Intelligent book self-service borrowing and returning system and method

Abstract

The invention discloses an intelligent book self-service borrowing and returning system and method. The system comprises a user interaction layer which provides an operation interface; the identity verification module adopts a three-level security mechanism, namely dynamic living body detection, cloud face comparison and mobile terminal OTP verification, and is combined with edge computing equipment to locally process feature extraction; the book management module integrates a damage detection system through RFID batch identification and weight sensor dual verification, a camera scans a cover / spine to identify tearing and doodling and detect obvious damage, and a pressure sensor detects page thickness abnormity; the overdue automatic fee deduction module is used for monitoring the book returning condition of the reader and automatically starting a fee deduction process once overdue is found; the reader reading preference analysis module is used for analyzing the borrowing history and behavior data of the readers; and the data security and background management module is used for guaranteeing privacy and data integrity through local encrypted storage, block chain evidence storage and federal learning.

Description

Technical Field

[0001] The present invention relates to the technical field of book management, and in particular to an intelligent self-service book borrowing and returning system and method. Background Art

[0002] With the rapid development of intelligent technology, the library industry is facing an urgent need to transform from traditional manual services to self-service. As an important component of intelligent services, self-service borrowing and returning systems can significantly improve library service efficiency and user experience, meeting readers' demand for convenient and efficient services.

[0003] In the existing technology, during the batch borrowing and returning process, systems that rely on a single technology (such as RFID or barcodes) for book identification may fail to identify books due to technical limitations or environmental interference, resulting in book omissions. Book omissions not only affect the library's inventory management, causing books to be lost or misplaced, but also reduce readers' borrowing experience. Therefore, an intelligent self-service book borrowing and returning system and method are proposed. Summary of the Invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an intelligent self-service book borrowing and returning system and method.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] An intelligent self-service book borrowing and returning system, comprising:

[0007] User interaction layer: This layer provides an operation interface that uses an infrared dual-camera module for dynamic liveness detection and facial feature collection. It also integrates an audio-visual prompt system and an emergency physical switch. The audio-visual prompt system uses sound prompts and light indicators to guide readers through borrowing and returning operations. The emergency physical switch switches to privacy protection mode with one click, disconnecting biometric feature collection.

[0008] Identity verification module: This module uses a three-level security mechanism – dynamic liveness detection, cloud-based face comparison, and mobile OTP verification – combined with local feature extraction processing on edge computing devices. Dynamic liveness detection uses dual infrared cameras to detect liveness features to prevent photo or video attacks. Cloud-based face comparison compares collected facial feature vectors with pre-stored ID photos for identity verification. When the recognition confidence level falls below a threshold, a one-time verification code is sent to the user's bound mobile phone to complete secondary verification.

[0009] Book Management Module: This module integrates a damage detection system using RFID batch identification and weight sensor verification. A camera scans the cover / spine to identify torn or graffiti, detecting obvious damage, while a pressure sensor detects abnormal page thickness.

[0010] Overdue automatic deduction module: monitors readers' book return status and automatically initiates the deduction process once an overdue payment is detected. Through Alipay account / bank card binding, RESTful API and bank deduction interface, the overdue amount is deducted from the Alipay account balance. If the balance is insufficient, the overdue amount is deducted from the bound bank card. After the deduction is successful, the reader receives a deduction notification and the overdue record is recorded.

[0011] Reader reading preference analysis module: This module uses questionnaires, interviews, and decision tree algorithms to analyze readers' borrowing history and behavior data, providing readers with personalized reading recommendations while optimizing the library's resource development and management strategies.

[0012] Data security and background management module: Ensure privacy and data integrity through local encrypted storage, blockchain evidence storage and federated learning, and support automatic upload of abnormal logs and blockchain traceability.

[0013] The above technical solution further includes:

[0014] Preferably, the operating interface is tilted at an angle of 44°-46°, and the screen content is visible to the reader when the reader is operating.

[0015] Preferably, the infrared dual-camera module includes a black and white camera module, a color camera module and an infrared lighting unit, which are assembled on the same circuit board substrate. A double-pass filter is arranged in front of the black and white camera module, which transmits visible light with a wavelength of 400nm to 650nm and infrared light with a wavelength of 700nm to 1000nm to capture facial images. An infrared cutoff filter is arranged in front of the color camera module, and the infrared lighting unit emits infrared light with a wavelength range of 700nm to 1000nm.

[0016] Preferably, the sound and light prompt system adopts a sound and light alarm, which includes two parts: sound prompt and light indication. The sound prompt part sends out sound instructions through a speaker, and the light indication part prompts the reader of the operation status through lights of different colors and flashing frequencies.

[0017] Preferably, the specific steps of the book management module for identifying and detecting the status of books are:

[0018] RFID batch identification: Each book is attached with an RFID tag, which stores the book's unique coding information. The reader places the book in the identifiable area of ​​the RFID reader, and the wireless radio frequency identification technology is used to read the unique coding information in the book tag;

[0019] Double verification of weight sensors: A weight sensor is placed on the operating platform of the book lending and returning machine to measure the total weight of the books. When books are placed on the operating platform, the sensor measures the total weight of the books and outputs an electrical signal. The output electrical signal is converted into a weight value. The number of books identified by RFID is compared with the weight measured by the weight sensor to see if they match. If they do not match, the verification passes; if not, an alarm is triggered, prompting staff to conduct manual verification.

[0020] Camera scanning and image recognition: Scan the book cover / spine with a camera to obtain image information, pre-process the collected image information, and use the improved Faster R-CNN to extract feature information in the image. Based on the extracted feature information, determine whether the book is damaged;

[0021] Pressure sensor detection: Use the pressure sensor to measure the pressure change of the book page on the sensor, and judge whether there is any abnormality in the thickness of the book page through data analysis;

[0022] Cross-validation and overall detection: The data from RFID identification, weight sensors, camera scanning, and pressure sensor detection are integrated. By comparing and analyzing the data from different sensors and detection technologies, cross-validation is performed, and a joint model is used to comprehensively analyze image features and pressure data.

[0023] Preferably, the overdue automatic deduction module uses the HTTPS protocol to establish a secure connection with the bank system and transmits parameters according to the requirements of the bank's deduction interface.

[0024] Preferably, the reader reading preference analysis module analyzes the reader's borrowing history and behavior data in the following specific steps:

[0025] Data collection: Using questionnaire survey and interview techniques, we collected multi-dimensional data from readers through questionnaires and interview outlines;

[0026] Data processing: Using decision tree algorithms and combining them with data preprocessing techniques, we process the collected raw data and build a decision tree model to explore readers' reading preferences and needs.

[0027] Personalized reading recommendations: Based on the analysis results of the decision tree model, combined with content-based recommendations and collaborative filtering recommendation algorithms, as well as popular book lists, we recommend books that match readers' reading preferences.

[0028] Optimize library resource construction and management strategies: By analyzing readers' borrowing behavior data, provide libraries with book procurement strategies and layout adjustment suggestions.

[0029] An intelligent self-service book borrowing and returning method, including a book borrowing process and a book returning process;

[0030] The specific steps of the book borrowing process are:

[0031] S11: The user completes identity authentication by performing facial recognition through the infrared dual-camera module set outside the operation interface;

[0032] S12: Scan books in batches using RFID to read the unique coding information in the book tags;

[0033] S13: Weight sensor verification, measuring the total weight of the books, comparing the number of books identified by RFID with the weight measured by the weight sensor to see if they match. If they match, the verification passes; if not, an alarm is triggered;

[0034] S14: Complete the borrowing after confirmation;

[0035] S15: Blockchain records borrowing information;

[0036] S16: If the damage detection system detects damage, the user is prompted to contact the staff for processing;

[0037] The specific steps of the book return process are:

[0038] S21: The user places the book into the scanning area;

[0039] S22: RFID identification bibliography;

[0040] S23: The damage detection system runs to check whether there is damage. If damage is detected, the staff will be contacted for processing;

[0041] S24: weight sensor calibration;

[0042] S25: Return completed;

[0043] S26: Blockchain records book return information;

[0044] S27: If the book is overdue, the operation interface will display the late payment fee information.

[0045] The present invention has the following beneficial effects:

[0046] 1. The present invention significantly improves the accuracy of book borrowing and returning operations by integrating RFID batch identification technology with a weight sensor dual verification mechanism, effectively preventing books from being left behind. At the same time, a multimodal damage detection system, combining cameras to scan covers / spines to identify torn graffiti and detect obvious damage, and a pressure sensor to detect abnormal page thickness, not only achieves a comprehensive inspection of the book's appearance and content, but also ensures the integrity of borrowed books.

[0047] 2. The present invention adopts a three-level identity authentication mechanism, including dynamic liveness detection, cloud comparison of facial features and ID photos, and optional real-time OTP verification on the mobile terminal, to build a multi-level security protection system, effectively eliminating the risk of borrowing books by impersonating others. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a system architecture diagram of an intelligent self-service book borrowing and returning system proposed by the present invention;

[0049] Figure 2 This is a flowchart of borrowing books in an intelligent self-service book borrowing and returning method proposed by the present invention;

[0050] Figure 3 The present invention provides a flowchart for returning books in an intelligent self-service book borrowing and returning method. DETAILED DESCRIPTION

[0051] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0052] like Figure 1-Figure 3 As shown, an intelligent self-service book borrowing and returning system includes:

[0053] User interaction layer: This layer provides an operation interface. This interface combines an infrared dual-camera module with a fill-light function for dynamic liveness detection (by analyzing facial micro-expressions and blinks to defend against photo or video attacks) and facial feature collection. It also integrates an audio-visual prompt system and an emergency physical switch. The audio-visual prompt system uses sound prompts and light indicators to guide readers through borrowing and returning items. The emergency physical switch switches to privacy protection mode with one click, disconnecting biometric feature collection.

[0054] Identity verification module: This module uses a three-level security mechanism: dynamic liveness detection (infrared liveness analysis), cloud-based face comparison (matching encrypted feature vectors with ID photos), and mobile OTP verification (triggering a one-time verification code when confidence is low). This mechanism is combined with local feature extraction processing on edge computing devices to reduce latency. Dynamic liveness detection uses dual infrared cameras to detect facial micro-expressions, blinking, and other liveness features to prevent photo or video attacks. Cloud-based face comparison compares the collected facial feature vectors with pre-stored ID photos for identity verification. When the recognition confidence falls below a threshold, a one-time verification code (OTP) is sent to the user's bound mobile phone to complete secondary verification.

[0055] Book Management Module: This module ensures operational accuracy through dual verification using RFID batch recognition (supporting simultaneous recognition of 50 book tags) and a weight sensor (monitoring weight changes and ensuring they match the RFID count to prevent missing books). It also integrates a damage detection system, where a camera scans the cover / spine to identify tears and graffiti, an AI algorithm detects obvious damage (such as tears and graffiti), and a pressure sensor detects abnormal page thickness (such as missing or folded pages).

[0056] Overdue automatic deduction module: monitors readers' book return status and automatically initiates the deduction process once an overdue payment is detected. Through Alipay account / bank card binding, RESTful API and bank deduction interface, the overdue amount is deducted from the Alipay account balance. If the balance is insufficient, the overdue amount is deducted from the bound bank card. After the deduction is successful, the reader receives a deduction notification including the deduction amount, the remaining balance, and other information, and the overdue record will be recorded;

[0057] Reader reading preference analysis module: This module uses questionnaires, interviews, and decision tree algorithms to analyze readers' borrowing history and behavior data, providing readers with personalized reading recommendations while optimizing the library's resource development and management strategies.

[0058] Data security and background management module: Privacy and data integrity are protected through local encrypted storage (only encrypted feature vectors are retained, and facial feature vectors are encrypted and stored on edge computing devices), blockchain evidence storage (all borrowing and returning records are timestamped and uploaded to a private chain, and timestamp records cannot be tampered with), and federated learning (each terminal device optimizes facial recognition through federated learning). It also supports automatic uploading of exception logs and blockchain traceability.

[0059] In one embodiment, the operation interface is tilted at an angle of 44°-46°, and the screen content is visible to the reader when the reader is operating.

[0060] In one embodiment, the infrared dual-camera module includes a black and white camera module, a color camera module and an infrared lighting unit, which are assembled on the same circuit board substrate. A dual-pass filter is arranged in front of the black and white camera module, which transmits visible light with a wavelength of 400nm to 650nm and infrared light with a wavelength of 700nm to 1000nm to capture facial images. An infrared cutoff filter is arranged in front of the color camera module to ensure that high-quality images can be obtained even in a visible light environment. The infrared light emitted by the infrared lighting unit has a wavelength range of 700nm to 1000nm, providing a fill light function to ensure clear imaging even in insufficient light conditions.

[0061] In one embodiment, the sound and light prompt system uses a sound and light alarm, which includes two parts: a sound prompt and a light indication. The sound prompt part issues a sound instruction through a speaker, such as "Please put the book down", "Book borrowed successfully", etc. The light indication part uses different colors of light (such as green for success and red for error) and flashing frequency to prompt the reader of the operation status.

[0062] The alarm circuit design adopts a single-chip microcomputer control system to realize the sound and light prompt function, among which: during the book borrowing process, the control system controls the LED to flash quickly, and the buzzer emits a short sound prompt; during the book returning process, the control system controls the LED to flash slowly, and the buzzer emits different sound prompts; when the book is detected to be overdue, the control system controls the LED to flash in a specific pattern, and the buzzer emits a continuous alarm sound.

[0063] In one embodiment, the specific steps of the book management module for identifying and detecting the status of books are as follows:

[0064] RFID batch identification: Each book is attached with an RFID tag, which stores the book's unique coding information. The reader places the book in the identifiable area of ​​the RFID reader, and the wireless radio frequency identification technology is used to read the unique coding information in the book tag;

[0065] Double verification of weight sensors: A weight sensor is placed on the operating platform of the book lending and returning machine to measure the total weight of the books. When books are placed on the operating platform, the sensor measures the total weight of the books and outputs an electrical signal. The output electrical signal is converted into a weight value. The number of books identified by RFID is compared with the weight measured by the weight sensor to see if they match. If they do not match, the verification passes; if not, an alarm is triggered, prompting staff to conduct manual verification.

[0066] Camera scanning and image recognition: Scan the book cover / spine with a camera to obtain image information, pre-process the collected image information, and use the improved Faster R-CNN to extract feature information in the image, such as tear locations and graffiti areas. Based on the extracted feature information, it is determined whether the book is damaged.

[0067] Pressure sensor detection: The pressure sensor is used to measure the pressure change of the book page on the sensor. Through data analysis, it is determined whether there is any abnormality in the thickness of the book page, such as missing pages, folds, etc., thus realizing a comprehensive detection of the appearance and internal status of the book;

[0068] Cross-validation and overall detection: The data from RFID identification, weight sensors, camera scanning, and pressure sensor detection are integrated, and cross-validation is performed by comparing and analyzing the data from different sensors and detection technologies. For example, the number of books identified by RFID is verified to be consistent with the weight measured by the weight sensor; whether there is a correlation between the damage identified by the image and the abnormal thickness of the book pages detected by the pressure sensor is verified, etc. The joint model is used to conduct a comprehensive analysis of image features (such as tear location) and pressure data (such as thickness changes).

[0069] In one embodiment, the overdue automatic deduction module uses the HTTPS protocol to establish a secure connection with the bank system and transmits parameters such as account ID, deduction amount, etc. according to the requirements of the bank's deduction interface.

[0070] In one embodiment, the reader reading preference analysis module analyzes the reader's borrowing history and behavior data in the following specific steps:

[0071] Data collection: Using questionnaire surveys and interviews, we collected multi-dimensional data on readers' borrowing history, reading types, reading frequency, etc.

[0072] Data processing: The decision tree algorithm is used in combination with data preprocessing technology to process the collected raw data. By building a decision tree model, readers' reading preferences and needs are explored. The most representative features are selected from the original features to construct a data set and the best split feature is selected. For discrete variables, information gain is used to select the best split feature. Assume that the information entropy H(D) of the data set D is calculated as follows: Among them, |Y| represents the number of classification categories, p k Represents the probability of the kth classification. For the feature A of the partitioned data set, information gain g(A) is used as the evaluation index, and its calculation formula is: Among them, V represents the number of values ​​of feature A, D v Represents the data sample set with the vth value in the selected feature A. Starting from the root node, the data set is divided into subsets according to the best splitting feature, and a subtree is recursively constructed. At each node, the data set is divided into two or more subsets according to the best splitting feature, and this process is recursively performed on each subset until the termination condition is met. The growth of the decision tree is stopped in advance during the growth process, and the decision tree is simplified after it is grown. Overfitting is avoided by deleting some subtrees or leaf nodes. The performance of the decision tree is evaluated using the test data set, and the evaluation indicators are calculated: the proportion of correctly classified samples to the total number of samples, the accuracy rate

[0073] The ratio of positive samples to the instances predicted as positive samples is the ratio of positive samples to the instances predicted as positive samples.

[0074] The ratio of true positive examples to all positive examples, and the recall rate is

[0075] The harmonic mean of precision and recall, F1 score is

[0076] Among them, TP (true positive), TN (true negative), FP (false positive), and FN (false negative) represent the number of samples of true positive, true negative, false positive, and false negative, respectively;

[0077] Adjust the decision tree based on the evaluation results. Once the decision tree is built, use the model to classify and predict new user feedback data to understand readers' preferences and attitudes towards books.

[0078] Personalized reading recommendations: Based on the analysis results of the decision tree model, combined with content-based recommendations and collaborative filtering recommendation algorithms, as well as popular book lists, we recommend books that match readers' reading preferences. The specific steps are: obtaining user profiles and literature information, identifying and removing data irrelevant to book recommendations, reasonably filling in missing values ​​in users' reading records, ratings, or comments, and converting users' reading behavior into a unified scale;

[0079] Extract text features from book titles, abstracts, content, or user reviews, convert them into numerical features, and construct user feature vectors based on user behavior;

[0080] Collaborative filtering: Constructing a user-book matrix: Based on the user's reading history and ratings, construct a user-book interaction matrix. Assuming there are m users and n books, the user-book matrix R can be expressed as:

[0081]

[0082] Among them, r ij Indicates the rating or reading times of book j by user i. If a user has not read a book, the corresponding element can be set to 0 or left blank.

[0083] Calculate similarity: Calculate the similarity between users with common reading preferences or between books with similar themes and content:

[0084]

[0085] Among them, I uv is the set of books that user u and user v have interacted with, r ui is the rating or some metric value of user u on book i, r viis the rating or metric of user v on book i. Based on the reading history and book similarity of similar users, a recommendation list is generated for the current user.

[0086] Content-based recommendation: Match the feature vector of the user's reading preference with the feature vector of the book and calculate the similarity: in, is the feature vector of user reading preference, is the feature vector of the book, which predicts the user's potential interest or satisfaction in the book that has not been read based on the content similarity;

[0087] Hybrid recommendation: Based on the actual application scenario and data characteristics, different weights are assigned to collaborative filtering and content-based recommendation. The scoring results of collaborative filtering and content-based recommendation are weighted and summed to obtain the final recommendation score.

[0088] Generate recommendations: Based on the final rating, select the book with the highest rating as the recommendation result;

[0089] Optimize library resource construction and management strategies: By analyzing readers' borrowing behavior data, the library is provided with book procurement strategies and layout adjustment suggestions, such as increasing the purchase volume of popular books and setting up thematic bookshelves, etc., which effectively optimizes the library's resource construction and management strategies and improves the library's service quality and resource utilization.

[0090] An intelligent self-service book borrowing and returning method, including a book borrowing process and a book returning process;

[0091] The specific steps of the book borrowing process are:

[0092] S11: The user completes identity authentication by performing facial recognition through the infrared dual-camera module set outside the operation interface;

[0093] S12: Scan books in batches using RFID to read the unique coding information in the book tags;

[0094] S13: Weight sensor verification, measuring the total weight of the books, comparing the number of books identified by RFID with the weight measured by the weight sensor to see if they match. If they match, the verification passes; if not, an alarm is triggered;

[0095] S14: Complete the borrowing after confirmation;

[0096] S15: Blockchain records borrowing information;

[0097] S16: If the damage detection system detects damage, the user is prompted to contact the staff for processing;

[0098] The specific steps of the book return process are:

[0099] S21: The user places the book into the scanning area;

[0100] S22: RFID identification bibliography;

[0101] S23: The damage detection system runs to check whether there is damage. If damage is detected, the staff will be contacted for processing;

[0102] S24: weight sensor calibration;

[0103] S25: Return completed;

[0104] S26: Blockchain records book return information;

[0105] S27: If the book is overdue, the operation interface will display the late payment fee information.

[0106] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent self-service book borrowing and returning system, characterized in that: include: User interaction layer: This layer provides an operation interface that uses an infrared dual-camera module for dynamic liveness detection and facial feature collection. It also integrates an audio-visual prompt system and an emergency physical switch. The audio-visual prompt system uses sound prompts and light indicators to guide readers through borrowing and returning operations. The emergency physical switch switches to privacy protection mode with one click, disconnecting biometric feature collection. Identity verification module: This module uses a three-level security mechanism – dynamic liveness detection, cloud-based face comparison, and mobile OTP verification – combined with local feature extraction processing on edge computing devices. Dynamic liveness detection uses dual infrared cameras to detect liveness features to prevent photo or video attacks. Cloud-based face comparison compares collected facial feature vectors with pre-stored ID photos for identity verification. When the recognition confidence level falls below a threshold, a one-time verification code is sent to the user's bound mobile phone to complete secondary verification. Book Management Module: This module integrates a damage detection system using RFID batch identification and weight sensor verification. A camera scans the cover / spine to identify torn or graffiti, detecting obvious damage, while a pressure sensor detects abnormal page thickness. Overdue automatic deduction module: monitors readers' book return status and automatically initiates the deduction process once an overdue payment is detected. Through Alipay account / bank card binding, RESTful API and bank deduction interface, the overdue amount is deducted from the Alipay account balance. If the balance is insufficient, the overdue amount is deducted from the bound bank card. After the deduction is successful, the reader receives a deduction notification and the overdue record is recorded. Reader reading preference analysis module: This module uses questionnaires, interviews, and decision tree algorithms to analyze readers' borrowing history and behavior data, providing readers with personalized reading recommendations while optimizing the library's resource development and management strategies. Data security and background management module: Ensure privacy and data integrity through local encrypted storage, blockchain evidence storage and federated learning, and support automatic upload of abnormal logs and blockchain traceability.

2. The intelligent self-service book borrowing and returning system according to claim 1, characterized in that: The operation interface is tilted at an angle of 44°-46°, and the screen content is visible to the reader when the reader is operating.

3. The intelligent self-service book borrowing and returning system according to claim 1, characterized in that: The infrared dual-camera module includes a black and white camera module, a color camera module and an infrared lighting unit, which are assembled on the same circuit board substrate. A dual-pass filter is arranged in front of the black and white camera module, which transmits visible light with a wavelength of 400nm to 650nm and infrared light with a wavelength of 700nm to 1000nm to capture facial images. An infrared cutoff filter is arranged in front of the color camera module, and the infrared lighting unit emits infrared light with a wavelength range of 700nm to 1000nm.

4. The intelligent self-service book borrowing and returning system according to claim 1, characterized in that: The sound and light prompt system adopts a sound and light alarm, which includes two parts: sound prompt and light indication. The sound prompt part sends out sound instructions through a speaker, and the light indication part prompts readers of the operation status through lights of different colors and flashing frequencies.

5. The intelligent self-service book borrowing and returning system according to claim 1, characterized in that: The specific steps of the book management module for identifying and detecting the status of books are as follows: RFID batch identification: Each book is attached with an RFID tag, which stores the book's unique coding information. The reader places the book in the identifiable area of ​​the RFID reader, and the wireless radio frequency identification technology is used to read the unique coding information in the book tag; Double verification of weight sensors: A weight sensor is placed on the operating platform of the book lending and returning machine to measure the total weight of the books. When books are placed on the operating platform, the sensor measures the total weight of the books and outputs an electrical signal. The output electrical signal is converted into a weight value. The number of books identified by RFID is compared with the weight measured by the weight sensor to see if they match. If they do not match, the verification passes; if not, an alarm is triggered, prompting staff to conduct manual verification. Camera scanning and image recognition: Scan the book cover / spine with a camera to obtain image information, pre-process the collected image information, and use the improved Faster R-CNN to extract feature information in the image. Based on the extracted feature information, determine whether the book is damaged; Pressure sensor detection: Use the pressure sensor to measure the pressure change of the book page on the sensor, and judge whether there is any abnormality in the thickness of the book page through data analysis; Cross-validation and overall detection: The data from RFID identification, weight sensors, camera scanning, and pressure sensor detection are integrated. By comparing and analyzing the data from different sensors and detection technologies, cross-validation is performed, and a joint model is used to comprehensively analyze image features and pressure data.

6. The intelligent self-service book borrowing and returning system according to claim 1, characterized in that: The overdue automatic deduction module uses the HTTPS protocol to establish a secure connection with the bank system and transmits parameters according to the requirements of the bank's deduction interface.

7. The intelligent self-service book borrowing and returning system according to claim 1, characterized in that: The specific steps of the reader reading preference analysis module analyzing the reader's borrowing history and behavior data are as follows: Data collection: Using questionnaire survey and interview techniques, we collected multi-dimensional data from readers through questionnaires and interview outlines; Data processing: Using decision tree algorithms and combining them with data preprocessing techniques, we process the collected raw data and build a decision tree model to explore readers' reading preferences and needs. Personalized reading recommendations: Based on the analysis results of the decision tree model, combined with content-based recommendations and collaborative filtering recommendation algorithms, as well as popular book lists, we recommend books that match readers' reading preferences. Optimize library resource construction and management strategies: By analyzing readers' borrowing behavior data, provide libraries with book procurement strategies and layout adjustment suggestions.

8. The intelligent self-service book borrowing and returning method corresponding to the intelligent self-service book borrowing and returning system according to claim 1 is characterized in that: Including the book borrowing process and the book returning process; The specific steps of the book borrowing process are: S11: The user completes identity authentication by performing facial recognition through the infrared dual-camera module set outside the operation interface; S12: Scan books in batches using RFID to read the unique coding information in the book tags; S13: Weight sensor verification, measuring the total weight of the books, comparing the number of books identified by RFID with the weight measured by the weight sensor to see if they match. If they match, the verification passes; if not, an alarm is triggered; S14: Complete the borrowing after confirmation; S15: Blockchain records borrowing information; S16: If the damage detection system detects damage, the user is prompted to contact the staff for processing; The specific steps of the book return process are: S21: The user places the book into the scanning area; S22: RFID identification bibliography; S23: The damage detection system runs to check whether there is damage. If damage is detected, the staff will be contacted for processing; S24: weight sensor calibration; S25: Return completed; S26: Blockchain records book return information; S27: If the book is overdue, the operation interface will display the late payment fee information.

Images