Book information analysis method based on Internet of Things

Abstract

The invention discloses a book information analysis method based on the Internet of Things, through deep learning and OCR technologies, even though the appearance of a book changes, classification information of the book can be accurately identified, and the book management accuracy and retrieval efficiency are improved. Meanwhile, boundary line and classification consistency check and a dual verification mechanism of RFID tags are combined, so that the accuracy of book placement and classification information is ensured, manual proofreading work is reduced, and the management efficiency is improved. Besides, by defining a unique folding angle identifier for each type of books and combining the RFID technology, automatic identification and rapid positioning of book types are realized, and the borrowing experience of readers and the management efficiency of libraries are greatly improved. According to the scheme, the folding angle length standard is introduced, the classification accuracy is further improved, and efficient classification, accurate positioning and automatic management of the library books are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of book information analysis, and in particular to a book information analysis method based on the Internet of Things. Background Art

[0002] In the construction and management of smart libraries, the integration of intelligent technology has greatly improved the automation and intelligence level of libraries, making library services no longer limited by physical space, and providing users with a more convenient and efficient reading experience. However, in the process of realizing intelligent book borrowing and returning and related information recommendation functions, technical challenges still exist, especially the accurate identification and positioning of books.

[0003] Traditional book recognition technologies often rely on the Internet of Things (IoT) and image recognition technologies, but these technologies face many challenges in practical applications. For example, when using IoT technology for book detection, a significant problem is the noise interference of book boundaries. Since the placement of books in the library may be affected by various factors such as light, angle, and occlusion, the boundaries of books may become blurred or irregular in the image, resulting in some books being unable to be accurately identified and then being incorrectly included in the data statistics.

[0004] This inaccurate identification not only affects the statistical accuracy of the self-service book borrowing and returning process, but may also interfere with the recommendation of book subscription service information for library users. Summary of the invention

[0005] This application provides a book information analysis method based on the Internet of Things. It integrates deep feature extraction, region candidate generation network, ROI alignment technology, OCR technology and RFID technology to address the above problems. By collecting book images and extracting visual features, candidate frames are generated and two position corrections are performed to ensure the accuracy of the book position. At the same time, the solution takes into account the noise problem of book boundary lines. After screening out valid books, the OCR technology and RFID technology are combined to identify book information, and the information is compared and updated with the RFID tag information. Finally, accurate book location information is output, providing reliable technical support for book borrowing and information recommendation in smart libraries.

[0006] This application provides a book information analysis method based on the Internet of Things, including:

[0007] S101, collecting images of books placed by library users according to specified book placement requirements, performing deep feature extraction on the book placement images, and obtaining visual features of the images;

[0008] S102, using the visual features extracted by the candidate box generation as input features, using the region candidate generation network, generating a candidate box for each target book in the book placement image, and performing a first position correction to obtain a corrected detection box;

[0009] S103, extracting a feature matrix of the same size from the detection frame corrected for the first time through ROI alignment and performing a second position correction;

[0010] S104, judging whether each target book does not exceed a preset boundary line position according to the previously acquired outline information of each target book;

[0011] S105, for the outline area of ​​the target book that does not exceed the boundary line, the OCR technology is used to recognize the spine and text information on the book;

[0012] S106, using an RFID reader to identify the RFID tag attached to each book, and obtaining basic information of the currently stored book;

[0013] S107, comparing the text information with the basic information of the RFID tag to update the output book position.

[0014] Preferably, the S105 includes:

[0015] Perform grayscale and denoising preprocessing operations on the extracted image area;

[0016] Delineate the text information area that needs to be identified;

[0017] Select the OCR recognition software Tesseract and load the preprocessed image;

[0018] Select the designated recognition area for recognition and obtain the book information of the library management library.

[0019] Preferably, in S105, the text information includes:

[0020] The title and author of the book.

[0021] Preferably, the S106 includes:

[0022] When the outline area of ​​the target book is identified, the RFID reader is started to read the RFID tag information;

[0023] When scanning the spine of a book, a fixed scanning RFID tag area is defined, and the book is placed within the reading range of the RFID reader; the read RFID tag information is stored and passed to the subsequent steps for comparison.

[0024] Preferably, the RFID tag in S106 includes:

[0025] The basic information of a book includes title, author name, book type, book subject terms and classification number.

[0026] Preferably, the S107 includes:

[0027] Compare the text information recognized by OCR technology with the basic information recognized by the RFID tag;

[0028] If the two information are consistent, the specific location of the target book is output;

[0029] If the two information are inconsistent, the RFID tag information is updated according to the recognized text information, and the specific location of the target book is output.

[0030] Preferably, the S106 further includes:

[0031] S201, defining a unique fold angle for each type of primary book classification;

[0032] S202, adding a digital code to the RFID tag, and assigning a unique fold angle of the corresponding book type to the code;

[0033] S203, connecting the RFID reader to the library management system. When a book is placed near the reader, the reader will automatically scan and read the information in the tag.

[0034] The S204 reader / writer reads and writes the digital code in the RFID tag, identifies the digital code and matches the book type folding angle;

[0035] S205, displaying and determining the location of the book in the library bookshelf according to the book type.

[0036] Preferably, the step S201 includes:

[0037] Analyze the library's collection types and classification standards, and determine the primary book classification types that need to be defined;

[0038] Design a unique corner mark for each category, store and record these corner marks, and associate them with the corresponding book categories;

[0039] First-level book classification types may include: literature, science and technology, art, education, and social sciences.

[0040] Preferably, the step S201, folding corners, includes:

[0041] The angle is the angle formed by the intersection of two straight line segments, and the intersection point is the center point O of the virtual two-dimensional coordinate axis;

[0042] The angle formed by taking point O as the starting point in any quadrant of the first, second, third or fourth quadrant or the angle formed across quadrants can be called a bend angle.

[0043] Preferably, the step S201 includes:

[0044] S301, performing secondary book classification for each primary book classification;

[0045] S302, setting different folding angle length standards for first-level book classification folding angles of different angles;

[0046] S303, in the library management system, recording the standard of the folding angle length of each secondary book classification, and establishing a mapping relationship between the folding angle mark and the secondary book classification;

[0047] S304, after the self-service book borrowing and returning device scans the spine and cover of the book, it identifies the folding angle and length of the RFID tag;

[0048] S305, displaying the type and location of the book according to the folded corner after the length standard is limited.

[0049] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0050] 1. The accuracy of book classification has been significantly improved by extracting deep features and strengthening classification features of book images, as well as introducing preliminary classification screening and fine classification steps. This not only helps libraries manage books more efficiently, but also provides readers with more accurate book retrieval and recommendation services. The combination of deep learning and OCR technology enables accurate identification of classification information even if the cover or spine of a book is worn, blocked, or the angle changes.

[0051] 2. Through the double verification mechanism of boundary line and classification consistency check, as well as RFID tag identification and classification information comparison, the correct placement of books in the library and the accuracy of classification information are ensured. This greatly reduces the manual proofreading and sorting work of library staff and improves the automation and intelligent level of book management. At the same time, the timely update of classification information also ensures the real-time and accuracy of library book information, providing strong support for the operation and management of the library.

[0052] 3. By defining unique corner marks for each type of book and combining it with RFID technology, the system can automatically identify the type of book and quickly locate its position in the bookshelf. This automated management method not only greatly improves the management efficiency of the library, but also reduces the error rate of manual classification and positioning, making the borrowing and returning of books smoother. This technical solution also significantly improves the borrowing experience of readers. Readers can quickly obtain the accurate location information of the required books through simple query operations without having to blindly search among the huge bookshelves. This convenience not only saves readers' time, but also enhances their satisfaction with library services. At the same time, the library can also better understand readers' borrowing needs through this technical solution, providing strong support for future book procurement and layout optimization.

[0053] 4. A corner marking system was introduced to assist in the classification and management of books. First, different corner length standards were set for the first-level book classifications at different angles. This standard not only takes into account the feasibility of actual operations, but also ensures the accuracy of classification. Subsequently, in the library management system, the corner length standards for each second-level book classification were recorded in detail, and a mapping relationship between the corner marking and the second-level book classification was established. When the self-service book borrowing and returning device scans the spine and book cover, it can quickly identify the corner angle and length of the RFID tag, and accurately display the type and location of the book according to the preset corner length standard. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 The present invention is a flowchart of a method for analyzing book information based on the Internet of Things. DETAILED DESCRIPTION

[0055] To facilitate the understanding of the present invention, the present application will be described more comprehensively below with reference to the relevant drawings; the drawings show preferred embodiments of the present invention, but the present invention can be implemented in many different forms and is not limited to the embodiments described herein; on the contrary, the purpose of providing these embodiments is to enable a more thorough and comprehensive understanding of the disclosed content of the present invention.

[0056] It should be noted that the terms “vertical”, “horizontal”, “up”, “down”, “left”, “right” and similar expressions used in this document are only for illustrative purposes and do not represent the only implementation method.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which the present invention belongs; the terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention; the term "and / or" used herein includes any and all combinations of one or more related listed items.

[0058] Embodiment 1:

[0059] Figure 1 The present invention is a flowchart of a method for analyzing book information based on the Internet of Things.

[0060] like Figure 1 As shown, a book information analysis method based on the Internet of Things includes the following steps:

[0061] S101, collecting images of books placed by library users according to specified book placement requirements, performing deep feature extraction on the book placement images, and obtaining visual features of the images.

[0062] Specifically, on the book scanning platform of the self-service book borrowing and returning device, the images of books placed by library users in the specified unobstructed white background scanning area are collected, and it is required that there are no other obstacles in front of the collection camera;

[0063] Then, the ResNet network structure is used as the basic AI network model to perform deep feature extraction on the collected book placement images to obtain the visual features of the images.

[0064] For the ResNet network structure, please adjust the number of network layers, convolution kernel size, step size and other parameters according to actual needs to ensure the accuracy and efficiency of feature extraction.

[0065] S102, using the visual features extracted by the candidate box generation as input features, using the region candidate generation network, generating a candidate box for each target book in the book placement image, and performing a first position correction to obtain a corrected detection box.

[0066] Among them, a sliding window is used to generate candidate boxes according to preset rules, and the RPN composed of convolutional layers is used to determine whether all generated candidate boxes belong to the foreground area or the background area. According to the judgment result, the position of the candidate box belonging to the foreground area is corrected to obtain the corrected detection box.

[0067] S103, extracting a feature matrix of the same size from the detection frame corrected for the first time through ROI alignment and performing a second position correction.

[0068] Specifically, the feature representations corresponding to all the detection frames obtained by the candidate frame generation module are obtained according to ROI alignment, and feature matrices of the same size are extracted for all the detection frames;

[0069] Taking the feature matrix extracted by ROI alignment as input, a second position correction is performed on each detection frame to generate the contour information of each target book after segmentation;

[0070] Among them, the process of the second position correction is:

[0071] Taking the feature matrix extracted by ROI alignment as input, the final position correction is performed for each detection box through the fully connected layer, and the feature matrix is ​​processed using a fully convolutional network to classify each pixel in the feature matrix, thereby generating segmented contour information for each detected target book.

[0072] S104: judging whether each target book does not exceed a preset boundary line position according to the previously acquired outline information of each target book.

[0073] If the target book does not exceed the boundary line position, the subsequent steps are continued; if it exceeds, a prompt message indicating that the target book recognition is unsuccessful is output, triggering rescanning or manual adjustment of the book position.

[0074] S105 , for the outline area of ​​the target book that does not exceed the boundary line, the OCR technology is applied to recognize the spine and text information on the book.

[0075] Specifically, the extracted image area is preprocessed by graying and denoising, the text information area to be recognized is delineated, the OCR recognition software Tesseract is selected, the preprocessed image is loaded, and the delineated recognition area is selected for recognition to obtain the book information of the library management library.

[0076] The text information includes the title and author of the book.

[0077] S106, using an RFID reader to identify the RFID tag attached to each book, and obtaining basic information of the currently stored book.

[0078] Specifically, when the outline area of ​​the target book is identified, the RFID reader is started to prepare to read the RFID tag information. When scanning the spine of the book, a fixed scanning RFID tag area is defined. The book is placed within the reading range of the RFID reader to ensure that the RFID tag can be read correctly. The read RFID tag information is stored and passed to the subsequent steps for comparison.

[0079] Among them, the RFID tag includes the basic information of the book, including the title, author name, book type, book subject terms and classification number.

[0080] S107, comparing the text information with the basic information of the RFID tag to update the output book position.

[0081] Specifically, the text information recognized by OCR technology is compared with the basic information recognized by the RFID tag. If the two information are consistent, the specific location of the target book is output; if the two information are inconsistent, the RFID tag information is updated according to the recognized text information, and the specific location of the target book is output.

[0082] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0083] 3. The accuracy of book classification has been significantly improved by extracting deep features and strengthening classification features of book images, as well as introducing preliminary classification screening and fine classification steps. This not only helps libraries manage books more efficiently, but also provides readers with more accurate book retrieval and recommendation services. The combination of deep learning and OCR technology allows accurate identification of classification information even if the cover or spine of a book is worn, blocked, or the angle changes.

[0084] 4. Through the double verification mechanism of boundary line and classification consistency check, as well as RFID tag identification and classification information comparison, the correct placement of books in the library and the accuracy of classification information are ensured. This greatly reduces the manual proofreading and sorting work of library staff and improves the automation and intelligent level of book management. At the same time, the timely update of classification information also ensures the real-time and accuracy of library book information, providing strong support for the operation and management of the library.

[0085] Embodiment 2:

[0086] The above-mentioned embodiment 1 realizes comprehensive and accurate recognition of book classification information by comprehensively applying deep learning, OCR technology and RFID technology, significantly enhances the accuracy of book classification, and improves the efficiency and automation and intelligent level of book management. In order to further improve the efficiency and accuracy of book classification, the embodiment 1 is further improved by adding digital codes in the RFID tags, and the codes correspond to the book corners to determine the specific classification and location of the books.

[0087] Now, further improvements are made on the basis of step S106 in the first embodiment, specifically:

[0088] S201, defining a unique fold angle for each type of first-level book classification.

[0089] Specifically, analyze the types of books in the library and the classification standards, and determine the first-level book classification types that need to be defined. The types of books are the first-level classifications of book types, including: literature, science and technology, art, education, and social sciences. Design a unique dog-ear mark for each classification, store these dog-ear marks, and associate them with the corresponding book classifications.

[0090] Among them, the angle is the angle formed by the intersection of two straight line segments, and the intersection point is the center point O of the circle in the virtual two-dimensional coordinate axis. The angle formed in any quadrant of the first, second, third, or fourth quadrant or across quadrants with point O as the starting point can be called the angle.

[0091] S202, adding a digital code to the RFID tag, and assigning a unique fold angle of the corresponding book type to the code.

[0092] S203, connecting the RFID reader to the library management system. When a book is placed near the reader, the reader will automatically scan and read the information in the tag.

[0093] The S204 reader / writer reads and writes the digital code in the RFID tag, identifies the digital code and matches the book type folding angle.

[0094] Specifically, the library management system receives the digital code sent by the reader, and the system searches the database for the book type corner corresponding to the code, and matches the query result with the book type corner.

[0095] S205, displaying and determining the location of the book in the library bookshelf according to the book type.

[0096] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0097] 1. By defining unique corner marks for each type of book and combining it with RFID technology, the system can automatically identify the type of book and quickly locate its position in the bookshelf. This automated management method not only greatly improves the management efficiency of the library, but also reduces the error rate of manual classification and positioning, making the borrowing and returning of books smoother. This technical solution also significantly improves the borrowing experience of readers. Readers can quickly obtain the accurate location information of the required books through simple query operations without having to blindly search among the huge bookshelves. This convenience not only saves readers' time, but also enhances their satisfaction with library services. At the same time, the library can also better understand readers' borrowing needs through this technical solution, providing strong support for future book procurement and layout optimization.

[0098] Embodiment three:

[0099] In the above-mentioned second embodiment, by defining a unique corner mark for each book type (such as literature, science, etc.), different types of books can be clearly divided for easy management and search. In order to further efficiently classify, locate and manage books, improve management efficiency and reader satisfaction, further improvements are made on the basis of step S201 of the second embodiment, specifically:

[0100] S301, performing secondary book classification for each primary book classification.

[0101] Among them, the secondary book categories include:

[0102] Secondary book categories for literature: novels, essays, and poems;

[0103] Secondary science and technology book classification: books on physics, chemistry, biology, computer science, engineering technology, medicine and other disciplines;

[0104] Secondary book classification of social sciences: books on political science, economics, sociology, psychology, history, geography and other disciplines;

[0105] Secondary book classification of art: In addition to music, painting, sculpture, and drama, it also includes books on other art forms such as dance, film, and photography;

[0106] Secondary book classification of education: textbooks and educational theory books for different educational stages such as early childhood education, primary education, secondary education, and higher education.

[0107] It should be noted that the classification of books is not fixed, it can be flexibly adjusted according to different purposes and needs.

[0108] S302, setting different folding angle length standards for first-level book classification folding angles of different angles.

[0109] S303, in the library management system, record the standard of the folding angle length of each secondary book classification, and establish a mapping relationship between the folding angle mark and the secondary book classification.

[0110] S304, after the self-service book borrowing and returning device scans the book spine and cover, it identifies the folding angle and length of the RFID tag.

[0111] S305, displaying the type and location of the book according to the folded corner after the length standard is limited.

[0112] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0113] 1. A corner marking system was introduced to assist in the classification and management of books. First, different corner length standards were set for the first-level book classifications at different angles. This standard not only takes into account the feasibility of actual operations, but also ensures the accuracy of classification. Subsequently, in the library management system, the corner length standards for each second-level book classification were recorded in detail, and a mapping relationship between the corner marking and the second-level book classification was established. When the self-service book borrowing and returning device scans the spine and book cover, it can quickly identify the corner angle and length of the RFID tag, and accurately display the type and location of the book according to the preset corner length standard.

[0114] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A book information analysis method based on the Internet of Things, characterized in that: include: S101, collecting images of books placed by library users according to specified book placement requirements, performing deep feature extraction on the book placement images, and obtaining visual features of the images; S102, using the visual features extracted by the candidate box generation as input features, using the region candidate generation network, generating a candidate box for each target book in the book placement image, and performing a first position correction to obtain a corrected detection box; S103, extracting a feature matrix of the same size from the detection frame corrected for the first time through ROI alignment and performing a second position correction; S104, judging whether each target book does not exceed a preset boundary line position according to the previously acquired outline information of each target book; S105, for the outline area of ​​the target book that does not exceed the boundary line, the OCR technology is used to recognize the spine and text information on the book; S106, using an RFID reader to identify the RFID tag attached to each book, and obtaining basic information of the currently stored book; S107, comparing the text information with the basic information of the RFID tag to update the output book position.

2. A book information analysis method based on the Internet of Things as claimed in claim 1, characterized in that: The step S105 includes: performing grayscale and denoising preprocessing operations on the extracted image area; Delineate the text information area that needs to be identified; Select the OCR recognition software Tesseract and load the preprocessed image; Select the designated recognition area for recognition and obtain the book information of the library management library.

3. A book information analysis method based on the Internet of Things as claimed in claim 1, characterized in that: The text information in S105 includes: The title and author of the book.

4. A book information analysis method based on the Internet of Things as claimed in claim 1, characterized in that: The step S106 includes: when the outline area of ​​the target book is identified, starting the RFID reader to prepare to read the RFID tag information; When scanning the spine of a book, a fixed scanning RFID tag area is defined, and the book is placed within the reading range of the RFID reader; the read RFID tag information is stored and passed to the subsequent steps for comparison.

5. A book information analysis method based on the Internet of Things as claimed in claim 1, characterized in that: The RFID tag in S106 includes: The basic information of a book includes title, author name, book type, book subject terms and classification number.

6. A book information analysis method based on the Internet of Things as claimed in claim 1, characterized in that: The step S107 includes: comparing the text information recognized by the OCR technology with the basic information recognized by the RFID tag; If the two information are consistent, the specific location of the target book is output; If the two information are inconsistent, the RFID tag information is updated according to the recognized text information, and the specific location of the target book is output.

7. A book information analysis method based on the Internet of Things as claimed in claim 1, characterized in that: The S106 further includes: S201, defining a unique fold angle for each type of primary book classification; S202, adding a digital code to the RFID tag, and assigning a unique fold angle of the corresponding book type to the code; S203, connecting the RFID reader to the library management system. When a book is placed near the reader, the reader will automatically scan and read the information in the tag. The S204 reader / writer reads and writes the digital code in the RFID tag, identifies the digital code and matches the book type folding angle; S205, displaying and determining the location of the book in the library bookshelf according to the book type.

8. A book information analysis method based on the Internet of Things as claimed in claim 5, characterized in that: The step S201 includes: analyzing the types of books in the library and the classification standards, and determining the primary book classification types that need to be defined; Design a unique corner mark for each category, store and record these corner marks, and associate them with the corresponding book categories; First-level book classification types may include: literature, science and technology, art, education, and social sciences.

9. A book information analysis method based on the Internet of Things as claimed in claim 7, characterized in that: The S201, corner fold, comprises: The angle is the angle formed by the intersection of two straight line segments, and the intersection point is the center point O of the virtual two-dimensional coordinate axis; The angle formed by taking point O as the starting point in any quadrant of the first, second, third or fourth quadrant or the angle formed across quadrants can be called a bend angle.

10. The method for analyzing book information based on the Internet of Things according to claim 7, characterized in that: The S201 includes: S301, performing secondary book classification for each primary book classification; S302, setting different folding angle length standards for first-level book classification folding angles of different angles; S303, in the library management system, recording the standard of the folding angle length of each secondary book classification, and establishing a mapping relationship between the folding angle mark and the secondary book classification; S304, after the self-service book borrowing and returning device scans the book spine and cover, it identifies the folding angle and length of the RFID tag; S305, displaying the type and location of the book according to the folded corner after the length standard is limited.

Images