0121] S203. Generate the second training sample for each age based on the image features and text data of each ancient book in each age.

[0122] In some embodiments, the second training sample is a training dataset composed of the image features of ancient books and the corresponding text data, classified by age, and is used to train the glyph recognition module for each age.

[0123] Exemplarily, embodiments of the present invention can generate training samples for ancient book samples within each era based on their image features and text data, ensuring that the number of training samples for each era is sufficient to improve the accuracy of the glyph recognition module.

[0124] S204. Based on the first training samples, conduct neural network training to obtain the era recognition module of the ancient book intelligent recognition model.

[0125] Exemplarily, embodiments of the present invention can select a suitable neural network architecture, such as a convolutional neural network (CNN), for training the era recognition module. Input the first training samples into the neural network, perform forward propagation and backward propagation, and adjust the network parameters to minimize the loss function. Through multiple iterations of training until the network converges, obtain the trained era recognition module.

[0126] S205. Based on the second training samples of each era, conduct neural network training to obtain the glyph recognition modules of each era.

[0127] Exemplarily, embodiments of the present invention can select a suitable neural network architecture, such as a recurrent neural network (RNN) or a convolutional neural network (CNN), for each era to train the glyph recognition module of that era. Input the second training samples corresponding to the era into the neural network, perform forward propagation and backward propagation, and adjust the network parameters to minimize the loss function. Through multiple iterations of training until the glyph recognition modules of each era converge, obtain the trained glyph recognition modules of each era.

[0128] In this way, embodiments of the present invention can pre-train the ancient book intelligent recognition model, achieve intelligent recognition of ancient books, and improve the recognition and restoration efficiency of ancient books. In addition, the present invention trains ancient books of each era separately, effectively distinguishing the glyph characters of each historical period and improving the accuracy of ancient book intelligent recognition.

[0129] Optionally, the method for intelligent recognition and restoration of ancient books based on machine learning provided by embodiments of the present invention further includes steps S301 - S305.

[0130] S301. Obtain the post - repair image and pre - repair image of the manually repaired ancient book.

[0131] Exemplarily, embodiments of the present invention can collect the images before and after manual repair from ancient book repair projects, ensuring that the images are clear, complete, and easy to compare. Pre - process the images, such as denoising and enhancing contrast, to improve the accuracy of subsequent analysis.

[0132] S302. Based on the pre - repair image, the ancient book intelligent recognition model, and the ancient book database, determine the estimated repair plan for manual repair.

[0133] Exemplarily, embodiments of the present invention can use an ancient book intelligent recognition model to analyze the pre - restoration image, identify text content, glyph features, etc. Combining historical information and style features in the ancient book database, one or more possible restoration schemes are generated. These schemes will serve as a benchmark for comparison with the manual restoration results.

[0134] S303. Based on the restoration scheme and the post - restoration image, make a comparison to determine the difference degree between manual restoration and intelligent recognition in each area of the manually restored ancient book.

[0135] In some embodiments, the difference degree is an index measuring the degree of inconsistency between the manual restoration result and the intelligent recognition restoration scheme, usually calculated by comparing the differences between the two in aspects such as text content, glyphs, layout, etc.

[0136] Exemplarily, embodiments of the present invention can compare the generated restoration scheme with the image after manual restoration, focusing on the differences in aspects such as text content, glyphs, layout, etc. Using image processing techniques and machine learning algorithms, quantify these differences and calculate the difference degree index. The difference degree can be obtained by calculating similarity, error rate or other appropriate metrics.

[0137] S304. Generate restoration quality inspection information based on the difference degree between manual restoration and intelligent recognition in each area.

[0138] In some embodiments, the restoration quality inspection information includes multiple controversial areas where there are disputes in manual restoration, and the controversial content of each controversial area.

[0139] In some embodiments, the controversial area is an area with significant differences between manual restoration and intelligent recognition, and these areas may require further review or restoration. The controversial content is the specific differences in text, glyphs, layout, etc. within the controversial area.

[0140] Exemplarily, embodiments of the present invention can identify controversial areas with significant differences according to the difference degree index. Conduct a detailed analysis of each controversial area, extract the controversial content, such as text errors, inconsistent glyphs, chaotic layout, etc. Organize the controversial areas and controversial content into a restoration quality inspection information report for subsequent review and processing.

[0141] S305. Present the restoration quality inspection information to the restoration personnel.

[0142] Exemplarily, embodiments of the present invention can present the restoration quality inspection information to the restoration personnel through an electronic screen, a printed report or an online platform, etc. Ensure that the information is clear and intuitive, facilitating the restoration personnel to understand and process the controversial areas. Provide necessary explanations and guidance to help the restoration personnel understand the controversial content and its possible reasons.

[0143] In this way, the embodiments of the present invention can verify and quality-check the ancient books repaired manually through the means of intelligent recognition and semantic analysis of ancient books, and present the controversial areas and response contents to the repair personnel for browsing and final determination by the repair personnel, thereby improving the accuracy of the recognition and repair of ancient books.

[0144] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The order of execution of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0145] The following are the device embodiments of the present invention. For the details not described in detail, reference can be made to the corresponding method embodiments above.

[0146] Figure 2 The structural schematic diagram of an ancient book intelligent recognition and repair device based on machine learning provided by the embodiments of the present invention is shown. The repair device 400 includes a communication module 401 and a processing module 402.

[0147] The communication module 401 is used to obtain the paper image of the ancient book to be repaired.

[0148] The processing module 402 is used to perform image segmentation on the paper image of the ancient book to be repaired to determine multiple regions to be recognized; extract features from the multiple regions to be recognized to determine the image features of each region to be recognized; the image features include color features, texture features, shape features and size features; based on the image features of each region to be recognized and the ancient book intelligent recognition model, determine multiple recognition results for each region to be recognized; based on the multiple recognition results of each region to be recognized and the ancient book database, perform semantic analysis and intelligent matching analysis on the context information of the ancient book to be repaired to determine the repair plan for the ancient book to be repaired; repair the ancient book to be repaired based on the repair plan of the ancient book to be repaired. [[ID=]18]

[0149] Figure 3 It is the structural schematic diagram of an electronic device provided by the embodiments of the present invention. As Figure 3 shown, the electronic device 500 includes: a processor 501, a memory 502, and a computer program 503 stored in the memory 502 and executable on the processor 501. When the processor 501 executes the computer program 503, the steps in the above method embodiments are implemented, such as Figure 1 the steps S101 - S106 shown. Or, when the processor 501 executes the computer program 503, the functions of each module / unit in the above device embodiments are implemented, for example, Figure 2 the functions of the communication module 401 and the processing module 402 shown.

[0150] Exemplarily, the computer program 503 can be divided into one or more modules / units, which are stored in the memory 502 and executed by the processor 501 to implement the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program 503 in the electronic device 500. For example, the computer program 503 can be divided into Figure 3 the communication module 401 and the processing module 402 as shown. Figure 2

[0151] The so-called processor 501 can be a central processing unit (CPU), or can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0152] The memory 502 can be an internal storage unit of the electronic device 500, such as the hard disk or memory of the electronic device �00. The memory 502 can also be an external storage device of the electronic device 500, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. equipped on the electronic device 500. Further, the memory 502 can also include both the internal storage unit and the external storage device of the electronic device 500. The memory 502 is used to store the computer program and other programs and data required by the terminal. The memory 502 can also be used to temporarily store the data that has been output or will be output.

[0153] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.​

Claims

1. A method for intelligent identification and restoration of ancient books based on machine learning, characterized in that: include: Obtain paper images of ancient books to be restored; Performing image segmentation on the paper image of the ancient book to be restored to determine a plurality of areas to be identified; Extracting features from the plurality of regions to be identified to determine image features of each region to be identified; the image features include color features, texture features, shape features, and size features; Determine multiple recognition results for each area to be recognized based on the image features of each area to be recognized and the ancient book intelligent recognition model; Based on the multiple recognition results of each area to be identified and the ancient book database, semantic analysis and intelligent matching analysis are performed on the context information of the ancient book to be repaired to determine the restoration plan for the ancient book to be repaired; Repairing the ancient books to be restored based on the restoration plan of the ancient books to be restored; The method of determining multiple recognition results for each area to be recognized based on the image features of each area to be recognized and the ancient book intelligent recognition model includes: determining the age of the ancient book to be restored based on the image features of each area to be recognized and the age recognition module of the ancient book intelligent recognition model; screening multiple glyph recognition modules of the ancient book intelligent recognition model based on the age of the ancient book to be restored to obtain a target glyph recognition module corresponding to the age of the ancient book to be restored; performing glyph recognition based on the image features of each area to be recognized and the target glyph recognition module to obtain multiple recognition results for each area to be recognized; wherein, the multiple recognition results are multiple recognition results whose probabilities are greater than a set value among the recognition results of the area to be recognized.

2. The method for intelligent identification and restoration of ancient books based on machine learning according to claim 1 is characterized in that: The step of segmenting the paper image of the ancient book to be restored to determine a plurality of areas to be identified includes: performing edge detection on the paper image to determine a plurality of edge points in the paper image; determining a plurality of text areas in the paper image based on the plurality of edge points; Image segmentation is performed based on the multiple text areas to obtain multiple areas to be identified; each area to be identified includes one or more text areas, and non-text areas adjacent to the text areas.

3. The ancient book intelligent identification and restoration method based on machine learning according to claim 1 is characterized in that: The extracting features of the plurality of to-be-identified regions to determine image features of each to-be-identified region includes: Calculating the color histogram of each area to be identified, and determining the color characteristics of each identification area based on the color histogram of each area to be identified; Determine a gray level co-occurrence matrix of each region to be identified, and calculate texture features of each region to be identified based on the gray level co-occurrence matrix of each region to be identified, wherein the texture features include contrast, homogeneity, and correlation; Determine the outline of each character region based on the edge points of each character region in each to-be-recognized region; determine the shape features of each to-be-recognized region based on the character outline of each character region; Calculating the size characteristics of each character based on the outline of each character area in each area to be recognized; determining the size characteristics of the non-character area based on the outline of the non-character area in each area to be recognized; determining the size characteristics of each area to be recognized based on the size characteristics of each character and the size characteristics of the non-character area; Based on the color features, texture features, shape features and size features of each area to be identified, feature fusion is performed to obtain the image features of each area to be identified.

4. The ancient book intelligent identification and restoration method based on machine learning according to claim 1 is characterized in that: Before determining multiple recognition results for each area to be recognized based on the image features of each area to be recognized and the ancient book intelligent recognition model, the method further includes: Obtain image features and text data of multiple ancient books with confirmed ages and font shapes; Generate a first training sample based on the image data and age of each ancient book; Based on the image features and text data of each ancient book in each era, generate the second training samples of each era; Performing neural network training based on the first training sample to obtain an age recognition module of the ancient book intelligent recognition model; Based on the second training samples of each era, neural network training is performed to obtain the glyph recognition modules of each era.

5. The ancient book intelligent identification and restoration method based on machine learning according to claim 1 is characterized in that: Based on the multiple recognition results of each area to be recognized and the ancient book database, semantic analysis and intelligent matching analysis are performed on the context information of the ancient book to be restored to determine the restoration plan of the ancient book to be restored, including: Determine the textual semantics of each recognition result in each area to be recognized based on the multiple recognition results in each area to be recognized and the ancient book database; Based on the text semantics of each recognition result in each area to be recognized, contextual semantic association is performed to determine multiple semantic schemes for the ancient books to be restored; Calculating evaluation indicators for each semantic scheme based on multiple semantic schemes of the ancient book to be restored, wherein the evaluation indicators include semantic coherence, historical accuracy, logical rationality, and text readability; Based on the evaluation indicators of each semantic scheme, the restoration plan for the ancient books to be restored is determined.

6. The ancient book intelligent identification and restoration method based on machine learning according to claim 5 is characterized in that: The method further includes: performing semantic analysis and intelligent matching analysis on the context information of the ancient book to be repaired based on the multiple recognition results of each area to be recognized and the ancient book database to determine the repair plan for the ancient book to be repaired; Based on the restoration plan, a restoration interface of the ancient book to be restored is preliminarily generated, and the restoration interface is displayed to the restoration personnel; Receiving annotation information on the repaired interface from the repairer; the annotation information includes text modification, semantic correction and / or text completion of the repaired interface; Based on the annotation information, semantic analysis and matching analysis are re-performed to regenerate a repair plan for the ancient book to be repaired.

7. The ancient book intelligent identification and restoration method based on machine learning according to claim 1 is characterized in that: The restoration plan based on the ancient book to be restored, and the restoration of the ancient book to be restored, includes: Determining the background color and text color of the ancient book to be restored based on the age of the ancient book to be restored and the current color of the ancient book to be restored; Determining the glyph data of the ancient book to be restored based on the restoration plan of the ancient book to be restored; Performing digital rendering based on the glyph data, background color, and text color to obtain a digital interface of the ancient book to be restored; Based on the digitized interface of the ancient book to be restored, feature extraction is performed to obtain image features of each area to be identified in the digitized interface; Based on the image features of each area to be identified in the paper image and the image features of each area to be identified in the digitized interface, the restoration method and restoration content of each area to be identified are determined.

8. The method for intelligent identification and restoration of ancient books based on machine learning according to any one of claims 1 to 7, characterized in that: The method further comprises: Obtaining images of manually restored ancient books after restoration and before restoration; Determining a restoration plan based on the pre-restoration image, the ancient book intelligent recognition model, and the ancient book database; Based on the restoration scheme and the restored image, a comparison is performed to determine the difference between the manual restoration and the intelligent recognition in each area of the manually restored ancient book; Generating restoration quality inspection information based on the difference between manual restoration and intelligent recognition in each area, wherein the restoration quality inspection information includes multiple disputed areas where manual restoration is disputed, and the disputed content of each disputed area; The restoration quality inspection information is presented to the restoration personnel.

9. An intelligent identification and restoration device for ancient books based on machine learning, characterized in that: include: A communication module, used to obtain paper images of the ancient books to be restored; a processing module, configured to segment the paper image of the ancient book to be restored to determine a plurality of regions to be identified; perform feature extraction on the plurality of regions to be identified to determine image features of each region to be identified; the image features include color features, texture features, shape features, and size features; Based on the image features of each area to be identified and the ancient book intelligent recognition model, multiple recognition results for each area to be identified are determined; based on the multiple recognition results for each area to be identified and the ancient book database, semantic analysis and intelligent matching analysis are performed on the context information of the ancient book to be repaired to determine a repair plan for the ancient book to be repaired; based on the repair plan for the ancient book to be repaired, the ancient book to be repaired is repaired; The processing module is specifically used to determine the age of the ancient book to be repaired based on the image features of each area to be recognized and the age recognition module of the ancient book intelligent recognition model; based on the age of the ancient book to be repaired, screen multiple glyph recognition modules of the ancient book intelligent recognition model to obtain a target glyph recognition module corresponding to the age of the ancient book to be repaired; perform glyph recognition based on the image features of each area to be recognized and the target glyph recognition module to obtain multiple recognition results for each area to be recognized; wherein, the multiple recognition results are multiple recognition results whose probabilities are greater than a set value among the recognition results of the area to be recognized.

Images