Data acquisition method and device, electronic equipment and storage medium

Abstract

The invention provides a data acquisition method and device, electronic equipment and a storage medium, and relates to the technical field of data processing, entity information of a to-be-acquired book is acquired, the entity information is coded according to a preset hierarchical coding rule, and a unique identification code corresponding to the to-be-acquired book is generated; based on a plurality of preset acquisition dimensions and the parameter range of each dimension, generating an acquisition parameter combination, and generating an acquisition task according to the unique identification code and the acquisition parameter combination, the acquisition task including the unique identification code of the target book and the corresponding acquisition parameter combination; and executing an acquisition task to obtain a book image, combining the unique identification code and the acquisition parameters as metadata in the acquisition process, and performing associated storage on the metadata and the acquired image to construct a book image data set. And the generalization ability of the image processing model trained based on the data set is obviously improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a data acquisition method, apparatus, electronic device, and storage medium. Background Technology

[0002] In the field of image data acquisition technology, especially in the area of ​​data acquisition for book image recognition applications, photo recognition technology has been widely used as a key auxiliary tool due to the rapid iteration of smart product functions. Book image data acquisition needs to meet the requirements of various application scenarios such as textbook recognition, picture book recognition, and extracurricular book recognition. These scenarios place high demands on the quality, diversity, and traceability of the acquired data.

[0003] Current technologies for collecting book image data typically rely on manual experience for scene setup and shooting. Collectors set up the shooting environment based on their personal judgment, randomly photographing different editions of books. After collection, the images are then manually categorized, organized, and stored. This traditional method attempts to cover various usage scenarios by increasing the number of images collected.

[0004] However, the aforementioned existing technologies are inefficient in collecting data from a large number of physical books such as textbooks and picture books with complex versions, and are prone to data corruption. Summary of the Invention

[0005] This invention provides a data acquisition method, device, electronic device, and storage medium to solve the problems of low acquisition efficiency and easy data disorder in the prior art when dealing with a large number of physical books such as textbooks and picture books with complex versions.

[0006] This invention provides a data acquisition method, comprising: Obtain the entity information of the book to be collected, encode the entity information according to the preset hierarchical coding rules, and generate a unique identifier code corresponding to the book to be collected; Based on multiple preset acquisition dimensions and the parameter range of each dimension, a combination of acquisition parameters is generated. The acquisition dimensions include shooting distance, book posture, lighting conditions, background complexity, line of sight offset angle, book offset distance, and book rotation angle. A collection task is generated based on the unique identifier and the collection parameter combination. The collection task includes the unique identifier of the target book and the corresponding collection parameter combination. The acquisition task is executed to obtain book images. During the acquisition process, the unique identifier and acquisition parameters are combined as metadata and stored in association with the acquired images to construct a book image dataset.

[0007] According to a data acquisition method provided by the present invention, the hierarchical coding rules include: The books to be collected are classified according to a hierarchical structure of type, subject, grade level, publisher, and serial number; Each level is assigned a corresponding coding segment, and the coding segments of each level are combined to form a complete unique identifier. The unique identifier is used to identify and track specific books during the collection, storage, and retrieval process.

[0008] According to a data acquisition method provided by the present invention, the step of generating a combination of acquisition parameters based on a preset plurality of acquisition dimensions and a parameter range for each dimension specifically includes: The shooting distance is set to a first parameter value of close distance, medium distance and long distance; the book posture is set to a second parameter value of flat state, natural curved state and obvious curled state; the lighting conditions are set to a third parameter value of normal lighting, strong lighting and weak lighting; and the background complexity is set to a fourth parameter value of simple background, cluttered background and occluded background. Based on the first parameter value, the second parameter value, the third parameter value, and the fourth parameter value, a discrete parameter set is obtained. By applying orthogonal experimental design strategies or random sampling strategies, parameter values ​​are extracted from the discrete parameter set and combined operations are performed to generate combinations of collected parameters covering different environmental scenarios.

[0009] According to a data acquisition method provided by the present invention, the method further includes: Configure the range of parameter variations for spatial geometric dimensions, wherein the spatial geometric dimensions include at least the line-of-sight offset angle, the book offset distance, and the book rotation angle; The line-of-sight offset angle is set as the offset of the shooting line of sight relative to the normal of the book center, the book offset distance is set as the displacement of the book center relative to the image center, and the book rotation angle is set as the rotation of the book relative to a preset reference direction. The parameters of the spatial geometric dimension are superimposed and combined with the acquisition parameters to generate the final acquisition parameter combination that simulates real shooting behavior.

[0010] According to a data acquisition method provided by the present invention, the step of performing the acquisition task to obtain book images includes: Adjust the shooting environment and equipment settings according to the combination of acquisition parameters in the acquisition task; For the same page of the same book, take multiple photos with different parameter combinations to collect positive sample images and rejection set images; Among them, the positive sample images are valid images that meet the recognition requirements, and the rejection set of images are invalid images that contain blurry, heavily occluded, or no book content; While acquiring images, the system automatically records and associates the current combination of acquisition parameters with the book's unique identifier.

[0011] According to a data acquisition method provided by the present invention, the unique identifier and acquisition parameters are combined as metadata and associated with the acquired image for storage, including: Embed task identifier, book unique identifier, and acquisition parameter combination information in the metadata field of the image file; Create an index file to record the mapping relationship between each image and its corresponding unique book identifier and acquisition parameter combination; The associated storage of the metadata supports subsequent data tracing, filtering, and analysis.

[0012] According to a data acquisition method provided by the present invention, the method further includes a step of performing quality detection on the acquired images: Image processing algorithms are used to automatically detect the blur, exposure, and integrity of images. For positive sample images, the recognizability of the book content and the compliance of the acquisition parameters are verified. For rejected images, it is confirmed that they contain the expected interference features. Images that fail the quality check are removed or marked as samples to be resampled.

[0013] The present invention also provides a data acquisition device, comprising the following modules: The first generation module is used to obtain the entity information of the book to be collected, encode the entity information according to the preset hierarchical encoding rules, and generate a unique identifier code corresponding to the book to be collected. The second generation module is used to generate a combination of acquisition parameters based on multiple preset acquisition dimensions and the parameter range of each dimension. The acquisition dimensions include shooting distance, book posture, lighting conditions, background complexity, line of sight offset angle, book offset distance, and book rotation angle. The third generation module is used to generate a collection task based on the unique identifier and the collection parameter combination. The collection task includes the unique identifier of the target book and the corresponding collection parameter combination. The acquisition module is used to perform the acquisition task to obtain book images. During the acquisition process, the unique identifier and acquisition parameters are combined as metadata and stored in association with the acquired images to construct a book image dataset.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the data acquisition method as described above.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the data acquisition method as described above.

[0016] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements any of the data acquisition methods described above.

[0017] The data acquisition method, device, electronic device, and storage medium provided by this invention achieve standardized management and precise tracking of the book entity information to be acquired by generating unique identifiers based on hierarchical coding rules, effectively solving the problem of chaotic management of multiple versions of books. By generating acquisition parameter combinations that comprehensively cover real-world scenarios based on preset parameters such as shooting distance, book posture, lighting conditions, background complexity, and spatial geometric dimensions, the constructed dataset has high diversity and real-world fit, thereby significantly improving the generalization ability and robustness of the image processing model trained on the dataset. At the same time, through the acquisition task-driven and metadata automatic association storage mechanism, a strong binding between image data and acquisition parameters and book identity is achieved, greatly improving the standardization and automation level of data acquisition, and providing a solid foundation for subsequent data traceability, screening, and the construction of high-quality datasets. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the data acquisition method provided by the present invention; Figure 2 A schematic diagram of the encoding rules provided by this invention; Figure 3 A schematic diagram of book image data acquisition provided for this application; Figure 4 This is a schematic diagram illustrating the required shooting distance for the present invention. Figure 5 A schematic diagram illustrating the book's posture requirements provided for this invention; Figure 6A schematic diagram illustrating the lighting requirements provided by this invention; Figure 7 A schematic diagram illustrating the book background complexity requirements provided for this invention; Figure 8 Specific implementation diagrams provided for this invention; Figure 9 A schematic diagram illustrating the required center line of sight offset angle for this invention; Figure 10 A schematic diagram illustrating the required offset distance between the camera and the center point of the book, provided by this invention. Figure 11 A schematic diagram of the book rotation angle provided by the present invention; Figure 12 This is a schematic diagram of the data acquisition device structure provided by the present invention; Figure 13 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0021] Figure 1 This is a flowchart illustrating the data acquisition method provided by the present invention, as shown below. Figure 1 As shown, the method includes the following: Step 110: Obtain the entity information of the book to be collected, encode the entity information according to the preset hierarchical coding rules, and generate a unique identifier code corresponding to the book to be collected. In this application, the entity information of the books to be collected refers to the basic attribute information that can describe and distinguish different books.

[0022] This entity information can be obtained in various ways, including but not limited to: manually entering basic book information, automatically obtaining information by scanning the book's ISBN code, retrieving information from the publisher's database, and automatically extracting text information from the book cover using image recognition technology. The content of entity information is quite broad, and can include multi-dimensional information such as the book's name, author, publisher, publication year, edition, subject category, applicable grade level, book type (such as textbook, supplementary teaching material, picture book, extracurricular reading material, etc.), number of pages, and book size.

[0023] Hierarchical coding rules are a system of rules for structuring and encoding information in books. These rules classify and code books according to a hierarchical relationship from broad categories to specific subcategories. Figure 2 This is a schematic diagram of the encoding rules provided by the present invention, such as... Figure 2 As shown, in this embodiment, the hierarchical coding rule adopts a five-digit coding system: the first digit indicates the type of book (e.g., 1 for textbooks, 2 for supplementary teaching materials, 3 for picture books, 4 for extracurricular reading materials, etc.); the second digit indicates the subject (e.g., 01 for English, 02 for Chinese, 03 for mathematics, etc.); the third digit indicates the grade level (e.g., 01 for primary school, 02 for junior high school, 03 for senior high school, etc.); the fourth digit indicates the publisher (e.g., 01 for People's Education Press, 02 for Foreign Language Teaching and Research Press, etc.); and the fifth digit is the book serial number, used to distinguish different books within the same category.

[0024] First, the obtained book entity information is parsed to extract key attributes. Then, according to a pre-defined mapping table, each attribute is mapped to its corresponding coded value. Finally, the coded values ​​at each level are combined in a prescribed order to form a complete and unique identifier. For example, the unique identifier of a primary school Chinese textbook published by the People's Education Press might be 10201001, where 1 represents textbook, 02 represents Chinese, 01 represents primary school, 01 represents the People's Education Press, and 001 is the serial number of this type of book.

[0025] In this application, the generation of a unique identifier ensures that each book has a unique identity throughout the entire data acquisition and management system. This not only facilitates the physical management and scheduling of books, but more importantly, it enables accurate tracking and association of each image with its corresponding book information during subsequent image acquisition, storage, retrieval, and analysis, avoiding data confusion and incorrect associations.

[0026] Step 120: Based on multiple preset acquisition dimensions and the parameter range of each dimension, generate a combination of acquisition parameters, wherein the acquisition dimensions include shooting distance, book posture, lighting conditions, background complexity, line of sight offset angle, book offset distance and book rotation angle. In this application, the acquisition dimension refers to the various environmental and conditional factors that affect the acquisition effect of book images. Figure 3 The illustration of book image data acquisition provided in this application is as follows: Figure 3 As shown, this embodiment sets seven main acquisition dimensions: shooting distance, book posture, lighting conditions, background complexity, line-of-sight offset angle, book offset distance, and book rotation angle. Each dimension represents a type of variable factor in a real-world usage scenario.

[0027] The shooting distance dimension simulates different distance relationships between the user and the book. Figure 4 This is a schematic diagram illustrating the shooting distance requirements provided by the present invention, such as... Figure 4 As shown, the parameter range includes close distance (15-25 cm), medium distance (25-40 cm), and long distance (40-60 cm). Different shooting distances will affect the sharpness, completeness, and proportion of the book content in the image.

[0028] The book's posture dimension reflects changes in its physical form. Figure 5 A schematic diagram illustrating the book posture requirements provided by this invention, such as... Figure 5 As shown, the parameter range includes a flat state (pages are completely flat and without bending), a naturally curved state (slight bending at the spine or natural curling at the edges of the pages), and a noticeably curled state (pages have obvious wrinkles, curled edges, or bulges). This dimension takes into account the actual shape of the book under different usage conditions.

[0029] The lighting conditions dimension simulates different ambient lighting conditions. Figure 6 A schematic diagram illustrating the lighting requirements provided by this invention, such as... Figure 6 As shown, the parameter range includes normal lighting (normal indoor lighting, illuminance of about 300-500 lux), strong lighting (direct strong light or close-range illumination from a desk lamp, illuminance exceeding 1000 lux), and weak lighting (dim environment or insufficient lighting at night, illuminance below 100 lux).

[0030] The background complexity dimension takes into account the interference factors in the surrounding environment of the book. Figure 7 A schematic diagram illustrating the book background complexity requirements provided by this invention, such as... Figure 7 As shown, the parameter range includes simple backgrounds (solid color desktop or simple texture background), cluttered backgrounds (with stationery, water cups, etc. scattered), and backgrounds with obstructions (fingers, pens, etc. partially obscure the content of the book).

[0031] For each acquisition dimension, the system sets specific parameter values ​​or parameter ranges based on its characteristics. The generation process of acquisition parameter combinations adopts a scientific experimental design method, which can be an orthogonal experimental design to ensure that the fewest combinations cover the most parameter interaction situations; or a random sampling strategy to randomly select representative combinations from the parameter space. Each generated parameter combination is a specific instance of the parameter value of each dimension, such as: [close distance, natural curvature, normal lighting, cluttered background, 15-degree line-of-sight offset, 5 cm left offset, 10-degree rotation].

[0032] Step 130: Generate a collection task based on the unique identifier and the collection parameter combination. The collection task includes the unique identifier of the target book and the corresponding collection parameter combination. In this application, the collection task is to associate a specific book with specific collection conditions to form a complete collection instruction.

[0033] The process of generating a data collection task includes: first, selecting the books to be collected from the book library and obtaining their unique identifiers; then, selecting appropriate parameter combinations from the pre-generated set of data collection parameter combinations; and finally, encapsulating the unique identifier, parameter combinations, and other necessary information, such as collection time, operator, and data collection device model, into a structured task object.

[0034] Each data collection task can be represented as a data structure, including but not limited to: Task ID (an automatically generated unique task identifier), a unique book identifier, a combination of collection parameters (containing specific parameter values ​​across seven dimensions), task status (pending execution, in progress, completed, etc.), planned collection quantity, actual collection quantity, and quality check results. This structured task design facilitates batch task management, progress tracking, and quality control.

[0035] Step 140: Execute the acquisition task to obtain book images. During the acquisition process, the unique identifier and acquisition parameters are combined as metadata and stored in association with the acquired images to construct a book image dataset.

[0036] During the task execution phase, operators or automated data acquisition equipment make corresponding adjustments and settings to the data acquisition environment and equipment according to the parameter combinations specified in the task.

[0037] Figure 8 The specific implementation diagrams provided for this invention are as follows: Figure 8 As shown, firstly, the physical distance between the acquisition device and the book is adjusted according to the shooting distance parameter; the book is adjusted to the appropriate shape according to the book posture parameter; the ambient light source or supplementary lighting device is adjusted according to the lighting condition parameter; the appropriate background environment is arranged according to the background complexity parameter; and the shooting angle of the acquisition device and the position of the book are adjusted according to the spatial geometry parameter.

[0038] During image acquisition, the system automatically triggers a capture to obtain images of the book. These book images can be either single-frame still images or continuous video sequences. The acquired image types include positive sample images and rejection set images.

[0039] Metadata describes image data, recording important attributes such as the image's source, acquisition conditions, and associated information. In this embodiment, the metadata includes: the book's unique identifier, seven parameter values ​​at the time of acquisition, acquisition timestamp, device information, and operator information.

[0040] Metadata can be associated and stored in several ways: it can be directly embedded into the EXIF ​​field of an image file; key information can be reflected in the file name; an independent metadata database can be established and associated with the image file through an index; or a matching XML or JSON format metadata file can be generated.

[0041] The construction of the book image dataset is a systematic process. All collected images are organized and stored according to a predetermined directory structure, such as a hierarchical structure of dataset root directory / book category / publisher / grade / subject / book ID / collection batch / . Simultaneously, the system generates statistical information for the dataset, including the total number of images, the distribution of parameters in each dimension, and book coverage, facilitating subsequent data analysis and model training.

[0042] In this application, a unique identifier code is generated based on hierarchical coding rules, achieving standardized management and precise tracking of the information of the book entities to be collected, effectively solving the problem of chaotic management of multiple versions of books. By generating a combination of collection parameters that comprehensively covers real-world scenarios based on preset parameters such as shooting distance, book posture, lighting conditions, background complexity, and spatial geometric dimensions, the constructed dataset has high diversity and real-world fit, thereby significantly improving the generalization ability and robustness of the image processing model trained on this dataset. At the same time, through a collection task-driven and metadata automatic association and storage mechanism, a strong binding between image data and collection parameters and book identity is achieved, greatly improving the standardization and automation level of data collection, and providing a solid foundation for subsequent data traceability, screening, and the construction of high-quality datasets.

[0043] Optionally, the hierarchical coding rules include: The books to be collected are classified according to a hierarchical structure of type, subject, grade level, publisher, and serial number; Each level is assigned a corresponding coding segment, and the coding segments of each level are combined to form a complete unique identifier. The unique identifier is used to identify and track specific books during the collection, storage, and retrieval process.

[0044] In this application, the hierarchical coding rules adopt a multi-level classification system to systematically classify and manage massive amounts of book resources.

[0045] The first level is category classification, which is the highest level of classification. Categories are divided based on the book's purpose and nature. Typical categories include: textbooks (textbooks used in formal school education), supplementary teaching materials (supplementary materials to complement textbooks), picture books (children's books primarily featuring illustrations), extracurricular reading materials (science, literature, and other extended reading materials), and reference books (dictionaries, encyclopedias, and other reference books). Each category is assigned a unique code; for example, textbooks are coded as 1, and supplementary teaching materials are coded as 2.

[0046] The second level is subject classification, which is further subdivided under the category level. For textbooks and supplementary teaching materials, subjects include Chinese, mathematics, English, physics, chemistry, biology, history, geography, and politics; for picture books and extracurricular reading materials, they can be classified according to content themes, such as science, literature, and art. Subject codes use two digits, such as Chinese 01, mathematics 02, and English 03.

[0047] The third level is the educational stage classification, indicating the educational stage to which the book is applicable. The educational stages include: preschool education, primary school (grades 1-6), junior high school (grades 7-9), senior high school (grades 10-12), and university. In actual coding, this can be further refined to specific grades, such as 011 for first grade and 012 for second grade.

[0048] The fourth level is the publisher classification, recording the publishing unit of the book. Major textbook publishers in China include People's Education Press, Beijing Normal University Press, and Jiangsu Education Press, each assigned a unique code. This level is particularly important for distinguishing different versions of textbooks.

[0049] The fifth level is the serial number, which is a sequential number determined after the first four levels and used to distinguish different books within the same category. The serial number starts from 001 and increments, and the number of digits can be expanded according to actual needs.

[0050] The allocation of coded segments follows a fixed length principle to ensure the standardization and readability of the encoding. A complete unique identifier is formed by sequentially combining coded segments at each level, resulting in an identifier in the format 1-02-011-03-015, where hyphens can be omitted and the segments are directly concatenated as 10201103015.

[0051] This hierarchical coding system has good scalability, allowing new classification levels to be added or new categories to be added under existing levels as needed; at the same time, the coding has semantic information, allowing readers to quickly understand the basic attributes of a book through the coding; finally, it facilitates batch management and retrieval, allowing readers to quickly filter books of specific categories by using the coding prefix.

[0052] In the actual collection, storage, and retrieval process, the unique identifier plays a crucial role. During the collection phase, operators scan or input the unique identifier, and the system automatically retrieves the complete information of the book, avoiding manual input errors. During the storage phase, all image data related to the book is linked through the unique identifier, forming a complete data relationship network. During the retrieval phase, the unique identifier can be used to quickly locate all relevant data for a specific book, and batch retrieval can be performed based on the hierarchical features of the encoding, such as retrieving image data from all elementary school math textbooks.

[0053] In this application, by implementing this hierarchical coding rule, the problem of chaotic management of multiple versions of textbooks is effectively solved, and intelligent and standardized management of book resources is realized, providing a reliable foundation for large-scale data collection.

[0054] Optionally, the step of generating a combination of acquisition parameters based on multiple preset acquisition dimensions and the parameter range of each dimension specifically includes: The shooting distance is set to a first parameter value of close distance, medium distance and long distance; the book posture is set to a second parameter value of flat state, natural curved state and obvious curled state; the lighting conditions are set to a third parameter value of normal lighting, strong lighting and weak lighting; and the background complexity is set to a fourth parameter value of simple background, cluttered background and occluded background. Based on the first parameter value, the second parameter value, the third parameter value, and the fourth parameter value, a discrete parameter set is obtained. By applying orthogonal experimental design strategies or random sampling strategies, parameter values ​​are extracted from the discrete parameter set and combined operations are performed to generate combinations of collected parameters covering different environmental scenarios.

[0055] In this application, the first parameter value of the shooting distance is set with reference to typical scenarios in actual use by users.

[0056] Close distance (15-25 cm) simulates the shooting behavior when a user needs to carefully examine specific content. At this time, the book content occupies a large part of the image, but it may not completely cover the page. Medium distance (25-40 cm) is the most common shooting distance, which can better balance the clarity of the content and the integrity of the page. Long distance (40-60 cm) simulates the scenario when a user needs to shoot multiple pages at the same time or include the surrounding environment.

[0057] The second parameter value of the book's posture is set based on the book's physical characteristics and usage status. The flat state represents the book's shape under ideal conditions, usually seen in new books or hardcover books; the naturally curved state is the most common state, especially for frequently flipped textbooks, where the spine will have a natural curve and the page edges may be slightly curled; the obviously curled state simulates the extreme cases of books that have been used for a long time, have poor storage conditions, or are softcover books, including complex deformations such as page wrinkles, curled edges, and local bulges.

[0058] The third parameter value for lighting conditions covers various indoor and outdoor lighting environments. Normal lighting (300-500 lux) corresponds to standard indoor lighting conditions, such as daily lighting in classrooms and studies; strong lighting (>1000 lux) includes direct sunlight, close-range illumination from a desk lamp, etc., which may cause local overexposure of the image or produce strong shadows; weak lighting (<100 lux) simulates nighttime study and insufficient light environments, and the image may have problems such as increased noise and loss of detail.

[0059] The fourth parameter of background complexity reflects the degree of interference from the surrounding environment of the book. A simple background refers to a desktop with a solid color or simple texture, with the fewest interfering factors; a cluttered background includes common items such as stationery, other books, and water cups, which increases the difficulty of image segmentation and recognition; an occluded background specifically refers to situations where objects such as fingers, pens, and rulers partially obscure the content of the book, which is a very common scenario in actual use but poses a significant challenge to recognition algorithms.

[0060] In this application, based on the parameter values ​​set in the above four dimensions, a discrete parameter set is constructed. This set is a subset of the Cartesian product of all dimension parameter values. For example, if all combinations are performed, 3 (shooting distance) × 3 (book posture) × 3 (lighting conditions) × 3 (background complexity) = 81 basic combinations will be generated.

[0061] In this application, the orthogonal experimental design strategy is an efficient experimental design method that can obtain more comprehensive information with fewer experiments.

[0062] Random sampling strategies randomly select combinations from the complete parameter space. Stratified random sampling can be used to ensure that a certain proportion of each parameter's values ​​are selected; alternatively, Monte Carlo sampling can be used, employing weighted random sampling based on the probability distribution of the actual usage scenario. For example, since mid-range shooting is more common than long-range shooting, a higher sampling weight can be set.

[0063] First, a basic combination sequence is generated based on the selected strategy. Then, a feasibility check is performed on each combination, eliminating combinations that are physically impossible or rarely occur in reality. Next, the number of combinations is optimized and adjusted based on the collection resources and time constraints. Finally, a final list of collection parameter combinations is generated, with each combination assigned a unique combination ID for tracking.

[0064] In this application, through systematic parameter setting and scientific combination strategy, the maximum scene coverage can be obtained with the minimum acquisition cost. The generated acquisition parameter combination not only ensures the diversity of data, but also avoids redundant acquisition, which significantly improves the efficiency and quality of data acquisition.

[0065] Optionally, the method further includes: Configure the range of parameter variations for spatial geometric dimensions, wherein the spatial geometric dimensions include at least the line-of-sight offset angle, the book offset distance, and the book rotation angle; The line-of-sight offset angle is set as the offset of the shooting line of sight relative to the normal of the book center, the book offset distance is set as the displacement of the book center relative to the image center, and the book rotation angle is set as the rotation of the book relative to a preset reference direction. The parameters of the spatial geometric dimension are superimposed and combined with the acquisition parameters to generate the final acquisition parameter combination that simulates real shooting behavior.

[0066] This embodiment further expands the process of generating the combination of acquisition parameters, especially by adding parameter configuration for spatial geometric dimensions, making the acquisition scheme closer to real user shooting behavior.

[0067] In this application, spatial geometric dimension refers to a set of parameters describing the spatial positional relationship between the shooting device and the book. The introduction of these parameters is based on an in-depth analysis of users' actual shooting behavior: when users shoot with a handheld device, it is difficult to maintain a standard upright relationship between the device and the book; various angles and positional offsets are the norm rather than the exception.

[0068] The line-of-sight offset angle is defined as the angle between the optical axis of the shooting device and the normal to the surface of the book. Figure 9 A schematic diagram illustrating the center line of sight offset angle requirement provided by the present invention, as shown below. Figure 9 As shown, the parameter variation range is set from -45 degrees to +45 degrees, where 0 degrees represents vertical shooting. Specifically, this can be further subdivided into: vertical shooting (-5 degrees to +5 degrees), slight tilt (-15 degrees to -5 degrees or +5 degrees to +15 degrees), moderate tilt (-30 degrees to -15 degrees or +15 degrees to +30 degrees), and significant tilt (-45 degrees to -30 degrees or +30 degrees to +45 degrees). Viewpoint shift can cause trapezoidal distortion in the image, which places demands on subsequent image correction algorithms.

[0069] The book offset distance describes the displacement of the book's center point relative to the image's center point. Figure 10 This is a schematic diagram illustrating the required offset distance between the camera and the center point of the book provided by the present invention, as shown below. Figure 10 As shown, this parameter includes two components: horizontal offset and vertical offset. The parameter range is set as a percentage of the image size, such as an offset of 0%, 10%, 20%, or 30% of the image width or height. The book offset simulates the compositional deviation when a user takes a quick photo, requiring the recognition algorithm to handle situations where the target is not in the center of the image.

[0070] The book rotation angle refers to the amount of rotation of the book within its own plane. Figure 11 A schematic diagram of the book rotation angle provided by the present invention, as shown below. Figure 11 As shown. The parameter range is set from -45 degrees to +45 degrees, which can be divided into: no rotation (-5 degrees to +5 degrees), slight rotation (-15 degrees to -5 degrees or +5 degrees to +15 degrees), moderate rotation (-30 degrees to -15 degrees or +15 degrees to +30 degrees), and significant rotation (-45 degrees to -30 degrees or +30 degrees to +45 degrees). Rotated book images require the algorithm to have rotation invariance or be able to perform effective rotation correction.

[0071] In this application, the process of superimposing and combining parameters begins by determining the applicable range of spatial geometric parameters for each combination of basic acquisition parameters. For example, when shooting from a distance, the influence of the line-of-sight offset angle is relatively small, and the sampling density in that dimension can be appropriately reduced; under strong lighting conditions, even slight angular offsets may produce severe reflections, requiring special annotation.

[0072] Secondly, a hierarchical combination strategy is applied. Spatial geometric parameters are divided into primary and secondary parameters, prioritizing sufficient sampling of primary parameters. Nested orthogonal design can be employed, where, based on the orthogonal design of the basic parameters, orthogonal design of spatial geometric parameters is performed for each basic combination.

[0073] Then, combinatorial optimization is performed. Cluster analysis is used to identify and merge combinations with high similarity; combinations are ranked by importance score, with high-value combinations being prioritized; and an upper limit is set on the total number of combinations to achieve optimal coverage under resource constraints.

[0074] Finally, the final acquisition parameter combinations are generated. Each final combination contains complete 7-dimensional parameter values, forming a complete parameter vector such as [mid-distance, natural curvature, normal lighting, cluttered background, 20-degree gaze offset, 15% left offset, -10-degree rotation]. These combinations comprehensively simulate real user shooting behavior, from standard head-on shooting to various non-ideal tilt, offset, and rotation situations.

[0075] By increasing the spatial geometric dimension and scientifically combining them, the acquisition parameter combination constructed in this invention can comprehensively cover various shooting situations in practical applications, especially non-standard shooting scenarios that are ignored by traditional acquisition methods, thereby significantly improving the robustness and accuracy of the model trained on this dataset in practical applications.

[0076] Optionally, performing the acquisition task to obtain book images includes: Adjust the shooting environment and equipment settings according to the combination of acquisition parameters in the acquisition task; For the same page of the same book, take multiple photos with different parameter combinations to collect positive sample images and rejection set images; Among them, the positive sample images are valid images that meet the recognition requirements, and the rejection set of images are invalid images that contain blurry, heavily occluded, or no book content; While acquiring images, the system automatically records and associates the current combination of acquisition parameters with the book's unique identifier.

[0077] In this application, the system needs to make precise adjustments to the shooting environment and equipment based on the combination of acquisition parameters specified in the acquisition task. This adjustment process can be manual, semi-automatic, or fully automatic, depending on the configuration level of the acquisition equipment.

[0078] Regarding the adjustment of the shooting distance, if a fixed acquisition platform is used, the vertical distance between the camera and the book plane can be precisely controlled by a motor-driven lifting device; if a handheld device is used for acquisition, distance information is fed back in real time through a laser rangefinder or visual ranging algorithm to guide the operator to adjust to the target distance range.

[0079] Adjusting the book's posture requires taking appropriate measures depending on the desired posture. For a flat state, use a book-pressing board or a transparent acrylic sheet to gently press the book down; for a naturally curved state, let the book open naturally without applying external force; for a noticeably curled state, the book can be pre-treated, or a book with existing natural wear can be selected.

[0080] Lighting conditions are controlled by combining ambient light sources and supplemental lighting. Normal lighting uses standard indoor lighting; strong lighting is achieved by adding LED supplemental lights or using natural light from windows; weak lighting is achieved by turning off the main light source and retaining only weak ambient light. Light intensity is monitored in real-time using a lux meter to ensure it remains within the preset range.

[0081] The arrangement of the background environment depends on the complexity requirements. A simple background uses a solid-color background board or backdrop; a cluttered background involves randomly placing stationery, books, or other items in the shooting area; and a background with obstructions requires the operator to cooperate by placing fingers or other objects in specific positions within the frame.

[0082] Taking multiple photos of the same page of the same book generates diverse image samples for the same content, thus increasing the richness of the dataset.

[0083] The execution methods for multiple shots include: First, parameter gradual shooting, which involves continuously changing the parameter within a certain range, such as gradually changing the viewing angle from -30 degrees to +30 degrees, taking one shot every 5 degrees; Second, parameter combination shooting, which involves shooting for multiple predetermined parameter combinations to ensure coverage of different scene conditions; Third, random disturbance shooting, which involves adding small random disturbances to the baseline parameters to simulate natural shooting shake.

[0084] Positive sample images must be acquired to ensure the book content is clear and identifiable. Specific criteria include: sharp text edges without significant blurring; proper image exposure with no overexposed or underexposed areas; complete or mostly visible book content; and no severe specular reflections or shadows obscuring key information. Positive sample images are used to train the model's primary recognition capabilities.

[0085] Rejection sets are intentionally created low-quality or invalid images. These include, but are not limited to: severely blurred images; severely occluded images; images without book content; and images unrecognizable due to extreme lighting conditions. The purpose of rejection sets is to train the model's ability to recognize and reject low-quality inputs, thus avoiding misidentification.

[0086] During image acquisition, the system automatically records all relevant metadata information. This recording process is fully automated and requires no manual intervention, thus avoiding human error and omissions.

[0087] The recorded information includes: timestamp, acquisition device information, environmental sensor data, operator identification, task ID and task sequence number, actual acquisition parameter values, and image quality pre-assessment results.

[0088] The association mechanism ensures the binding relationship between each image and its metadata. The main association methods include: First, filename association: key information is encoded into the filename, such as BookID_TaskID_ParamCode_Timestamp.jpg; Second, database association: create image tables and metadata tables in a relational database and establish associations using unique keys; Third, metadata embedding, which directly stores information using the metadata area of ​​the image file format; fourth, associated files, which generate a metadata file with the same name for each image (such as .xml or .json format).

[0089] In this application, multiple association mechanisms are typically employed to ensure data reliability and redundancy. Even if one association method fails, the correspondence between the image and metadata can still be restored through other means.

[0090] In this application, through precise environmental control, a scientific multiple-shot strategy, and a reliable automatic recording mechanism, the present invention achieves high-quality, traceable book image acquisition. Each image carries complete identity information and a birth certificate, providing a solid data foundation for subsequent data processing, model training, and result analysis.

[0091] Optionally, the unique identifier and acquisition parameters are combined and stored as metadata associated with the acquired image, including: Embed task identifier, book unique identifier, and acquisition parameter combination information in the metadata field of the image file; Create an index file to record the mapping relationship between each image and its corresponding unique book identifier and acquisition parameter combination; The associated storage of the metadata supports subsequent data tracing, filtering, and analysis.

[0092] In this application, embedding information in the metadata fields of image files is a direct association method. For JPEG format, the system uses the EXIF ​​user comment field to store the task identifier, unique book identifier, and acquisition parameter combination, serializing this information into a JSON string and writing it; for PNG format, custom keywords are created using tEXt data blocks to store the corresponding information; for TIFF format, private tags are defined to directly store parameter values. The embedding process is automatically completed by the image processing library and does not affect the visual content of the image.

[0093] The index file provides centralized management of mapping relationships. The system establishes an index table containing image file paths, unique book identifiers, task IDs, and parameter values ​​for each dimension, which can be in CSV, JSON, or SQLite database format.

[0094] The index is automatically updated each time a new image is added, and multi-condition queries are supported, such as "find all images of a specific book under low-light conditions". Index updates use a transaction mechanism to ensure data consistency.

[0095] In this application, the mapping relationships include multi-level associations such as image-to-book, image-to-parameter, and image-to-task. This mapping supports flexible data filtering, rapid traceability, and statistical analysis. Through comprehensive metadata association storage, this invention achieves refined management of image data, with each image possessing complete traceable information, supporting efficient retrieval and analysis of large-scale datasets.

[0096] Optionally, the method further includes a step of performing quality detection on the acquired images: Image processing algorithms are used to automatically detect the blur, exposure, and integrity of images. For positive sample images, the recognizability of the book content and the compliance of the acquisition parameters are verified. For rejected images, it is confirmed that they contain the expected interference features. Images that fail the quality check are removed or marked as samples to be resampled.

[0097] In this application, the image processing algorithm automatically evaluates three key metrics. Blur detection uses the Laplacian operator to calculate image sharpness, and images with a variance below a threshold (e.g., 100) are considered blurry. Exposure detection analyzes the histogram distribution to check the proportion of overexposed (brightness > 250) and underexposed (brightness < 5) pixels. Integrity detection identifies the outline of the book through edge detection to determine the integrity of the content.

[0098] For positive sample images, OCR pre-detection is used to assess text recognizability; images with a confidence level below 0.8 need to be resampled. The compliance of the acquisition parameters is verified by estimating the actual parameters using computer vision; images with deviations exceeding the allowable range are marked as inconsistent. For images in the rejection set, it is confirmed that their interference features meet the standards, such as sufficiently high blurriness and occlusion area exceeding 50%.

[0099] Images that fail quality checks are processed in a tiered manner: images with minor issues are marked with a quality tag but retained; images with serious issues are directly removed; images with issues related to key parameters are marked as requiring resampling and supplementary tasks are generated. The system maintains inspection logs to support quality analysis. Through automated quality control, this invention ensures dataset quality, reduces training noise, and improves model performance.

[0100] In an optional embodiment, the process of training a book recognition model using the dataset constructed according to the present invention is as follows: Data preparation involved extracting training images from the metadata index and dividing them into training, validation, and test sets in an 8:1:1 ratio to ensure a balanced parameter distribution across the sets. Image preprocessing included size normalization to 416×416 pixels and targeted data augmentation based on the acquired parameters.

[0101] The model configuration uses deep learning architectures such as ResNet or YOLO, with the number of output layer nodes set according to the number of book categories. A rejection branch is added to the loss function, enabling the model to identify and reject low-quality inputs.

[0102] The training strategy employs a course-based learning approach, gradually transitioning from simple to complex scenarios; it weights samples based on collected parameters to improve performance in challenging scenarios; and it utilizes the rejection set to train the model's quality judgment capabilities.

[0103] Performance evaluation includes metrics such as overall accuracy, accuracy under various parameter conditions, false negative accuracy, and robustness score. Metadata analysis of failure cases identifies weaknesses and guides supplementary data collection.

[0104] This data-driven training fully utilizes rich scene information, making the training process controllable and interpretable, and the resulting model exhibits excellent robustness in real-world scenarios.

[0105] The data acquisition device provided by the present invention is described below. The data acquisition device described below and the data acquisition method described above can be referred to in correspondence.

[0106] Figure 12 This is a schematic diagram of the data acquisition device structure provided by the present invention, as shown below. Figure 12 As shown, it includes the following modules: The first generation module 1210 is used to obtain the entity information of the book to be collected, encode the entity information according to the preset hierarchical encoding rules, and generate a unique identifier code corresponding to the book to be collected. The second generation module 1220 is used to generate a combination of acquisition parameters based on multiple preset acquisition dimensions and the parameter range of each dimension. The acquisition dimensions include shooting distance, book posture, lighting conditions, background complexity, line of sight offset angle, book offset distance, and book rotation angle. The third generation module 1230 is used to generate a collection task based on the unique identifier and the collection parameter combination, wherein the collection task includes the unique identifier of the target book and the corresponding collection parameter combination. The acquisition module 1240 is used to perform the acquisition task to obtain book images. During the acquisition process, the unique identifier and acquisition parameters are combined as metadata and stored in association with the acquired images to construct a book image dataset.

[0107] In this application, a unique identifier code is generated based on hierarchical coding rules, achieving standardized management and precise tracking of the information of the book entities to be collected, effectively solving the problem of chaotic management of multiple versions of books. By generating a combination of collection parameters that comprehensively covers real-world scenarios based on preset parameters such as shooting distance, book posture, lighting conditions, background complexity, and spatial geometric dimensions, the constructed dataset has high diversity and real-world fit, thereby significantly improving the generalization ability and robustness of the image processing model trained on this dataset. At the same time, through a collection task-driven and metadata automatic association and storage mechanism, a strong binding between image data and collection parameters and book identity is achieved, greatly improving the standardization and automation level of data collection, and providing a solid foundation for subsequent data traceability, screening, and the construction of high-quality datasets.

[0108] Figure 13This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 13 As shown, the electronic device may include a processor 1310, a communications interface 1320, a memory 1330, and a communication bus 1340. The processor 1310, communications interface 1320, and memory 1330 communicate with each other via the communication bus 1340. The processor 1310 can call logical instructions in the memory 1330 to execute a data acquisition method. This method includes: acquiring entity information of the book to be acquired; encoding the entity information according to a preset hierarchical encoding rule; and generating a unique identifier code corresponding to the book to be acquired. Based on multiple preset acquisition dimensions and the parameter range of each dimension, a combination of acquisition parameters is generated. The acquisition dimensions include shooting distance, book posture, lighting conditions, background complexity, line of sight offset angle, book offset distance, and book rotation angle. A collection task is generated based on the unique identifier and the collection parameter combination. The collection task includes the unique identifier of the target book and the corresponding collection parameter combination. The acquisition task is executed to obtain book images. During the acquisition process, the unique identifier and acquisition parameters are combined as metadata and stored in association with the acquired images to construct a book image dataset.

[0109] Furthermore, the logical instructions in the aforementioned memory 1330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0110] On the other hand, the present invention also provides a computer program product, the computer program product including a computer program, the computer program being stored on a non-transitory computer-readable storage medium, the computer program being executed by a processor, the computer being able to execute the data acquisition method provided by the above methods, the method including: acquiring entity information of the book to be acquired, encoding the entity information according to a preset hierarchical encoding rule, and generating a unique identifier code corresponding to the book to be acquired; Based on multiple preset acquisition dimensions and the parameter range of each dimension, a combination of acquisition parameters is generated. The acquisition dimensions include shooting distance, book posture, lighting conditions, background complexity, line of sight offset angle, book offset distance, and book rotation angle. A collection task is generated based on the unique identifier and the collection parameter combination. The collection task includes the unique identifier of the target book and the corresponding collection parameter combination. The acquisition task is executed to obtain book images. During the acquisition process, the unique identifier and acquisition parameters are combined as metadata and stored in association with the acquired images to construct a book image dataset.

[0111] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, it implements the data acquisition method provided by the above methods. The method includes: acquiring entity information of a book to be acquired, encoding the entity information according to a preset hierarchical encoding rule, and generating a unique identifier code corresponding to the book to be acquired. Based on multiple preset acquisition dimensions and the parameter range of each dimension, a combination of acquisition parameters is generated. The acquisition dimensions include shooting distance, book posture, lighting conditions, background complexity, line of sight offset angle, book offset distance, and book rotation angle. A collection task is generated based on the unique identifier and the collection parameter combination. The collection task includes the unique identifier of the target book and the corresponding collection parameter combination. The acquisition task is executed to obtain book images. During the acquisition process, the unique identifier and acquisition parameters are combined as metadata and stored in association with the acquired images to construct a book image dataset.

[0112] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0113] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0114] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions 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.

Claims

1. A data acquisition method, characterized in that, include: Obtain the entity information of the book to be collected, encode the entity information according to the preset hierarchical coding rules, and generate a unique identifier code corresponding to the book to be collected; Based on multiple preset acquisition dimensions and the parameter range of each dimension, a combination of acquisition parameters is generated. The acquisition dimensions include shooting distance, book posture, lighting conditions, background complexity, line of sight offset angle, book offset distance, and book rotation angle. A collection task is generated based on the unique identifier and the collection parameter combination. The collection task includes the unique identifier of the target book and the corresponding collection parameter combination. The acquisition task is executed to obtain book images. During the acquisition process, the unique identifier and acquisition parameters are combined as metadata and stored in association with the acquired images to construct a book image dataset.

2. The data acquisition method according to claim 1, characterized in that, The hierarchical encoding rules include: The books to be collected are classified according to a hierarchical structure of type, subject, grade level, publisher, and serial number; Each level is assigned a corresponding coding segment, and the coding segments of each level are combined to form a complete unique identifier. The unique identifier is used to identify and track specific books during the collection, storage, and retrieval process.

3. The data acquisition method according to claim 1, characterized in that, The step of generating a combination of acquisition parameters based on multiple preset acquisition dimensions and the parameter range of each dimension specifically includes: The shooting distance is set to a first parameter value of close distance, medium distance and long distance; the book posture is set to a second parameter value of flat state, natural curved state and obvious curled state; the lighting conditions are set to a third parameter value of normal lighting, strong lighting and weak lighting; and the background complexity is set to a fourth parameter value of simple background, cluttered background and occluded background. Based on the first parameter value, the second parameter value, the third parameter value, and the fourth parameter value, a discrete parameter set is obtained. By applying orthogonal experimental design strategies or random sampling strategies, parameter values ​​are extracted from the discrete parameter set and combined operations are performed to generate combinations of collected parameters covering different environmental scenarios.

4. The data acquisition method according to claim 3, characterized in that, The method further includes: Configure the range of parameter variations for spatial geometric dimensions, wherein the spatial geometric dimensions include at least the line-of-sight offset angle, the book offset distance, and the book rotation angle; The line-of-sight offset angle is set as the offset of the shooting line of sight relative to the normal of the book center, the book offset distance is set as the displacement of the book center relative to the image center, and the book rotation angle is set as the rotation of the book relative to a preset reference direction. The parameters of the spatial geometric dimension are superimposed and combined with the acquisition parameters to generate the final acquisition parameter combination that simulates real shooting behavior.

5. The data acquisition method according to claim 1, characterized in that, The process of performing the acquisition task to obtain book images includes: Adjust the shooting environment and equipment settings according to the combination of acquisition parameters in the acquisition task; For the same page of the same book, take multiple photos with different parameter combinations to collect positive sample images and rejection set images; Among them, the positive sample images are valid images that meet the recognition requirements, and the rejection set of images are invalid images that contain blurry, heavily occluded, or no book content; While acquiring images, the system automatically records and associates the current combination of acquisition parameters with the book's unique identifier.

6. The data acquisition method according to claim 5, characterized in that, The unique identifier and acquisition parameters are combined and stored as metadata associated with the acquired image, including: Embed task identifier, book unique identifier, and acquisition parameter combination information in the metadata field of the image file; Create an index file to record the mapping relationship between each image and its corresponding unique book identifier and acquisition parameter combination; The associated storage of the metadata supports subsequent data tracing, filtering, and analysis.

7. The data acquisition method according to claim 1, characterized in that, The method also includes a step of quality detection of the acquired images: Image processing algorithms are used to automatically detect the blur, exposure, and integrity of images. For positive sample images, the recognizability of the book content and the compliance of the acquisition parameters are verified. For rejected images, it is confirmed that they contain the expected interference features. Images that fail the quality check are removed or marked as samples to be resampled.

8. A data acquisition device, characterized in that, include: The first generation module is used to obtain the entity information of the book to be collected, encode the entity information according to the preset hierarchical encoding rules, and generate a unique identifier code corresponding to the book to be collected. The second generation module is used to generate a combination of acquisition parameters based on multiple preset acquisition dimensions and the parameter range of each dimension. The acquisition dimensions include shooting distance, book posture, lighting conditions, background complexity, line of sight offset angle, book offset distance, and book rotation angle. The third generation module is used to generate a collection task based on the unique identifier and the collection parameter combination. The collection task includes the unique identifier of the target book and the corresponding collection parameter combination. The acquisition module is used to perform the acquisition task to obtain book images. During the acquisition process, the unique identifier and acquisition parameters are combined as metadata and stored in association with the acquired images to construct a book image dataset.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the data acquisition method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the data acquisition method as described in any one of claims 1 to 7.

Images