WO2022111549A1

WO2022111549A1 - Document recognition method and apparatus, and readable storage medium

Info

Publication number: WO2022111549A1
Application number: PCT/CN2021/132930
Authority: WO
Inventors: 徐青松; 李青
Original assignee: 杭州睿胜软件有限公司
Priority date: 2020-11-25
Filing date: 2021-11-24
Publication date: 2022-06-02
Also published as: CN112308036A

Abstract

The present invention provides a document identification method and apparatus, and a readable storage medium, the method comprising: scaling to a preset first node size the length of an original image containing a document in a first direction, and compensating, to a preset second node size, for the length of the scaled original image in a second direction, wherein the first direction is perpendicular to the second direction, and the length of the original image in the first direction is not less than the length thereof in the second direction; then, acquiring a marking box of the document, and enlarging the marking box of the document according to a preset proportion; and finally, segmenting an image of the document on the basis of the enlarged marking box of the document, and outputting the image, such that deformation of images can be prevented, and the subsequent processing speed is increased by unifying original document pictures of various sizes to a preset size, in addition, the marking box is enlarged to prevent the loss of document edge areas, thereby lowering the difficulty of recognizing the document in a picture.

Description

Bill identification method, device and readable storage medium

technical field

The invention relates to the technical field of artificial intelligence, in particular to a bill identification method, a bill identification device and a readable storage medium.

Background technique

With the continuous development of the economy and the continuous improvement of people's consumption level, in order to protect people's consumption rights and interests, bills have become a powerful guarantee for consumers and an effective reimbursement certificate. Therefore, financial personnel need to deal with a large number of bills every day. At the same time, more and more people are keeping track of their consumption through accounting and classification.

In recent years, bill recognition technology has been developing continuously, but it is still difficult to accurately identify bills in pictures, especially when there are multiple bills distributed on a picture, that is, when a picture includes multiple bills, It is difficult to recognize the multiple bills in the picture.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a bill identification method, a bill identification device and a readable storage medium to solve the problem of difficulty in bill identification.

In order to solve the above-mentioned technical problems, the present invention provides a bill identification method, including:

Preprocessing the original image containing the ticket, the preprocessing includes: after scaling the length of the original image in the first direction to a preset first node size, and then placing the scaled original image in the first The lengths in the two directions are complemented to a preset second node size, wherein the first direction is perpendicular to the second direction, and the length of the original image in the first direction is not less than that in the second direction length;

acquiring the labeling frame of the bill, and enlarging the labeling box of the bill according to a preset ratio;

Based on the enlarged annotation frame of the bill, the image of the bill is segmented and output.

Optionally, in the bill identification method, the bill identification method further includes:

A plurality of the first node sizes and a plurality of the second node sizes are preset, and the first node sizes and the second node sizes are in one-to-one correspondence; when scaling the original image, the The length of the original image in the first direction is scaled to the closest numerical value of the first node size; and,

The length of the scaled original image in the second direction is padded to the second node size corresponding to the scaled first node size.

Optionally, in the bill identification method, the size of the first node and the size of the second node are the same.

Optionally, in the bill recognition method, the method for compensating the length of the scaled original image in the second direction to a preset second node size includes:

Filling blank areas along the side of the scaled original image along the second direction.

Optionally, in the bill identification method, the method for obtaining the marked frame of the bill includes:

obtaining location area information for the ticket; and,

Based on the location area information of the ticket, the callout frame of the ticket is acquired.

Optionally, in the bill identification method, before outputting the image of the bill, the bill identification method further includes:

The orientation of the image of the bill is adjusted so that the orientation of the characters on the bill is a preset direction.

Optionally, in the bill identification method, after acquiring the image of the bill, the bill identification method further includes:

identifying the edge of the note; and,

Based on the recognition result, the image of the bill is trimmed.

Correcting the image content of the ticket, the correction includes global correction and local correction.

The present invention also provides a bill identification device, comprising:

The image preprocessing module is used for preprocessing the original image containing the bill, and the preprocessing includes: after scaling the length of the original image in the first direction to a preset first node size, The length of the original image in the second direction is complemented to a preset second node size, wherein the first direction is perpendicular to the second direction, and the length of the original image in the first direction not less than the length in the second direction;

A labeling frame acquisition and adjustment module, used to obtain the labeling frame of the bill, and enlarge the labeling frame of the bill by a preset ratio;

The image post-processing module is used for segmenting the image of the bill based on the enlarged annotation frame of the bill, and outputting the image.

Optionally, in the bill identification device, the bill identification device further includes a node size setting module, and the node size setting module is used to preset a plurality of the first node sizes and a plurality of the second node sizes. node size, the first node size and the second node size are in one-to-one correspondence;

When scaling the original image, the image preprocessing module scales the length of the original image in the first direction to the first node size with the closest value; The length of the image in the second direction is padded to the second node size corresponding to the scaled first node size.

Optionally, in the bill identification device, the size of the first node and the size of the second node are the same.

Optionally, in the bill recognition device, the image preprocessing method for compensating the length of the scaled original image in the second direction to a preset second node size includes: along the second The direction fills the blank space on the sides of the scaled original image.

Optionally, in the bill recognition device, the image post-processing module includes an image segmentation module and an image output module, and the image segmentation module is used to segment the bill based on the enlarged label frame of the bill. The image output module is used for outputting the image of the bill that has been segmented.

Optionally, in the bill recognition device, the image post-processing module further includes an orientation adjustment module, and the orientation adjustment module is used to adjust the orientation of the image of the bill, so that the characters on the bill are directional. The orientation is the default direction.

Optionally, in the bill identification device, the image post-processing module further includes an edge processing module, the edge processing module is used to identify the edge of the bill, and based on the identification result, the image of the bill is Edge trimming.

Optionally, in the bill recognition device, the image post-processing module further includes an image correction module, and the image correction module is used to correct the image content of the bill, and the correction includes global correction and local correction. Correction.

The present invention also provides a readable storage medium, characterized in that, the readable storage medium stores a computer program, and when the computer program is executed, the above-mentioned bill identification method is implemented.

To sum up, in the method for identifying bills, the device for identifying bills, and the readable storage medium provided by the present invention, first, the original image containing bills is preprocessed, and the preprocessing includes: storing the original image in a first After the length in the direction is scaled to the preset first node size, the length of the scaled original image in the second direction is filled to the preset second node size, wherein the first direction is perpendicular to the second direction, and the length of the original image in the first direction is not less than the length in the second direction; that is, by adjusting the original image containing the bill to a preset size, and then performing subsequent segmentation, and Before dividing, the labeling frame of each bill is enlarged according to the preset ratio. In this way, the speed of subsequent processing is improved by unifying the original bill images of various sizes to the preset size, and the image size adjustment method adopted is not It will bring the trouble of image deformation. In addition, by enlarging the label frame, the loss of the edge area of the bill is avoided, thus reducing the difficulty of identifying the bill in the picture.

Description of drawings

FIG. 1 is a flowchart of a method for identifying a ticket provided by an embodiment of the present invention;

FIG. 2 is a schematic diagram of an exemplary original image including multiple notes provided by an embodiment of the present invention;

3 is a schematic diagram of resizing an original image in an embodiment of the present invention;

FIG. 4 is a schematic diagram of forming a labeling frame of each bill in an embodiment of the present invention;

5 is a schematic diagram of enlarging each marked frame in an embodiment of the present invention;

6 is a schematic diagram of an image of a bill formed by cutting in an embodiment of the present invention;

7 is a schematic diagram of performing edge trimming processing on an image of a bill in an embodiment of the present invention;

FIG. 8 is a block diagram of the composition of a bill identification device provided by an embodiment of the present invention;

Wherein, each reference sign is described as follows:

P1, P2, P3, P4 - bills; A1, A2 - blank area; Z1, Z2, Z3, Z4 - callout boxes; Z1', Z2', Z3', Z4' - text boxes;

10-image preprocessing module; 20-marking frame acquisition and adjustment module; 30-image post-processing module; 301-segmentation module; 302-image output module; 303-direction adjustment module; 304-edge processing module; 305-image correction module.

Detailed ways

The bill identification method, bill identification device and readable storage medium proposed by the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The advantages and features of the present invention will become more apparent from the following description. It should be noted that, the accompanying drawings are all in a very simplified form and in inaccurate scales, and are only used to facilitate and clearly assist the purpose of explaining the embodiments of the present invention. Furthermore, the structures shown in the drawings are often part of the actual structure. In particular, each drawing needs to show different emphases, and sometimes different scales are used.

Unless otherwise defined, technical or scientific terms used in the present invention should have the ordinary meaning as understood by one of ordinary skill in the art to which the present invention belongs. The terms "first," "second," and similar terms used herein do not denote any order, quantity, or importance, but are merely used to distinguish different components. "Comprises" or "comprising" and similar words mean that the elements or things appearing before the word encompass the elements or things recited after the word and their equivalents, but do not exclude other elements or things.

In order to solve the problems in the prior art, the embodiments of the present invention provide a bill identification method, a bill identification device and a readable storage medium.

It should be noted that the topic search method of the embodiment of the present invention can be applied to the topic search apparatus of the embodiment of the present invention, and the topic search apparatus can be configured on an electronic device. Wherein, the electronic device may be a personal computer, a mobile terminal, etc., and the mobile terminal may be a hardware device with various operating systems, such as a mobile phone, a tablet computer, and the like.

As shown in FIG. 1 , this embodiment provides a bill identification method, and the bill identification method includes the following steps:

S11. Preprocess the original image containing the bill, the preprocessing includes: after scaling the length of the original image in the first direction to a preset first node size, and then scaling the scaled original image The length in the second direction is complemented to a preset second node size, wherein the first direction is perpendicular to the second direction, and the length of the original image in the first direction is not less than that in the second direction the length in the direction;

S12, acquiring the labeling frame of the bill, and enlarging the labeling frame of the bill according to a preset ratio;

S13 , segment the image of the bill based on the enlarged annotation frame of the bill, and output it.

In the bill recognition method provided in this embodiment, after the original image containing the bill is adjusted to a preset size, subsequent segmentation is performed, and before the segmentation, the labeling frame of the bill is enlarged according to a preset ratio. By unifying the original bill images of various sizes to the preset size, the speed of subsequent processing is improved, and the image size adjustment method adopted will not cause the trouble of image deformation. Edge regions are lost, thus reducing the difficulty of identifying multiple bills in the picture.

The above steps are further described in detail below.

In step S11, preferably, a plurality of the first node sizes and a plurality of the second node sizes are preset, and the first node sizes and the second node sizes are in one-to-one correspondence; When the image is zoomed, the length of the original image in the first direction is scaled to the size of the first node with the closest value; and the length of the scaled original image in the second direction is complemented to the second node size corresponding to the scaled first node size. Further preferably, the size of the first node is the same as the size of the corresponding second node. After scaling the original image and supplementing the blank area, the image is not deformed, and the obtained image is consistent with the preset size, so that in subsequent processing (such as the bill area recognition model described below) When obtaining the position information of each bill), the processing speed is significantly improved, and by setting the size of the first node and the size of the second node to be the same, so that the image is square, the processing speed of the model can be further improved.

For example, set multiple node sizes to 800X800, 1600X1600... (other sizes are also possible), first determine which node size the length of the original image in the first direction is close to, such as the length of the original image in the first direction If the length of the original image in the first direction is 1400 or 1800, it is scaled to 1600.

After the user takes a picture, if the image of the final output ticket is too different from the image seen at the time of shooting, it will bring a bad use experience to the user. The size adjustment range of the image is as small as possible, so that the image size of the final output ticket is as consistent as possible with the original image, so that a better user experience can always be guaranteed.

It should be understood that in step S11, scaling the original image refers to scaling the entire original image, not only adjusting the length in the first direction, but also adjusting the length in the second direction along with the length in the first direction And adjustment, in this way, can ensure that the image does not deform.

It should be noted that, if the length in the first direction of the original image is the same as the preset first node size, it is not necessary to perform scaling, but directly fill in the length in the second direction to the corresponding second node size. For example, if the length of the original image in the first direction is 800, no scaling is performed, but the length of the original image in the second direction is directly filled to 800.

It should be noted that, if the length in the first direction of the original image is the same as the length in the second direction, and the preset first node size and the second node size are the same, the length in the first direction of the original image After scaling to the preset first node size, there is no need to fill in the length in the second direction. For example, if the original image is 1000×1000, after scaling the length of the original image in the first direction to 800, the length in the second direction of the original image is also changed to 800, so there is no need to make up.

In step S11, a blank area can be supplemented along the side of the scaled original image along the second direction to make up the length of the scaled original image in the second direction to a preset second node size.

FIG. 2 shows an exemplary original image including four bills provided in this embodiment, and the included bills are: bill P1, bill P2, bill P3, and bill P4. Since the length of the original image in the first direction is greater than or equal to the length in the second direction, for the original image shown in the figure, the first direction can be understood as the X direction shown in FIG. 2 , The second direction can be understood as the Y direction shown in FIG. 2 .

As shown in FIG. 3 , in this embodiment, the length of the scaled original image in the second direction is filled by supplementing blank areas A1 and A2 on both sides of the original image along the second direction To be the same as the length in the first direction, and preferably, the supplementary blank areas A1 and A2 on both sides have equal areas. In other embodiments, blank areas may also be supplemented on one side of the original image along the second direction. When blank areas are supplemented on one side of the original image, the area of the supplemented blank areas is equal to The sum of the areas of the blank areas supplemented on both sides is equal. In other embodiments, the supplemented area may also be an area filled with an image, such as a grid line area and the like.

In step S12, the marked frame of the bill can be obtained by obtaining the position area information of the bill. Specifically, the method for obtaining the labeling frame of the bill may include: acquiring the location area information of the bill, and acquiring the labeling box of the bill based on the location area information of the bill. For the original image shown in FIG. 2 , through step S12 , the column frame of each of the bills obtained is as shown in FIG. 4 . The text box of the bill P1 is Z1, the text box of the bill P2 is Z2, the text box of the bill P3 is Z3, and the text box of the bill P4 is Z4.

Wherein, a bill area identification model can be used to obtain the position area information of each of the bills. The ticket area identification model may employ machine learning techniques and run, for example, on a general purpose computing device or a special purpose computing device. For example, the bill region recognition model can be implemented by using a neural network such as a deep convolutional neural network (DEEP-CNN). In some embodiments, the image is input to the bill region recognition model, and the bill region identification model can identify the boundaries of each bill in the input image, and then mark out the identified boundaries, so as to obtain each of the bills. The callout box for the ticket.

After obtaining the labeling frame of each bill, as shown in Figure 5, the labeling box of each bill is enlarged according to the preset ratio. It is guaranteed that the entire area of the ticket is contained within the callout box. It should be understood that the enlargement of the marked frame of each bill here refers to the enlargement of the marked frame along the periphery.

In some embodiments, each of the bill marked boxes is enlarged by 5%, and in other embodiments, it can also be enlarged by 3%, 7%, 9%, and so on. In addition, different magnification ratios can be matched for bills of different sizes. For example, for small bills, the marked frame can be enlarged by 2%, and for large bills, the marked frame can be enlarged by 6%, and so on.

In step S13, based on the enlarged text frame Z1' of the text frame Z1 in step S12, the enlarged text frame Z2' of the text frame Z2, the enlarged text frame Z3' of the text frame Z3, and the enlarged text frame Z4. In the text box Z4', all the bills are divided, and the image of the bill P1 obtained from the segmentation is shown in Figure 6. In addition to the bill itself, it also includes a partial image of the bill P2. The partial image, for the bill P1, is Redundant images.

In view of this, preferably, after acquiring the image of each of the bills, the edge of each of the bills is identified; and, based on the identification result, the image of each of the bills is subjected to edge trimming processing. The image of the bill P1 obtained by the edge trimming process is shown in FIG. 7 , and it can be seen from FIG. 7 that the redundant image is cut out through the edge trimming process.

In this embodiment, when identifying the edge of any one of the bills, the following methods are used:

processing the image of the bill to obtain a line drawing of grayscale contours in the image of the bill;

Combine multiple lines in the line drawing to obtain multiple reference boundary lines;

Identify the boundary area of the bill image through a boundary area model, wherein the boundary area model and the bill area identification model may use the same model;

Calculate the number of pixels belonging to the boundary area of each of the reference boundary lines, and according to the number of the reference boundary lines, the number of pixels belonging to the boundary area of the reference boundary lines, and the Boundary area, identify the edge of the note.

Among them, the image of the bill can be processed by the edge detection algorithm to obtain the line drawing of the grayscale outline. For example, an image can be processed by an OpenCV-based edge detection algorithm to obtain a line drawing of the grayscale contours in the image. OpenCV is an open source computer vision library. Edge detection algorithms based on OpenCV include Sobel, Scarry, Canny, Laplacian, Prewitt, Marr-Hildresh, scharr and other algorithms. For example, the Canny edge detection algorithm is used in this embodiment. The Canny edge detection algorithm is a multi-stage algorithm, that is, the Canny edge detection algorithm consists of multiple steps. For example, the Canny edge detection algorithm includes: 1. Image noise reduction: using Gaussian Filter to smooth the image; 2. Calculate the image gradient: use the first-order partial derivative finite difference to calculate the gradient magnitude and direction; 3. Non-maximum suppression: perform non-maximum suppression on the gradient amplitude; 4. Threshold filtering: use A dual threshold algorithm detects and connects edges.

It should be noted here that in other embodiments, other edge identification methods known to those skilled in the art may also be used, and the selection of specific edge identification methods does not constitute a limitation to the present application.

In step S13, further preferably, after the image of each of the bills is acquired, the image content of each of the bills is corrected, and the correction includes global correction and local correction. In this embodiment, in order to reduce the range of correction, the correction step may be performed after the edge trimming process, and in other embodiments, the correction step may also be performed before the edge trimming process.

In the process of converting a paper document into a text image, the text image may be inclined, and the inclination may adversely affect the analysis of the text image (for example, character recognition, etc.) and other processing. Therefore, in this embodiment, the image content of each of the bills is corrected, so as to avoid the inclination of the text image from adversely affecting the analysis and processing of the original image.

In this embodiment, the image content of any bill can be corrected by using the following steps: performing global correction processing on the image of the bill to obtain an intermediate corrected image; performing local adjustment on the intermediate corrected image to obtain a target corrected image ; wherein, performing local adjustment on the intermediate correction image to obtain a target correction image, including:

According to the intermediate correction image, the lower boundaries of M character lines corresponding to the M character rows of the image of the bill are determined; based on the intermediate correction image and the lower boundaries of the M character rows, local adjustment reference lines and M reservation coefficient groups, wherein each reservation coefficient group in the M reservation coefficient groups includes a plurality of reservation coefficients; according to the lower boundary of the M character lines, the local adjustment reference line and the reservation coefficient group , determine M local adjustment offset groups corresponding to the M character lines, wherein each local adjustment offset group in the M local adjustment offset groups includes a plurality of local adjustment offsets ; Perform local adjustment on the M character lines in the intermediate correction image according to the M local adjustment offset groups to obtain the target correction image.

In other embodiments, other methods well known to those skilled in the art may also be used to correct the image content of each of the bills, which will not be repeated here.

In addition, when sticking bills, there may be a phenomenon that the bills are placed in different directions (some are sticking up, some are sticking upside down or sideways). When the characters are sideways or upside down, it is not conducive to statistical observation. In view of this, in this embodiment, preferably, before outputting any image of the bill, the bill recognition method further includes: adjusting the direction of the image of the bill, so that the characters on the bill The orientation is the default direction. Preferably, the preset direction is the positive Y direction in the plane coordinate system, so as to facilitate subsequent identification.

Based on the same idea, this embodiment also provides a bill identification device, as shown in FIG. 8 , the bill identification device includes:

The image preprocessing module 10 is configured to perform preprocessing on the original image containing the bill, and the preprocessing includes: after scaling the length of the original image in the first direction to a preset first node size, scaling the The length of the original image in the second direction is then complemented to a preset second node size, wherein the first direction is perpendicular to the second direction, and the length of the original image in the first direction is The length is not less than the length in the second direction;

The labeling frame obtaining and adjusting module 20 is used for obtaining the labeling frame of the bill, and enlarging the labeling frame of the bill according to a preset ratio;

The image post-processing module 30 is configured to segment and output the image of the bill based on the enlarged annotation frame of the bill.

Wherein, the bill identification device further includes a node size setting module (not shown in the figure), and the node size setting module is used for setting the first node size and the second node size. When a plurality of the first node sizes and a plurality of the second node sizes are preset, the first node sizes and the second node sizes are in one-to-one correspondence. When scaling the original image, the image preprocessing module 10 scales the length of the original image in the first direction to the first node size with the closest value; The length of the original image in the second direction is padded to the second node size corresponding to the scaled first node size. The method for the image preprocessing module 10 to fill in the length of the scaled original image in the second direction to a preset second node size includes: along the second direction on the side of the scaled original image; Fill in the blank space.

The image post-processing module 30 specifically includes a segmentation module 301 and an image output module 302. The image segmentation module 301 is configured to segment the image of the ticket based on the enlarged annotation frame of the ticket. The image output module 302 is used for outputting the image of the segmented bill.

Preferably, the image post-processing module 30 further includes: an orientation adjustment module 303, which is used to adjust the orientation of the image of the bill, so that the orientation of the characters on the bill is a preset direction . Preferably, the preset direction is the positive Y direction in the plane coordinate system, so as to facilitate subsequent identification.

Further, the bill identification device may further include an edge processing module 304, and the edge processing module 304 is configured to identify the edge of the bill, and based on the identification result, perform edge trimming processing on the image of the bill.

Further, the bill identification device may further include an image correction module 305; the image correction module 305 is used to correct the image content of the bill, and the correction includes global correction and local correction. Wherein, the correction of the image content of the bill by the image correction module 305 can be performed after the edge processing module 304 performs edge trimming processing on the image of the bill, and the edge processing module 304 also performs the edge processing on the bill. The image of the ticket is processed before edge trimming.

It should be noted that each module in the bill identification device provided in this embodiment is respectively used to implement each step of the bill identification method provided in this embodiment. Therefore, for the specific description of the functions that each module can implement, please refer to the above The relevant descriptions of the corresponding steps of the bill identification method will not be repeated where repeated. In addition, the bill identification device can achieve the same technical effect as the bill identification method described above, which will not be repeated here.

It can be understood that, the bill recognition device, the image preprocessing module 10, the frame acquisition and adjustment module 20, and the image post-processing module 30 can be combined in one device, or any one of the modules can be split into A plurality of sub-modules, or, in the bill recognition device, at least part of the functions of one or more modules in the image preprocessing module 10, the frame acquisition and adjustment module 20, and the image post-processing module 30 can be combined with at least part of the other modules. The functions are combined and implemented in one functional module. According to an embodiment of the present invention, at least one of the bill identification device, the statistical analysis module 11 and the calibration module 12 may be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), system-on-chip, system-on-substrate, system-on-package, application specific integrated circuit (ASIC), or any other reasonable means of integrating or packaging circuits, implemented in hardware or firmware, or in software, It can be realized by an appropriate combination of the three implementations of hardware and firmware. Alternatively, in the bill recognition device, at least one of the image preprocessing module 10, the frame acquisition and adjustment module 20, and the image postprocessing module 30 may be at least partially implemented as a computer program module, and when the program is run by a computer , can execute the function of the corresponding module.

In addition, this embodiment further provides a readable storage medium, where a computer program is stored in the readable storage medium, and when the computer program is executed, the bill identification method described in this embodiment is implemented.

The readable storage medium can be a tangible device that can hold and store instructions for use by the instruction execution device, such as, but not limited to, electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or the above. any suitable combination. More specific examples (non-exhaustive list) of readable storage media include: portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or Flash memory), static random access memory (SRAM), portable compact disc read only memory (CD-ROM), digital versatile disc (DVD), memory sticks, floppy disks, mechanical encoding devices, and any suitable combination of the foregoing. The computer programs described herein can be downloaded to various computing/processing devices from readable storage media, or to external computers or external storage devices over a network such as the Internet, a local area network, a wide area network, and/or a wireless network. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and/or edge servers. A network adapter card or network interface in each computing/processing device receives the computer program from the network and forwards the computer program for storage in a readable storage medium in the respective computing/processing device. The computer program for carrying out the operations of the present invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or any other program in one or more programming languages. Combining source or object code written in programming languages including object-oriented programming languages such as Smalltalk, C++, etc., and conventional procedural programming languages such as the "C" language or similar programming languages. The computer program may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server . Where a remote computer is involved, the remote computer may be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (eg, through the Internet using an Internet service provider) connect). In some embodiments, electronic circuits, such as programmable logic circuits, field programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), that can execute computer programmable logic circuits, are personalized by utilizing state information from a computer program. Program instructions are read to implement various aspects of the present invention.

Aspects of the present invention are described herein with reference to flowchart illustrations and/or block diagrams of methods, systems and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by a computer program. These computer programs may be provided to the processor of a general purpose computer, special purpose computer or other programmable data processing apparatus to produce a machine which, when executed by the processor of the computer or other programmable data processing apparatus, produces a Means implementing the functions/acts specified in one or more blocks of the flowchart and/or block diagrams. These computer programs can also be stored in a readable storage medium, and these computer programs cause computers, programmable data processing devices and/or other devices to operate in a specific manner, so that the readable storage medium storing the computer program includes a An article of manufacture comprising instructions for implementing various aspects of the functions/acts specified in one or more blocks of the flowchart and/or block diagrams.

A computer program can also be loaded onto a computer, other programmable data processing apparatus, or other equipment, causing a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other equipment to produce a computer-implemented process that causes A computer program executing on a computer, other programmable data processing apparatus, or other device implements the functions/acts specified in one or more blocks of the flowcharts and/or block diagrams.

To sum up, in the method for identifying bills, the device for identifying bills, and the readable storage medium provided by the present invention, first, the original image containing bills is preprocessed, and the preprocessing includes: storing the original image in the first After the length in one direction is scaled to the preset first node size, the length of the scaled original image in the second direction is filled to the preset second node size, wherein the first direction is vertical in the second direction, and the length of the original image in the first direction is not less than the length in the second direction; then, obtain the marked frame of each of the bills, and press the marked frame of each of the bills The preset ratio is enlarged; finally, all the bills are segmented based on the enlarged annotation frame of each of the bills, so as to obtain an image of each of the bills and output them. That is, by adjusting the original image containing multiple bills into a square, and then performing subsequent segmentation, and before dividing, enlarging the marked frame of each bill according to a preset ratio, so that the original bills of various sizes can be divided into different sizes. The image is unified to the same size, which improves the speed of subsequent processing, and the image size adjustment method adopted will not bring about the trouble of image deformation. In addition, by enlarging the annotation frame, the loss of the edge area of the bill is avoided. Difficulty in identifying multiple bills in a picture.

The above description is only a description of the preferred embodiments of the present invention, and is not intended to limit the scope of the present invention. Any changes and modifications made by those of ordinary skill in the field of the present invention based on the above disclosure all belong to the protection scope of the claims.

Claims

A method for identifying bills, comprising:

Preprocessing the original image containing the ticket, the preprocessing includes: after scaling the length of the original image in the first direction to a preset first node size, and then placing the scaled original image in the first The lengths in the two directions are complemented to a preset second node size, wherein the first direction is perpendicular to the second direction, and the length of the original image in the first direction is not less than that in the the length in the second direction;

acquiring the labeling frame of the bill, and enlarging the labeling box of the bill according to a preset ratio;

Based on the enlarged annotation frame of the bill, the image of the bill is segmented, and the segmented image of the bill is output.
The bill identification method according to claim 1, wherein the bill identification method further comprises:

A plurality of the first node sizes and a plurality of the second node sizes are preset, and the first node sizes and the second node sizes are in one-to-one correspondence; when scaling the original image, the The length of the original image in the first direction is scaled to the closest numerical value of the first node size; and,

The length of the scaled original image in the second direction is padded to the second node size corresponding to the scaled first node size.
The bill identification method according to claim 1, wherein the size of the first node and the size of the second node are the same.
The bill recognition method according to claim 1, wherein the method for compensating the length of the scaled original image in the second direction to a preset second node size comprises:

Blank areas are supplemented along the side of the scaled original image along the second direction.
The bill identification method according to claim 1, wherein the method for obtaining the marked frame of the bill comprises:

obtaining location area information for the ticket; and,

Based on the location area information of the ticket, the callout frame of the ticket is acquired.
The bill identification method according to claim 1, wherein before outputting the image of the bill, the bill identification method further comprises:

The orientation of the image of the bill is adjusted so that the orientation of the characters on the bill is a preset direction.
The bill identification method according to claim 1, wherein after acquiring the image of the bill, the bill identification method further comprises:

identifying the edge of the note; and,

Based on the recognition result, the image of the bill is trimmed.
The bill identification method according to claim 1, wherein after acquiring the image of the bill, the bill identification method further comprises:

The image content of the bill is corrected, and the correction includes global correction and local correction.
A bill identification device, characterized in that it includes:

The image preprocessing module is used for preprocessing the original image containing the bill, and the preprocessing includes: after scaling the length of the original image in the first direction to a preset first node size, The length of the original image in the second direction is padded to a preset second node size, wherein the first direction is perpendicular to the second direction, and the original image is in the first direction The length is not less than the length in the second direction;

An annotation frame acquisition and adjustment module, used for acquiring the annotation frame of the bill, and enlarging the annotation frame of the bill according to a preset ratio;

The image post-processing module is used for segmenting the image of the ticket based on the enlarged annotation frame of the ticket, and outputting the segmented image of the ticket.
The bill identification device according to claim 9, wherein the bill identification device further comprises a node size setting module, and the node size setting module is used to preset a plurality of the first node sizes and a plurality of the a second node size, the first node size and the second node size are in one-to-one correspondence;

When scaling the original image, the image preprocessing module scales the length of the original image in the first direction to the first node size with the closest value; The length of the original image in the second direction is padded to the second node size corresponding to the scaled first node size.
The bill identification device of claim 9, wherein the size of the first node and the size of the second node are the same.
The bill recognition device according to claim 9, wherein the method for the image preprocessing module to complement the length of the scaled original image in the second direction to a preset second node size comprises: Blank areas are supplemented at the sides of the scaled original image along the second direction.
The bill recognition device according to claim 9, wherein the image post-processing module comprises an image segmentation module and an image output module, and the image segmentation module is configured to segment out the the image of the bill, and the image output module is configured to output the image of the bill that has been segmented.
The bill recognition device according to claim 13, wherein the image post-processing module further comprises an orientation adjustment module, and the orientation adjustment module is used to adjust the orientation of the image of the bill, so that the The orientation of the characters is the preset direction.
The bill identification device according to claim 13, wherein the image post-processing module further comprises an edge processing module, the edge processing module is used for identifying the edge of the bill, and based on the identification result, image is trimmed.
The bill recognition device according to claim 13, wherein the image post-processing module further comprises an image correction module, the image correction module is used to correct the image content of the bill, and the correction includes global correction and local correction.
A readable storage medium, characterized in that the readable storage medium stores a computer program, and when the computer program is executed, the bill identification method according to any one of claims 1 to 8 is implemented.