TWI423154B - Book Inventory Management System and Its Method - Google Patents

Description

Book inventory management system and method thereof

The present invention relates to an inventory management technology, and more particularly to a vision-based book inventory management system and method thereof.

In the process of opening or opening a library or bookstore, there must be cases of lending, selling, returning, purchasing, loading, losing, etc. After a period of time, the number of books is bound to change; in addition, the library or The books in the bookstore will be scattered in the corners of the hall (shop) due to the user's reading and incorrect homing, and the placement of the book may also change. Therefore, in order to maintain a large amount of library materials, it is often necessary to pay a relatively high personnel cost; for example, to maintain the current data of all books by means of manual inventory.

In order to automate the management of inventories, a conventional inventory system, such as the "inventory system for the use of radio frequency (RFID) tags), as disclosed in U.S. Patent No. 6,693,539, is incorporated herein by reference. A passive RFID tag is installed on the item (book), and then the RFID reader and the RFID scanner are used to establish and update the circulation status of the item (book), and the inventory database is used to store and provide the circulation status of the book for tracking and management.

Although the use of RFID technology for book management can automatically manage the inventory of books and reduce the personnel costs required. However, due to the nature of the radio signal, its signal strength is easily affected by the following factors: the absorption and shadowing effects of radio signals caused by obstacles in the room, and the interference of various electronic devices in the room affect the radio signals. Stability. Moreover, as the profit from the sale of books is decreasing year by year, the installation of passive RFID tags on each book is equivalent to increasing the cost of each book, and the related hardware devices need to be set up in the library (shop). Not economically viable.

Accordingly, it is an object of the present invention to provide a book inventory management system.

Therefore, the book inventory management system of the present invention is adapted to be connected to a photographing device, the system comprising a memory unit, an image processing unit, and an inventory data management unit. The memory unit stores at least one inventory data set that has been established, and the inventory data group includes a back image of an inventory book. The image processing unit is electrically connected to the memory unit, and the image processing unit includes a cutting module and a matching module, and the cutting module is configured to receive a frame of the bookshelf captured by the camera device, and The frame performs image cutting to obtain a back image of the book on the bookshelf. When the system is in an inventory mode, the comparison module is used in the memory unit to perform a book about the book on the shelf. The comparison between the back image of the book and the back image of the book of the stock book. The inventory data management unit is electrically connected to the memory unit and the image processing unit. When the system is in the inventory mode, the inventory data management unit is configured to generate a book on the shelf according to the comparison result of the comparison module. An inventory status and store it to the memory unit.

Another object of the present invention is to provide a book inventory management method.

Thus, the book inventory management method of the present invention is adapted to be executed by a system connected to a photographing apparatus, the method comprising the following steps. A) receiving one of the books to identify the information; B) initializing the inventory status of the book; C) receiving a frame of the bookshelf captured by the camera; D) performing image cutting on the frame to obtain the a back image of the book on the shelf; and E) establishing an inventory data group corresponding to the book, wherein the inventory data group includes the identification information, the inventory status, and the back image of the book.

Still another object of the present invention is to provide a book inventory management method.

Thus, the book inventory management method of the present invention is adapted to be executed by a system connected to a photographing apparatus, the method comprising the following steps. A) providing at least one inventory data set that has been established, wherein the inventory data set includes a back image of an inventory book; B) receiving a frame of a book shelf captured by the camera device; C) the book shelf The frame is image-cut to obtain a back image of a book on the shelf; D) using the back image of the book on the shelf to compare with the back image of the book; and E) The result of the comparison of step D) produces an inventory status of the books on the shelf.

The above and other technical contents, features and advantages of the present invention will be apparent from the following detailed description of the preferred embodiments.

Referring to FIG. 1, a preferred embodiment of the book inventory management system 1 of the present invention is connected to a plurality of camera devices 2 disposed near the bookshelf via a network, wherein the camera device 2 is a PTZ (pan, tilt, zoom) network. The number of the photographing devices 2 is related to environmental parameters in the library or bookstore (for example, the size of the indoor space, the number of bookshelves).

The book inventory management system 1 of the present invention comprises a human machine interface 11, a parameter processing unit 12, a memory unit 13 electrically connected to the human machine interface 11 and the parameter processing unit 12, and an image processing electrically connected to the memory unit 13. The unit 14 is electrically connected to the memory unit 13 and the inventory data management unit 15 of the image processing unit 14. The human machine interface 11 is used for the user to input and set system parameters. The parameter processing unit 12 is configured to process parameters related to the book inventory management system 1 and the photographing device 2. The image processing unit 14 is configured to receive an image captured by the photographing device 2 and perform image processing related to subsequent inventory management. The image processing unit 14 includes a geometric correction module 141, a cutting module 142, and a The feature capture module 143 and a comparison module 144. The inventory data management unit 15 is configured to perform corresponding book inventory management according to the processing result of the image processing unit 14, and store the result in the memory unit 13.

In the preferred embodiment, the human machine interface 11 includes a peripheral input device (for example, a keyboard, a mouse, not shown) for the user to operate and input data, and a scanner for reading the barcode of the book. (not shown). The parameter processing unit 12, the image processing unit 14, and the inventory data management unit 15 are integrated into a software and installed in a computer to execute the book inventory management method of the present invention. The memory unit 13 is a database or other storage device.

The book inventory management method of the present invention can be divided into the following three stages for explanation.

System initialization phase

Referring to FIG. 1, FIG. 2 and FIG. 3, before the book inventory management system 1 starts the relevant steps of the shelf stage and the inventory stage, the required parameters (groups) must be set because of the setting of each camera 2 Since the operation is substantially the same as that of the operation, the single imaging device 2 will be described below as an example.

First, since the photographing device 2 is preliminarily mounted at a known position, one of the photographing devices 2 has a camera identification information (CID) as a preset information; and the photographing range of the photographing device 2 is also set up in the photographing device 2 It is fixed afterwards, so the number of rows and the number of columns related to the number of frames of the bookshelf in the shooting range can be predicted. The column and row in which each frame 3 is located is defined as one. For example, FIG. 3 is all the frame 3 of the bookshelf in the shooting range of the photographing device 2, and the corresponding column number is 3, the number of rows is 4, and the book position information of each frame 3 can be expressed as (column i , line j ). The photographic device identification information and the book position information are used for subsequently generating a book location information of one of the books. In step S11, the user sets the photographic device by using the peripheral input device of the human interface 11 The information and the bookshelf location information are identified and stored in the memory unit 13.

Then, in step S12, the user moves the lens of the photographing device 2 to the bookshelf position at four corners of the shooting range thereof (for example, corresponding to (1, 1), (1, 4), (3, 1 in FIG. 3). And (3, 4) the four bookshelf positions of the frame 3), the parameter processing unit 12 obtains one of the bookshelf positions corresponding to each corner and a plurality of control point coordinates and stores them in the Inside the memory unit 13. In the preferred embodiment, each frame 3 corresponds to four control point coordinates of the real world coordinate system and four control point coordinates of the imaging coordinate system. The details of the PTZ parameter set and control point coordinates are well known to those skilled in the art and will not be described here.

Finally, in step S13, the parameter processing unit 12 calculates one of the PTZ parameter sets and the plurality of control point coordinates of each of the other bookshelf positions in the shooting range of the photographing device 2 by using the interpolation method, and stores them in the Inside the memory unit 13.

It is worth mentioning that the steps of the system initialization phase do not need to be performed frequently, and only need to be re-executed when the position of the bookshelf or the photographing device 2 is moved.

Shelf stage

Referring to FIG. 1 , FIG. 4 , and FIG. 5A to FIG. 5C , after the system is initialized, the user can switch the system to a shelf mode through the peripheral input device of the human interface 11 to start the related processing in the shelf stage. The shelf stage includes the following steps.

As shown in step S201, the user reads the bar code of each book to be put up by the scanner of the man-machine interface 11, and each bar code corresponds to the identification information of a book. Generally, the identification information of the book It is the International Standard Book Number (ISBN). The book inventory management system 1 receives the identification information of each book to be put on the shelf and stores it in the memory unit 13.

Then, the user puts the books to be put on shelves according to a preset shelf order.

As shown in step S202, the inventory management unit 15 initializes an inventory status of each of the books that have been placed. In the preferred embodiment, the initialized inventory status is "on-state".

As shown in step S203, the inventory management unit 15 forms the book location information of each book that has been put on the shelf by the camera identification information, the shelf location identification information, and the shelf order.

As shown in step S204, the parameter processing unit 12 provides a PTZ parameter set stored in advance in the memory unit 13 to control the photographing device 2 through the network to capture the frame 3 of the bookshelf.

Since the book inventory management system 1 processes the same frame 3 of the bookshelf captured by each of the photographing devices 2, for the sake of convenience of description, in the subsequent description, one of the frames of the bookshelf is used. Be explained.

As shown in step S205, because the photographing device 2 captures the relationship between the position and the angle, the content of the captured frame 3 is skewed (eg, FIG. 5A), so the image geometric correction must be performed on the captured frame 3. . The geometric correction module 141 of the image processing unit 14 performs image geometric correction on the frame 3 by using a pre-established frame geometry transformation matrix, wherein the frame geometry transformation matrix is framed in the memory unit 13 3 corresponding to the control point coordinates of a pre-established, wherein, with respect to the frame geometric transformation method for finding based matrices according to the prior art technique, reference may be "" Digital Image Processing Using MATLAB ", RC Gonzalez and RE Woods, 3 rd Edition, Prentice Hall. Therefore, I will not repeat them here.

As shown in step S206, the cutting module 142 of the image processing unit 14 performs image cutting on the corrected frame 3 ' (eg, FIG. 5B) to obtain a book back image of the book on the shelf; wherein the image cutting includes The following substeps.

a) Perform vertical edge detection on the corrected frame 3 ' and perform a histogram projection in the horizontal direction.

b) judging by using the execution result of a), if the result of the histogram projection is greater than a preset threshold, it indicates that it is the title of the book, and in this way, find an upper boundary of the title range. And the next boundary.

c) Perform horizontal edge detection on the corrected frame 3 ' and project the edged results in a horizontal histogram.

d) using the execution result of c), and matching the upper and lower boundaries of the identified title range, finding the possible bookshelf edge 31, and then performing image cutting on the bookcase edge 31 to obtain as shown in FIG. 5C. Box 3 " .

Among them, the vertical and horizontal edge detection in a) and c) above are Sobel edge detection.

Before proceeding with the description of the sub-steps below, make the following definitions for easy understanding. A state transition model is defined. The state transition model is a finite state automation having six distinct state categories, namely a left boundary (represented by q ₁ ) and a right book edge (right Boundary, denoted by q ₂ ), left spine (represented by q ₃ ), right spine (represented by q ₄ ), title (indicated by q ₅ ), and background (background, denoted by q ₆ ) , organized into a set of classes _{Q = {q 1, ...,} q 6}, the state s frame 3 'positions of a plurality of points 32 on a central horizontal line _i-i respectively correspond to the above-described state wherein one category, And the most likely top position of the book placement structure is denoted by p _i .

e) Using the state transition model in conjunction with the use of a dynamic programming algorithm to obtain a book back image of a book on a shelf.

For a description of the state transition model and the dynamic programming algorithm, refer to "A model-based book boundary detection technique for bookshelf image analysis." by E. Taira et al., and "Applied" by R. Bellman et al. Dynamic Programming.".

It is worth mentioning that the present invention further adds a plurality of hue function values corresponding to the position point i as the weight of the dynamic programming algorithm, and the hue average value and state of each hue function value and its corresponding position point. At least one of them is related, and the definition of the hue function d _T [( s _i , p _i )|( s _{i -1} , p _{i -1} )] is as shown in Equation 1.

Wherein, d _H ( H _i , H _{i -1} ) is as defined in Formula 2.

d _H ( H _i , H _{i -1} )=1-cos(2×π×( H _i - H _{i -1} )).................... . . . Equation 2 and H _i is the hue average of the pixels on the line from the position point i to the top position p _i .

As shown in step S207, the geometric correction module 141 of the image processing unit 14 further cuts the back image of the book cut in step S206 by using a back geometry conversion matrix, where the image geometric correction is performed. Similar to step S205; the difference is only that: according to the four boundaries of the back image of the book cut out e), the coordinates of the four book back control points are obtained; and then according to the coordinates of the four control points of the real world coordinate system, and The book back control point coordinates are used to find the book back geometry transformation matrix.

The back image of the book cut out in step S206 may still be misjudged due to the following three situations, so the present invention further post-processes the back image to obtain a more accurate cutting result.

Case 1, non-bookback images (eg, gaps between books, bookcase compartments) are misjudged into the back of the book; situation 2, the title of the book back is too small and misjudged as the background; and the situation 3, many books misjudged as one Ben, this situation often appears in the same series or multiple books with the same color on the back of the book.

As shown in step S208, in accordance with the above three cases, the cutting module 142 of the image processing unit 14 performs post-processing on the corrected book back image to further cut the corrected book back image. Among them, the post-processing includes the following sub-steps.

f) Perform horizontal edge detection on the back image of the book, and project the edged result into a histogram projection in the vertical direction.

g) using the execution result of f), performing the judgment on the above three cases: if the width of the book back image corresponding to the execution result is smaller than the preset book back width threshold, it is regarded as a non-book back image; The result is not corresponding to the title, and is regarded as a non-bookback image; if the execution result corresponds to the number of titles in the back image of the book being greater than one, the width of each title in the back image can be further determined.

h) further image cutting according to the width of each title.

As shown in step S209, the feature capturing module 143 of the image processing unit 14 performs Scale Invariant Feature Transform (SIFT) on the book back image further cut in step S208 to obtain at least one book back of the book. feature. Among them, the scale-invariant conversion system is a conventional technique, and the details thereof can be referred to "Object recognition from local scale-invariant features." proposed by D. G. Lowe, and therefore will not be described here.

As shown in step S210, the inventory data management unit 15 establishes an inventory data group corresponding to each book that has been put on the shelf, and stores the inventory data group to the memory unit 13; wherein each inventory data group includes its identification information, Inventory status, book location information, book back image, and book back features.

Referring to FIG. 4 and FIG. 6, it is worth mentioning that, in the shelf stage, the book inventory management system 1 can also automatically retrieve each of the books according to the identification data of each book (for example, the above-mentioned ISBN). A meta-data of a book, each interpretation material including the actual width and actual height of the book, as well as the hardcover or paperback of the book and the number of pages (which can be used to estimate the actual thickness of the book). Since the shelf stage can clearly know the book position identification information and the shelf order corresponding to each book, and the book location information is known; as shown in FIG. 6, if all the books are erect and closely arranged on the bookshelf, then the next The perfect cutting program, which automatically cuts perfectly, obtains the most accurate back image of the book, and accumulates information about the back image of these books, which is used as a subsequent inventory stage. This perfect cutting procedure includes the following sub-steps.

i) converting the actual height of each book and the actual thickness of the same image coordinate system "(shown in FIG frame 36), to afford the length and width of each book (in pixels).

j) Using the execution result similar to d) of the above step 206, performing horizontal edge detection and projecting the edged result into a histogram projection in the vertical direction.

k) using j) of the results of the determination, if the result of the histogram of a projection is greater than a predetermined threshold value, it indicates that, in this manner to find the correct position of the frame 3, the title of the book "each Books The scope of the title.

l) Using the execution result of k), find out the positive central position of one of the title ranges of each book, whereby the positive central position extends to the left and right directions by the same distance D _k , and the distance D _k is i) Half of the width, in this way, find a left boundary and a right boundary of the book and perform image cutting according to the left and right borders.

The above sub-steps i)~l), the main principle is to "use the interpretation data of all books to accumulate information about the actual length, width, thickness of each book and the proportion range of pixels of the image coordinate system". When the accuracy of the above sub-steps e) to h) is within the usable range (more than 90%), the perfect cutting procedure facilitates the improvement of the cutting accuracy in the future when the inventory is taken to complement the above sub-steps e) to h) The perfect point is that some of the book back images that are not accurately identified and cut (because the position and state of the book placed at the time of inventory are more complicated) can be further cut out using the above perfect cutting program.

It is worth mentioning that we can also design a book scanning machine (embedded system with hardware), measure the actual length, width and thickness of the book when it is put on the shelf, and combine the barcode scanner and the photographic device. By obtaining the barcode of the book together with the image of the back of the book, the interpretation data of the book is more in line with the actual situation and is complete.

Counting stage

Referring to FIG. 1 , FIG. 5A to FIG. 5C , and FIG. 6 , when the book is put on the shelf, it is an inventory book, and the memory unit 13 stores the established multiple inventory data groups, and the user can access the periphery of the human machine interface 11 . The input device switches the system to an inventory mode to begin the relevant processing of the inventory phase, wherein the inventory phase includes the following steps.

As shown in step S31, the parameter processing unit 12 provides a PTZ parameter set stored in advance in the memory unit 13 to control the photographing device 2 through the network to capture the frame 3 of the bookshelf.

As shown in step S32, the geometric correction module 141 of the image processing unit 14 performs image geometric correction on the frame 3, and the correction manner is similar to the step S205, and thus will not be described herein.

As shown in step S33, the cutting module 142 of the image processing unit 14 performs image cutting on the corrected frame 3 ' to obtain a book back image of the book on the shelf, and the image cutting manner is similar to step S206. Not to repeat here.

As shown in step S34, the geometric correction module 141 of the image processing unit 14 performs image geometric correction on the book back image cut in step S33, and the correction manner is similar to step S207, and thus will not be described herein.

As shown in step S35, the cutting module 142 of the image processing unit 14 performs post-processing on the back image corrected in step S34 to further cut the corrected book back image, and the processing manner is similar to step S208. Not to repeat here.

As shown in step S36, the feature capturing module 143 of the image processing unit 14 performs scale-invariant feature conversion on the back image processed after step S35 to obtain the book back feature of the book on the shelf, and the conversion manner is similar to Step S209, therefore, it will not be described here.

As shown in step S37, the comparison module 144 of the image processing unit 14 compares the back image of the book on the bookshelf with the back image of the book; in the preferred embodiment, the comparison The module 144 compares the book back features of the books on the bookshelf with the book back features of the stock books.

As shown in step S38, the inventory management unit 15 generates an inventory status of the books on the bookshelf based on the comparison result. In the preferred embodiment, the inventory status may be one of three states: "on-state", "new state", and "lost or sold state". The back image of the book defined on the bookshelf in the inventory stage is the query image set A = { a ₁ , a ₂ , ..., a _n }, and the back image of the book in the memory unit 13 is the stock image set. D = { d ₁ , d ₂ , ..., d _m }, the definitions of the above three states can be organized into Table 1.

For the book on the shelf status, the inventory management unit 15 can also analyze which book positions are disordered according to the sequence of the image sets A and D ; and provide the analysis result to the user for manually returning the book to the original position. In order to maintain the various states when the book is originally put on the shelf, the system can then return to the first step S31 of the inventory phase.

As shown in step S39, the inventory management unit 15 creates an inventory data group corresponding to each book on the bookshelf, and stores the inventory data group to the memory unit 13; wherein each inventory data group includes its inventory status, book Back image, and book back features. It is worth mentioning that the inventory data group can be further provided to the user as a reference to make them aware of the various states of the current book.

In practical applications, the present invention can also provide the current image data of the book in real time when the user wants to find a certain book, for example, using the input information of the book to input the book inventory management system 1, The inventory management unit 15 is informed by the memory unit 13 of the book location information of the book and its corresponding PTZ parameter group according to the input identification information, and then the PTZ parameter group is provided by the parameter processing unit 12 to control The photographing device can obtain the current image data of the book in real time, and the application is similar to the application, and can be achieved by the book inventory management system 1 and the method thereof provided by the present invention, and is not limited to the preferred embodiment.

In summary, the book inventory management system 1 and method thereof according to the present invention utilizes the photographic device 2 near the existing bookshelf to perform automated inventory management of the visual-based book, which not only reduces personnel costs, but also avoids The shortcomings encountered in the management of books by RFID technology; in addition, the book back image cutting technology used in the present invention is further matched with the hue average as a reference, and post-processing is added to greatly improve the book back image cutting. The accuracy of this also helps to improve the performance and management capabilities of the system; therefore, the object of the present invention can be achieved.

The above is only the preferred embodiment of the present invention, and the scope of the invention is not limited thereto, that is, the simple equivalent changes and modifications made by the scope of the invention and the description of the invention are All remain within the scope of the invention patent.

1‧‧‧Book Inventory Management System

11‧‧‧Human Machine Interface

12‧‧‧Parameter Processing Unit

13‧‧‧ memory unit

14‧‧‧Image Processing Unit

141‧‧‧Geometry Correction Module

142‧‧‧ cutting module

143‧‧‧Feature capture module

144‧‧‧ comparison module

15‧‧‧Stock Data Management Unit

2‧‧‧Photographing device

S11~S13‧‧‧Steps

3‧‧‧ frame

S201~S210‧‧‧Steps

3 ' ‧‧‧ corrected frame

3 " ‧‧‧ Frames after image cutting on the edge of the bookshelf

31‧‧‧ bookshelf edge

32‧‧‧Central horizontal line

D _k ‧‧‧Distance

S31~S39‧‧‧Steps

1 is a block diagram showing a preferred embodiment of the book inventory management system of the present invention;

2 is a flow chart illustrating a system initialization phase of one of the book inventory management methods of the present invention;

Figure 3 is a schematic view showing all the frames of the bookshelf in the shooting range of a photographing device;

4 is a flow chart showing one stage of the book inventory management method of the present invention;

5A-5C are schematic diagrams illustrating image geometric correction and image cutting of one of the frames;

Figure 6 is a schematic view showing the frames in which the books are erect and closely arranged on the bookshelf; and

Figure 7 is a flow chart showing an inventory stage of the book inventory management method of the present invention.

1. . . Book inventory management system

11. . . Human machine interface

12. . . Parameter processing unit

13. . . Memory unit

14. . . Image processing unit

141. . . Geometric correction module

142. . . Cutting module

143. . . Feature capture module

144. . . Alignment module

15. . . Inventory data management unit

2. . . Photography device

Claims (29)

A book inventory management system, which is adapted to be connected to a photographic device, the system comprising: a memory unit storing at least one inventory data set that is established, the inventory data set comprising a book back image of an inventory book; an image processing The unit is electrically connected to the memory unit, and the image processing unit includes a cutting module and a matching module, and the cutting module is configured to receive a frame of the bookshelf captured by the camera device, and the image is The frame performs image cutting to obtain a back image of the book on the bookshelf. When the system is in an inventory mode, the comparison module is used in the memory unit to perform a book back on the book on the shelf. Comparing the image with the back image of the book of the inventory book, wherein the cutting module utilizes a predefined state transition model and uses the dynamic programming algorithm to perform image cutting to obtain the book back of the book on the shelf Image, the state transition model is a finite state machine, and has a plurality of different state categories, and states of the plurality of location points in the frame of the bookshelf respectively correspond to the states In one of the above, the cutting module utilizes a plurality of hue function values respectively corresponding to the position points in the frame of the bookshelf as the weight of the dynamic programming algorithm, and each hue function value corresponds to the corresponding position point. Corresponding to at least one of a hue average and a state; and an inventory data management unit electrically connected to the memory unit and the image processing unit, wherein when the system is in the inventory mode, the inventory data management unit is configured to The comparison result of the comparison module produces an inventory status of the books on the shelf and stores them in the memory unit. The book inventory management system according to claim 1, wherein when the system is in the inventory mode, the inventory data management unit is further configured to create an inventory data group corresponding to the books on the shelf, and The storage unit is stored, wherein the inventory data group includes a book back image and an inventory status of the book on the bookshelf. According to the book inventory management system of claim 1, wherein the image processing unit further comprises a feature capture module, wherein the feature capture module is configured to scale the book back image of the book on the shelf Invariant feature conversion to obtain at least one book back feature of the book. The book inventory management system of claim 3, wherein the inventory data set further comprises at least one book back feature of the inventory book, and when the system is in the inventory mode, the comparison module is The book back feature of the book on the bookshelf is compared with the back feature of the book of the stock book. According to the book inventory management system of claim 3, wherein the inventory data management unit is configured to create an inventory data group corresponding to the books on the shelf and store the same when the system is in a shelf mode. To the memory unit, wherein the inventory data set includes a back image of a book on the shelf. According to the book inventory management system of claim 5, wherein, when the system is in the shelf mode, the inventory data management unit is further configured to initialize an inventory status of the books on the shelf, wherein, corresponding to the shelf The inventory data group of the book includes its initialized inventory status. According to the book inventory management system of claim 6, wherein the inventory data group corresponding to the books on the bookshelf further includes the book back feature of the books on the bookshelf. According to the book inventory management system of claim 1, wherein the cutting module of the image processing unit is further configured to perform horizontal edge detection on the back image of the book on the shelf, and then to marginalize the book. The back image is subjected to histogram projection, and the histogram projection result is used for correlation judgment to determine the number of a title in the back image of the book and the width of each title, and then further image the back of the book according to the width of each title Perform image cutting. The book inventory management system of claim 1, wherein the image processing unit further comprises a geometric correction module for performing image geometric correction on the frame of the bookshelf. According to the book inventory management system of claim 9, wherein the geometric correction module of the image processing unit performs image geometric correction on the frame of the bookshelf by using a pre-established frame geometric transformation matrix. The frame geometry conversion matrix is pre-established according to preset preset control point coordinates. The book inventory management system according to claim 9 , wherein the geometric correction module of the image processing unit is further configured to perform image geometric correction on the back image of the book on the bookshelf. According to the book inventory management system of claim 11, wherein the geometric correction module of the image processing unit determines a plurality of book back control point coordinates according to multiple boundaries of the book back image of the book on the bookshelf , Then, according to the coordinates of the back control points of the book, a back geometry conversion matrix is obtained, and then the back geometry of the book is used to perform image geometric correction on the back image of the book. A book inventory management method is adapted to be executed by a system connected to a photographing device, the photographing device corresponding to a photographing device identifying information, the method comprising the steps of: A) receiving one of the books to identify information; B) initializing the An inventory state of the book; C) receiving a frame of a bookcase captured by the camera device; D) utilizing a predefined state transition model, and performing image cutting on the frame by using a dynamic programming algorithm Obtaining a back image of the book on the bookshelf, wherein the state transition model is a finite state machine, and has a plurality of different state categories, and states of the plurality of location points in the frame respectively correspond to the states One of the categories, and using a plurality of hue function values respectively corresponding to the corresponding position points as the weight of the dynamic programming algorithm, and each hue function value is at least one of a hue average value and a state of the corresponding position point Corresponding; and E) establishing an inventory data group corresponding to the book, wherein the inventory data group includes the identification information, the inventory status, and the back image of the book. The book inventory management method according to claim 13 further includes a step F) before the step E), the information identifying the information, the preset book position information, and a preset shelf order, Forming a book location information of the book, and the inventory data group of the book further includes The book location information. The book inventory management method according to claim 14, wherein the identification information of the book is an international standard book number. According to the book inventory management method of claim 13, further comprising a step G) between step C) and step D), performing image geometric correction on the frame, and step D) correcting the corrected The frame is image cut. According to the book inventory management method of claim 16, wherein the step G) performs image geometric correction on the frame by using a pre-established frame geometric transformation matrix, and the frame geometric transformation matrix is preset according to the preset Multiple control point coordinates are pre-established. The book inventory management method according to claim 13 further includes a step H) after the step D), performing image geometric correction on the back image of the book. The book inventory management method according to claim 18, wherein the step H) comprises the following sub-steps: h-1) determining a plurality of book back control point coordinates according to the plurality of boundaries of the back image of the book; -2) determining a book back geometry transformation matrix according to the coordinates of the back control points of the book; and h-3) performing image geometric correction on the back image of the book by using the book back geometry transformation matrix. According to the book inventory management method described in claim 13 of the patent application, further comprising a step I) between step D) and step E), the back of the book Like performing a scale-invariant feature conversion to obtain at least one book back feature, the inventory data set of the book further includes the book back feature. According to the book inventory management method of claim 13, further comprising a step J) after the step D), and further performing post-processing on the back image of the book, the post-processing comprising the following sub-steps: j-1) The back image of the book performs horizontal edge detection; j-2) the histogram projection of the marginal book back image; j-3) using the histogram projection result of substep j-2) to perform correlation judgment to determine the The number of titles in the back image and the width of each title; and j-4) further image cutting of the back image of the book according to the width of each title. A book inventory management method is adapted to be executed by a system connected to a photographing apparatus, the method comprising the steps of: A) providing at least one inventory data set that has been established, wherein the stock data set includes one of the stock books a back image; B) receiving a frame of a bookcase captured by the camera; C) using a predefined state transition model, and using the dynamic programming algorithm to image the frame of the bookcase, Obtaining a back image of a book on the bookshelf, wherein the state transition model is a finite state machine, and has a plurality of different state categories, and the states of the plurality of location points in the frame of the bookshelf respectively correspond to One of the state categories, and using the plurality of hue function values corresponding to the corresponding position points as the weights of the dynamic programming algorithm, each hue function value is Corresponding to the at least one of the hue average and the state of the position point; D) using the back image of the book on the shelf to compare with the back image of the book; and E) according to step D The result of the comparison produces an inventory status of the books on the shelf. The book inventory management method according to claim 22, further comprising a step F) after the step E), establishing an inventory data group corresponding to the books on the shelf, the inventory data group including the books on the shelf Back image and inventory status. According to the book inventory management method of claim 22, further comprising a step G) between step B) and step C), performing image geometric correction on the frame of the bookshelf, and step C) correcting the image This frame performs image cutting. According to the book inventory management method of claim 24, the step G) performs image geometric correction on the frame of the bookshelf by using a pre-established frame geometric transformation matrix, and the frame geometric transformation matrix is based on Pre-set multiple control point coordinates are pre-established. The book inventory management method according to claim 22, further comprising a step H) after the step C), performing image geometric correction on the back image of the book on the bookshelf. The book inventory management method according to claim 26, wherein the step H) comprises the following sub-steps: h-1) according to a plurality of boundaries of the back image of the book on the bookshelf, Obtaining a plurality of book back control point coordinates; h-2) determining a book back geometry conversion matrix according to the book back control point coordinates; and h-3) using the book back geometry transformation matrix to image the book back image Geometric Correction. According to the book inventory management method of claim 22, wherein each inventory data group includes at least one book back feature of the inventory book, and step D) includes the following sub-steps: d-1) on the shelf The back image of the book is converted into a scale-invariant feature to obtain at least one book back feature of the book on the shelf; and d-2) the book back feature of the book on the shelf, and the back of each book Features are compared for features. The book inventory management method according to claim 22, further comprising a step I) after the step C), post-processing the book back image of the book on the bookshelf, the post-processing comprising the following sub-steps: i -1) horizontal edge detection of the back image of the book; i-2) histogram projection of the marginal book back image; i-3) using the histogram projection result of substep i-2) for correlation judgment, To determine the number of a title in the back image of the book and the width of each title; and i-4) further perform image cutting on the back image of the book according to the width of each title.