Book borrowing and returning management system based on social security card recognition

Abstract

The present invention relates to the technical field of book borrowing and returning management, and in particular to a book borrowing and returning management system based on social security card recognition. The present invention comprises a front-end module, a lending mark module, a return detection module and a determination module. The present invention can flexibly calculate the remaining preparation time according to the scheduled pickup time or departure time provided by the user, and intelligently select the optimal processing method, so as to balance the energy consumption when ensuring that the books are borrowed and allocated on time, and realize the efficient allocation planning of the book borrowing and returning management system.

Description

Technical Field

[0001] The invention relates to the technical field of book borrowing and returning management, and in particular to a book borrowing and returning management system based on social security card recognition. Background Art

[0002] In order to further promote the application of the "one-card" social security card for resident services, expand the application scope of social security cards in the field of "cultural experience", and improve the quality and efficiency of public cultural services, it is planned to implement the "Book Cabinet Application Service Project". The project aims to use physical social security cards and electronic social security cards as the main means of identity authentication to achieve intelligent, automated, networked and convenient management of book borrowing and returning, thereby significantly improving the convenience and intelligence of book borrowing and returning management services and creating a more superior cultural reading environment for readers.

[0003] The existing book cabinet borrowing and returning management system usually combines the user's account information during use so that the user can issue a request to borrow or return a book in the book cabinet, thereby realizing convenient book borrowing and returning management. However: When users return books, if the books are damaged, they must be repaired in a timely manner to ensure that the subsequent user experience is not affected; However, in actual operation, if the book is determined to be for repair but is booked for borrowing by a subsequent user before the repair is carried out, then if this situation occurs, how to intelligently select an appropriate repair method according to different situations and ensure that the book can be prepared at the time the user's booked borrowing time is available, so as to facilitate subsequent book borrowing and allocation planning, and improve the efficiency of the book borrowing and returning management system and user experience is a problem that needs to be solved at present. Summary of the invention

[0004] In view of the above-mentioned shortcomings of the prior art, the present invention provides a book borrowing and returning management system based on social security card recognition, which can effectively solve the problem that when returning wet books, the existing book borrowing and returning management system in the prior art may fail to flexibly select an appropriate drying method, which may cause delays in the time for subsequent users to obtain borrowed books.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: The present invention provides a book borrowing and returning management system based on social security card recognition. The system is arranged on a bookcase and includes a front-end module, a lending mark module, a return detection module, a determination module and a decision module, wherein: Loan marking module: After the user determines the book they need, it helps the user obtain the book and change the storage status of the book; The return detection module includes a data acquisition module, a 3D reconstruction module, a corner point matching module and a judgment module. The data acquisition module obtains the depth image of the books in the return detection area of ​​the book cabinet and extracts the depth information according to the depth image; the 3D reconstruction module converts the depth information into 3D point cloud data and generates a 3D model of the book return; the corner point matching module performs corner point matching to determine the difference area; the judgment module determines the type and degree of damage according to the characteristics of the difference area; Function of the decision-making module: When the humidity of the returned book exceeds the preset value, it is judged as a repairable book, and the subsequent user reservation information in the bookcase server is collected to determine the required books of the subsequent users. When the required books are repairable books, the processing time for the repairable books is obtained, and the remaining preparation time is obtained according to the difference between the reservation information and the current time, and the corresponding processing method is selected according to the remaining preparation time.

[0006] Furthermore, the front-end module also includes a login unit and a query unit, wherein the login unit is used to help users access the book cabinet server, and the query unit is used for users to query required books in the book cabinet server. It also includes a binding unit and a verification login unit. The binding unit can help users associate and bind their accounts on the book cabinet server with the user's own social security card account, electronic social security card, Alipay and WeChat, and all the above accounts after binding can access the book cabinet server through the verification login unit.

[0007] Furthermore, the lending marking module also includes an RFID recognition unit, which reads the RFID code of the book and marks the RFID code on the book as a lending state.

[0008] Furthermore, the data acquisition module also includes a conventional camera component and an image preprocessing unit. The conventional camera component is composed of multiple RGB-D cameras. When the user places a book in the detection area of ​​the bookcase, the RGB-D camera shoots the book from multiple angles to obtain a depth image of the book. The image preprocessing unit segments the depth image of the book based on the deep learning model GrabCut. According to the segmentation result, the overall appearance of the book is extracted to obtain a valid depth image. The valid depth image is loaded, the depth information of the effective depth image is obtained, and the depth value of each pixel point of the effective depth image is determined in turn.

[0009] Furthermore, the 3D reconstruction module converts the depth value of each pixel of the effective depth image into 3D point cloud data according to the depth value of each pixel of the effective depth image through the KinectFusion 3D reconstruction algorithm, and filters and downsamples the 3D point cloud data to remove noise and redundant information to obtain effective 3D point cloud data, and generates a 3D model of the book when it is returned by converting the effective 3D point cloud data into the 3D point cloud data.

[0010] Furthermore, the corner matching module extracts the corner features of the three-dimensional model of the book when borrowed and returned respectively based on the Harris corner detection algorithm, and performs SIFT descriptor calculation on the corner features, and matches and compares the corner points of the three-dimensional model of the book when returned with the corner points of the three-dimensional model of the book when borrowed based on the FLANN feature matching algorithm, determines the matching point pairs, calculates the difference area between the matching point pairs, and generates a difference area image.

[0011] Furthermore, the determination module is based on the convolutional neural network algorithm, and determines the type and degree of damage according to the features of the difference area image. The confidence of each damage type is calculated for the features of the difference area image, and the damage type is determined according to the confidence. The damage types include cover damage, spine damage, and back cover damage. The determination module segments the difference area image based on the MaskR-CNN deep learning model, identifies the damaged area and the normal area, extracts the contour of the damaged area, and calculates the actual area of ​​the damaged area according to the following relationship:

[0012] Where Areacm is the actual area of ​​the damaged area, Areaapixels is the area of ​​the damaged area, and DPI is the resolution of the image; Sum the depth values ​​of each point in the damaged area and divide it by the number of points in the damaged area to obtain the average depth value Ddamaged of the damaged area; Sum the depth value of each point in the normal area and divide it by the number of points in the normal area to obtain the average depth value Dnormal of the normal area; The depth change value is calculated according to the following formula: ΔD=Ddamaged-Dnormal Where ΔD is the depth change value; The damage index is calculated according to the following relationship: DI = α × Area cm + β × ΔD Where Areacm is the actual area of ​​the damaged area, ΔD is the depth change value, DI is the damage index, and α and β are the corresponding weight coefficients respectively; When the damage index DI is less than the preset value, the book is judged as qualified; when the damage index DI is greater than or equal to the preset value, the book is judged as unqualified.

[0013] Furthermore, the page images of qualified books are collected, damaged books or safe books are determined based on the page images, the humidity of safe books is detected to obtain the humidity of safe books, and healthy books or repairable books are determined according to the humidity threshold.

[0014] Compared with the known prior art, the technical solution provided by the present invention has the following beneficial effects: The binding unit helps users to associate and bind their accounts on the book cabinet server with their social security card accounts, electronic social security cards, Alipay and WeChat. After binding, all of the above accounts can access the book cabinet server through the verification login unit, realizing convenient book borrowing and returning management.

[0015] Flexible calculation of remaining preparation time based on the scheduled pickup time or departure time provided by the user, intelligent selection of the optimal drying method, intelligently balancing energy consumption and efficiency while ensuring that the books are ready on time; In addition, the drying method based on the intelligent allocation of books can make a comprehensive judgment based on the user's actual book-taking conditions in the future, so that the book cabinet can normally allocate repairable books based on the user's book-taking conditions, ensure the normal flow of data, and form an efficient resource allocation plan for books. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings required to be based on the embodiments or the prior art description are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 It is the overall module block diagram of the present invention. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] The present invention will be further described below in conjunction with the embodiments.

[0020] Example 1 (see Figure 1 ): A book borrowing and returning management system based on social security card recognition, which is set on the book cabinet and includes a front-end module, a borrowing mark module, a return detection module, a determination module and a decision module, among which: The front-end module also includes a login unit and a query unit, wherein the login unit is used to help users access the book cabinet server, and the query unit is used for users to query required books in the book cabinet server. It also includes a binding unit and a verification login unit. The binding unit can help users associate and bind their account on the book cabinet server with the user's own social security card account, electronic social security card, Alipay and WeChat, and all the above accounts after binding can access the book cabinet server through the verification login unit.

[0021] It is worth noting that users can choose to bind multiple accounts to the same library account. Users can choose their usual identity authentication method to access the library server without having to rely on the social security card every time.

[0022] The lending mark module further includes an RFID recognition unit, which reads the RFID code of the book and marks the RFID code on the book as a lending state.

[0023] Loan marking module: After the user determines the book they need, it helps the user obtain the book and change the storage status of the book; In the process of borrowing books, the user first determines the required book to be borrowed, and the lending mark module obtains the RFID code corresponding to the required book in the library server, obtains the storage coordinates of the required book in the library server according to the RFID code corresponding to the book, and verifies whether the book at this coordinate is the required book through the RFID identification unit corresponding to the storage coordinate. After the verification is passed, the storage status of the required book is changed from in-stock to borrowed, and the cabinet door corresponding to the storage coordinate is controlled to open through the lending mark module, so that the user can obtain the required book; Through the setting of the RFID identification unit, users only need to place the target book within the RFID identification range, and the system can automatically read the RFID code of the book and update the borrowing status. Compared with the traditional manual borrowing method (such as scanning barcodes), it is faster and easier to operate. The borrowing marking module can also determine the storage coordinates of the required book in the bookcase after the user selects the required book, and open the cabinet door corresponding to the storage coordinates, so that the user can take the book by himself.

[0024] The return detection module includes a data acquisition module, a three-dimensional reconstruction module, a corner point matching module and a judgment module.

[0025] Data acquisition module: when borrowed books are returned, the depth image of the books in the book return detection area of ​​the book cabinet is obtained, and the depth information is extracted according to the depth image; 3D reconstruction module: converts depth information into 3D point cloud data and generates a 3D model of book return; corner matching module: performs corner matching through the corner matching module to determine the difference area; the judgment module determines the type and degree of damage based on the characteristics of the difference area.

[0026] Specifically, the data acquisition module also includes a conventional camera component and an image preprocessing unit. The conventional camera component is composed of multiple RGB-D cameras. When the user places a book in the detection area of ​​the bookcase, the RGB-D camera shoots the book from multiple angles to obtain a depth image of the book. The image preprocessing unit segments the depth image of the book based on the deep learning model GrabCut. According to the segmentation result, the overall appearance of the book is extracted to obtain a valid depth image. The valid depth image is loaded, the depth information of the effective depth image is obtained, and the depth value of each pixel point of the effective depth image is determined in turn.

[0027] It is worth noting that the depth value refers to the gray value of each pixel in the effective depth image, which can be directly expressed as the depth value of the point. When the user returns the book, the user first places the book in the detection area of ​​the bookcase, and the RFID sensor reads the RFID tag of the book to confirm that the book has been placed in the detection area; After determining that there are books in the detection area of ​​the bookcase, the RGB-D camera is used to capture the depth image of the book to obtain the depth information of the book. The deep learning model GrabCut is used to segment the obtained depth image of the book, extract the overall appearance of the book and record it as the effective depth image.

[0028] The 3D reconstruction module converts the depth value of each pixel of the effective depth image into 3D point cloud data through the KinectFusion 3D reconstruction algorithm, and filters and downsamples the 3D point cloud data to remove noise and redundant information to obtain effective 3D point cloud data, and generates a 3D model of the book when it is returned based on the effective 3D point cloud data.

[0029] The corner matching module extracts the corner features of the three-dimensional model of the book when borrowing and returning respectively based on the Harris corner detection algorithm, and performs SIFT descriptor calculation on the corner features. Based on the FLANN feature matching algorithm, the corner points of the three-dimensional model of the book when returned are matched and compared with the corner points of the three-dimensional model of the book when borrowing, the matching point pairs are determined, the difference area between the matching point pairs is calculated, and the difference area image is generated.

[0030] It is worth noting that the Harris corner point detection calculation formula is as follows:

[0031] in, is the determinant of the matrix M, is the trace of the matrix M, k is an empirical constant, is the corner point response value; The structure tensor matrix M is defined as:

[0032] in, is the gradient of the image in the x and y directions, Weights the local area for the Gaussian window function.

[0033] By image gradient and , and construct the structure tensor matrix M, calculate the corner point response value R by combining the determinant and the trace, when the point with a large R value represents a corner point (R>0 and large), the final corner point is screened by non-maximum suppression; The SIFT descriptor calculation formula is as follows:

[0034] in, is the gradient amplitude, reflecting the local feature strength, are the image gradients in the neighborhood of the key points respectively;

[0035] in, is the gradient direction, which is used to count the direction distribution; the gradient amplitude and direction are calculated in the area around the key point, the area is divided into sub-blocks, the gradient direction histogram of each sub-block is counted, the histograms of all sub-blocks are combined to generate a 128-dimensional vector, and normalized to improve illumination invariance; The FLANN feature matching algorithm relationship is as follows:

[0036] Among them, d is the distance between descriptors, reflecting the similarity. They are two feature descriptors to be matched (such as 128-dimensional vectors of SIFT); like , then accept the match, build an index for the feature descriptor, quickly find candidate matches for each descriptor based on approximate nearest neighbor search (ANN), filter out incorrect matches through distance ratio test, retain high confidence matching point pairs, Harris corner detection determines the corner position through local gradient statistics, SIFT descriptor characterizes local texture through gradient direction histogram, and FLANN matching selects correct matching point pairs through efficient search and ratio test.

[0037] The judgment module is based on the convolutional neural network algorithm. It determines the type and degree of damage according to the features of the difference area image, calculates the confidence of each damage type based on the features of the difference area image, and determines the damage type based on the confidence. The damage types include cover damage, spine damage, and back cover damage. The judgment module segments the difference area image based on the MaskR-CNN deep learning model, identifies the damaged area and the normal area, extracts the contour of the damaged area, and calculates the actual area of ​​the damaged area according to the following relationship:

[0038] Where Areacm is the actual area of ​​the damaged area, Areaapixels is the area of ​​the damaged area, and DPI is the resolution of the image; Sum the depth values ​​of each point in the damaged area and divide it by the number of points in the damaged area to obtain the average depth value Ddamaged of the damaged area; Sum the depth value of each point in the normal area and divide it by the number of points in the normal area to obtain the average depth value Dnormal of the normal area; The depth change value is calculated according to the following formula: ΔD=Ddamaged-Dnormal Where ΔD is the depth change value; The damage index is calculated according to the following relationship: DI = α × Area cm + β × ΔD Where Areacm is the actual area of ​​the damaged area, ΔD is the depth change value, DI is the damage index, and α and β are the corresponding weight coefficients respectively; When the damage index DI is less than the preset value, the book is judged as qualified. When the damage index DI is greater than or equal to the preset value, the book is judged as unqualified. Unqualified books can be scrapped.

[0039] Collect the page images of qualified books, and judge whether they are damaged books (damaged books can be scrapped) or safe books. It is worth noting that the method of collecting the page images of qualified books includes but is not limited to collecting the page images inside the qualified books based on the cooperation of automatic page turners and high-speed scanners, and unifying the size, normalizing the pixel values, denoising and correcting the deformation of the collected page images through the Autoencoder model. The encoder compresses the input page images into low-dimensional potential features (such as 128-dimensional vectors), extracts key information (such as text layout, edge structure), and calculates the mean square error through the relational formula :

[0040] in, The original image pixel values, To reconstruct the image pixel values, , if the mean square error If the value exceeds this threshold, the page is judged to be damaged, and valid and invalid pages are obtained. Represents the total number of pixels in the image; A threshold number of valid pages corresponding to a book is set. If the valid pages of a qualified book reach the threshold number of valid pages, it is recorded as a repairable book (usually the pages are not damaged and can be used again). Otherwise, it is scrapped. Furthermore, the humidity of the repairable books is detected. When the humidity of the returned qualified books exceeds the preset value, they are determined to be repairable books (the repairable books need to be dried before the subsequent distribution). The decision module collects the subsequent user reservation information in the bookcase server and determines the required books of the subsequent users. When the required books are repairable books, the processing time of different processing methods for the repairable books is determined. The remaining preparation time is obtained according to the difference between the reservation information and the current time, and the corresponding processing method is selected based on the remaining preparation time, as follows: Treatment methods include equipment drying and natural air drying: Natural air drying, based on the evaporation rate method to predict the estimated time of natural air drying; According to the formula:

[0041] In the formula, Estimated time for natural drying, is the initial humidity of the book, is the target humidity, is the ambient humidity, is the evaporation coefficient.

[0042] During equipment drying: The drying rate is predicted according to the linear regression model, and the drying temperature, ambient relative humidity, and air flow velocity are used as characteristic factors to calculate the drying rate. The specific expression of the linear regression model is not described in detail. The estimated time for equipment drying is based on the following relationship:

[0043] in, Estimated time for drying the equipment, is the drying rate.

[0044] The repairable books, pickup time, departure time and home address of subsequent users are determined based on the reservation information, and the handling method of the repairable books is determined according to the following method: When the subsequent user cannot determine the departure time but can only determine the scheduled pickup time, the estimated device drying time and natural air drying time corresponding to the repairable book under device drying and natural air drying are obtained respectively, and the difference between the current time of the repairable book and the scheduled pickup time registered by the subsequent user is calculated to obtain the remaining preparation time; If the subsequent user cannot determine the appointment pickup time and can only determine the departure pickup time, the subsequent user's home address is recorded as the initial coordinates, and the bookcase address is recorded as the destination coordinates. The best path from the initial coordinates to the destination coordinates is obtained according to the Dijkstra algorithm, and the corresponding predicted journey time is determined according to the best path. The predicted journey time at this time is regarded as the remaining preparation time. By limiting an optimal and fastest path, the fastest predicted journey time for the subsequent user can be determined, so as to accurately select the processing method for the subsequent time. When the remaining preparation time is ≤ the estimated time for natural air drying, the repairable books are dried by equipment; when the remaining preparation time is greater than the estimated time for natural air drying, the repairable books are dried by natural air drying.

[0045] In this way, the corresponding actual processing method can be determined based on the subsequent users' demand for repairable books and the actual processing of the repairable books, ensuring that the repairable books can be correctly allocated to subsequent users and facilitate subsequent users to borrow them.

[0046] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the embodiments of the present invention.

Claims

1. A book borrowing and returning management system based on social security card recognition, the system is set on a bookcase, and is characterized in that: It includes a front-end module, a lending mark module, a return detection module, a determination module and a decision module, among which: The lending mark module marks the borrowed books; The return detection module includes a data acquisition module, a three-dimensional reconstruction module, a corner point matching module and a judgment module; The data acquisition module acquires the depth image of the book in the book return detection area of ​​the book cabinet when the borrowed book is returned, and extracts the depth information according to the depth image; The 3D reconstruction module converts the depth information into 3D point cloud data and generates a 3D model of the book return; Corner matching module, which performs corner matching to determine the difference area; The determination module determines the type and extent of damage based on the characteristics of the difference area; The decision-making module determines that a returned book is a repairable book when the humidity of the returned book exceeds a preset value, collects the reservation information of subsequent users in the bookcase server, determines the required books of subsequent users, and obtains the processing time of the repairable book by the processing method when the required book is a repairable book. The remaining preparation time is obtained according to the difference between the reservation information and the current time, and the corresponding processing method is selected according to the remaining preparation time.

2. The book borrowing and returning management system based on social security card recognition according to claim 1 is characterized in that: The front-end module also includes a login unit and a query unit, wherein the login unit is used to help users access the book cabinet server, and the query unit is used for users to query required books in the book cabinet server. The front-end module also includes a binding unit and a verification login unit.

3. The book borrowing and returning management system based on social security card recognition according to claim 1 is characterized in that: The lending mark module further includes an RFID recognition unit, which reads the RFID code of the book and marks the RFID code on the book as a lending state.

4. The book borrowing and returning management system based on social security card recognition according to claim 1 is characterized in that: When the user places a book in the detection area of ​​the bookcase, a depth image of the book is taken. The image preprocessing unit segments the depth image of the book based on the deep learning model GrabCut to obtain an effective depth image, obtains the depth information of the effective depth image, and determines the depth value of each pixel.

5. The book borrowing and returning management system based on social security card recognition according to claim 4 is characterized in that: The 3D reconstruction module converts the depth value into 3D point cloud data, and pre-processes the 3D point cloud data to obtain effective 3D point cloud data, and generates a 3D model of the book based on the effective 3D point cloud data.

6. The book borrowing and returning management system based on social security card recognition according to claim 5 is characterized in that: The corner point matching module extracts corner point features of the three-dimensional model of the book when borrowed and returned respectively based on the Harris corner point detection algorithm, performs SIFT descriptor calculation on the corner point features, matches and compares the corner points of the three-dimensional model of the book when returned with the corner points of the three-dimensional model of the book when borrowed, determines the matching point pairs, calculates the difference area between the matching point pairs, and generates a difference area image.

7. The book borrowing and returning management system based on social security card recognition according to claim 6 is characterized in that: The determination module determines the damage type and degree based on the convolutional neural network algorithm, calculates the features of the difference area image to obtain the confidence of each damage type, determines the damage type according to the confidence, identifies the damaged area and the normal area, extracts the contour of the damaged area, and calculates the actual area of ​​the damaged area: Sum the depth values ​​of each point in the damaged area and divide it by the number of points in the damaged area to obtain the average depth value Ddamaged of the damaged area; Sum the depth value of each point in the normal area and divide it by the number of points in the normal area to obtain the average depth value Dnormal of the normal area; Calculate the damage index: DI = α × Area cm + β × ΔD Where Areacm is the actual area of ​​the damaged area, ΔD is the depth change value, DI is the damage index, and α and β are the corresponding weight coefficients respectively; When the damage index DI is less than the preset value, the book is judged as qualified; when the damage index DI is greater than or equal to the preset value, the book is judged as unqualified.

8. The book borrowing and returning management system based on social security card recognition according to claim 7 is characterized in that: The page images of qualified books are collected to determine whether the books are damaged or safe, the humidity of the safe books is detected to obtain the humidity, and the healthy books or the repairable books are determined according to the humidity threshold.

9. The book borrowing and returning management system based on social security card recognition according to claim 8 is characterized in that: The decision module can process the repairable books, including equipment drying and natural air drying. In the natural air drying, the estimated time of natural air drying is predicted based on the evaporation rate method: ； In the formula, Estimated time for natural drying, is the initial humidity of the book, is the target humidity, is the ambient humidity, is the evaporation coefficient; During equipment drying: Drying rate was predicted based on linear regression model; The estimated time for equipment drying is based on the following relationship: ； in, Estimated time for drying the equipment, is the drying rate.

10. The book borrowing and returning management system based on social security card recognition according to claim 9 is characterized in that: Obtain reservation information, and determine the subsequent user's required books, pickup time, departure time and home address based on the reservation information. When the required books are repairable books, the following steps are also included: When the subsequent user cannot determine the departure time but can only determine the scheduled pickup time, the estimated time for equipment drying and natural air drying of the required books are calculated respectively, and the difference between the current time of the required books and the scheduled pickup time registered by the subsequent user is calculated to obtain the remaining preparation time; When the subsequent user cannot determine the scheduled pickup time but can only determine the departure pickup time, the subsequent user's home address is recorded as the initial coordinates, and the bookcase address is recorded as the destination coordinates. The optimal path from the initial coordinates to the destination coordinates is obtained according to the Dijkstra algorithm, and the corresponding predicted journey time is determined according to the optimal path. The predicted journey time is regarded as the remaining preparation time. When the remaining preparation time is less than or equal to the estimated natural drying time, use equipment drying method; When the remaining preparation time is greater than the estimated natural drying time, natural drying should be adopted.

Images