CN116935398A - Draft information identification method, device, equipment and medium - Google Patents

Draft information identification method, device, equipment and medium Download PDF

Info

Publication number
CN116935398A
CN116935398A CN202310911695.8A CN202310911695A CN116935398A CN 116935398 A CN116935398 A CN 116935398A CN 202310911695 A CN202310911695 A CN 202310911695A CN 116935398 A CN116935398 A CN 116935398A
Authority
CN
China
Prior art keywords
sub
text
sequence
image
coordinates
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310911695.8A
Other languages
Chinese (zh)
Inventor
张文宇
卜丽
陆佳庆
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Construction Bank Corp
CCB Finetech Co Ltd
Original Assignee
China Construction Bank Corp
CCB Finetech Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China Construction Bank Corp, CCB Finetech Co Ltd filed Critical China Construction Bank Corp
Priority to CN202310911695.8A priority Critical patent/CN116935398A/en
Publication of CN116935398A publication Critical patent/CN116935398A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10—Character recognition
    • G06V30/14—Image acquisition
    • G06V30/148—Segmentation of character regions
    • G06V30/153—Segmentation of character regions using recognition of characters or words
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10—Character recognition
    • G06V30/19—Recognition using electronic means
    • G06V30/191—Design or setup of recognition systems or techniques; Extraction of features in feature space; Clustering techniques; Blind source separation
    • G06V30/19173—Classification techniques

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • General Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Databases & Information Systems (AREA)
  • Medical Informatics (AREA)
  • Character Input (AREA)

Abstract

The present application relates to the field of image processing technologies, and in particular, to a method, an apparatus, a device, and a medium for identifying draft information. The method is used for solving the problem of low auditing efficiency when auditing the draft in the prior art. According to the embodiment of the application, the electronic equipment identifies the text sequence in the image containing the draft to be identified through the identification model, and obtains each audit information and the type of each audit information in the draft, so that each audit information and the type of each audit information for audit can be rapidly and accurately screened, further, the audit of subsequent audit personnel is facilitated, and the audit efficiency is improved.

Description

Draft information identification method, device, equipment and medium
Technical Field
The present application relates to the field of image processing technologies, and in particular, to a method, an apparatus, a device, and a medium for identifying draft information.
Background
An international settlement draft is a settlement draft for monetary bonds and liabilities between countries due to trade and non-trade economic associations, and a domestic settlement draft is a draft issued and paid in the home country. In international trade, the drawer of the draft is typically an exporter, the payer is typically an importer or an issuer, and the payee is typically a bargained, a collection or an exporter. The international settlement draft generally appears in banking operations such as credit card payment and export collection, and the information in the draft issued by each bank is the same or different, and generally the included information includes a draft word, a drawer signature, a payer name, a payee name, an unconditional payment order, a definite amount, a ticket date, a ticket place, a payment date, a payment place, and any recorded item generally includes a draft number, a ticket term, a collection term, a price term, and the like.
When a bank audit person examines the draft, the important audit information in the draft is generally read and understood in a purely manual mode according to the years of audit experience. The audit personnel screens important audit information in the draft, so that the problem of low audit efficiency can occur. The method consumes a great deal of labor cost, and the period for culturing one inspector often needs more than three years. In the related technology, a rule-based method is also provided for auditing, and generally, an auditor summarizes audit rules, summarizes common forms of audit information of all types in a regular expression form, and identifies audit information of all types in a draft by using a rule matching mode. However, the auditing information of each type is difficult to be exhausted by using rules, so that the screened auditing information is inaccurate, and the auditing efficiency is reduced.
Disclosure of Invention
The embodiment of the application provides a draft information identification method, device, equipment and medium, which are used for solving the problem of low auditing efficiency when auditing a draft in the prior art.
The embodiment of the application provides a draft information identification method, which comprises the following steps:
receiving an image containing a draft to be identified;
Identifying a text sequence in the image; inputting the text sequence into a pre-trained recognition model, and acquiring each audit information and the type of each audit information contained in the text sequence output by the recognition model;
and outputting the type of each audit information.
Further, the identifying the text sequence in the image includes:
determining each sub-text sequence contained in the image and coordinates of a circumscribed rectangle of each sub-text sequence in the image through a text recognition (Optical Character Recognition, OCR) technology;
and according to the coordinates of the circumscribed rectangle of each sub-character sequence in the image, arranging characters in each sub-character sequence to generate a character sequence.
Further, the determining, by the character recognition OCR technology, coordinates of each sub-character sequence in the image and a circumscribed rectangle of each sub-character sequence in the image includes:
determining each character in the image and the coordinates of the circumscribed rectangle of each character in the image by an OCR technology;
determining the distance between the circumscribed rectangles of each character, and determining the characters with the distance smaller than the preset distance as characters in the same sub-character sequence;
And determining the coordinates of the circumscribed rectangle of the sub-text sequence in the image according to the coordinates of the circumscribed rectangle of each text in the sub-text sequence aiming at each sub-text sequence.
Further, the arranging the characters in each sub-character sequence according to the coordinates of the circumscribed rectangle of each sub-character sequence in the image to generate the character sequence includes:
identifying the sub-text sequences in the same row according to the distance between the ordinate of the circumscribed rectangles of the sub-text sequences, and sequentially arranging the sub-text sequences in the same row according to the size of the abscissa of the circumscribed rectangles of the sub-text sequences in the same row to obtain the sub-text sequences in each row;
and splicing the sub-text sequences of each row in sequence according to the size sequence of the ordinate to obtain the text sequence.
Further, the identifying the sub-text sequences located in the same row according to the distance between the ordinate of the circumscribed rectangle of the sub-text sequences includes:
and determining the sub-text sequences with the distance between the ordinate of the circumscribed rectangle smaller than the preset value as the sub-text sequences positioned in the same row.
Further, the preset value is determined by:
Determining a first height of the sub-text sequence according to coordinates of the circumscribed rectangle of the sub-text sequence in the image;
determining a second height of the other sub-text sequences according to coordinates of the circumscribed rectangles of the other sub-text sequences in the image;
and determining a preset value according to the sum value of the first height and the second height.
Further, the identification model is a conditional random field (Conditional random field, CRF) model.
Further, the recognition model is trained by:
any sample sequence in a sample set is obtained, and each standard audit information and the standard type of each standard audit information contained in the sample sequence are obtained;
inputting the sample sequence into an original recognition model, and acquiring each output audit information and the output type of each output audit information contained in the sample sequence;
and training the original recognition model according to the standard audit information, the output audit information, the standard type and the output type.
The embodiment of the application also provides a draft information identification device, which comprises:
The receiving module is used for receiving the image containing the draft to be identified;
the processing module is used for identifying the text sequence in the image; inputting the text sequence into a pre-trained recognition model, and acquiring each audit information and the type of each audit information contained in the text sequence output by the recognition model;
and the output module is used for outputting each audit information and the type of each audit information.
Further, the processing module is specifically configured to determine, through a text recognition OCR technology, coordinates of each sub-text sequence included in the image and a circumscribed rectangle of each sub-text sequence in the image; and according to the coordinates of the circumscribed rectangle of each sub-character sequence in the image, arranging characters in each sub-character sequence to generate a character sequence.
Further, the processing module is specifically configured to determine, through OCR technology, coordinates of each text in the image and a rectangle circumscribing each text in the image; determining the distance between the circumscribed rectangles of each character, and determining the characters with the distance smaller than the preset distance as characters in the same sub-character sequence; and determining the coordinates of the circumscribed rectangle of the sub-text sequence in the image according to the coordinates of the circumscribed rectangle of each text in the sub-text sequence aiming at each sub-text sequence.
Further, the processing module is specifically configured to identify the sub-text sequences located in the same row according to a distance between the ordinate of the circumscribed rectangles of the sub-text sequences, and sequentially arrange the sub-text sequences located in the same row according to the size of the abscissa of the circumscribed rectangles of the sub-text sequences located in the same row, so as to obtain the sub-text sequences of each row; and splicing the sub-text sequences of each row in sequence according to the size sequence of the ordinate to obtain the text sequence.
Further, the processing module is specifically configured to determine, as a sub-text sequence located in the same row, a sub-text sequence with a distance between the ordinate of the circumscribed rectangle smaller than a preset value.
Further, the processing module is further configured to determine a first height of the sub-text sequence according to coordinates of an circumscribed rectangle of the sub-text sequence in the image; determining a second height of the other sub-text sequences according to coordinates of the circumscribed rectangles of the other sub-text sequences in the image; and determining a preset value according to the sum value of the first height and the second height.
Further, the processing module is further configured to obtain any sample sequence in the sample set, and each standard audit information and a standard type of each standard audit information included in the sample sequence; inputting the sample sequence into an original recognition model, and acquiring each output audit information and the output type of each output audit information contained in the sample sequence; and training the original recognition model according to the standard audit information, the output audit information, the standard type and the output type.
The embodiment of the application also provides electronic equipment, which at least comprises a processor and a memory, wherein the processor is used for realizing the steps of the draft information identification method according to any one of the above steps when executing the computer program stored in the memory.
The embodiment of the application also provides a computer readable storage medium storing a computer program, which when executed by a processor, implements the steps of the draft information identification method according to any one of the above.
Embodiments of the present application also provide a computer program product comprising: computer program code which, when run on a computer, causes the computer to perform the steps of the draft information identification method as set forth in any one of the preceding claims.
Because in the embodiment of the application, the electronic equipment receives the image containing the draft to be identified; recognizing a text sequence in the image; inputting the text sequence into a recognition model which is trained in advance, and acquiring each audit information and the type of each audit information contained in the text sequence output by the recognition model; outputting each audit information and the type of each audit information. According to the embodiment of the application, the electronic equipment identifies the text sequence in the image containing the draft to be identified through the identification model, and obtains each audit information and the type of each audit information in the draft, so that each audit information and the type of each audit information for audit can be rapidly and accurately screened, further, the audit of subsequent audit personnel is facilitated, and the audit efficiency is improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic diagram of a draft information identification process according to an embodiment of the present application;
FIG. 2 is a schematic diagram illustrating a detailed process for identifying draft information according to an embodiment of the present application;
FIG. 3 is a schematic diagram of a draft information identification apparatus according to an embodiment of the present application;
fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Detailed Description
The present application will be described in further detail below with reference to the attached drawings, wherein it is apparent that the embodiments described are only some, but not all embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
In order to improve the efficiency of checking the draft, the embodiment of the application provides a draft information identification method, device, equipment and medium.
The draft information identification method comprises the following steps: the electronic equipment receives an image containing a draft to be identified; recognizing a text sequence in the image; inputting the text sequence into a recognition model which is trained in advance, and acquiring each audit information and the type of each audit information contained in the text sequence output by the recognition model; outputting each audit information and the type of each audit information. Therefore, the draft information is accurately identified, and the auditing efficiency is improved.
Example 1:
fig. 1 is a schematic diagram of a draft information identification process according to an embodiment of the present application, where the process includes the following steps:
s101: an image is received that includes a draft to be identified.
The draft information identification method provided by the embodiment of the application is applied to electronic equipment, and the electronic equipment can be intelligent equipment such as a PC (personal computer) or a server.
For draft information identification, the electronic device may identify the received image containing the draft to be identified. Specifically, after the to-be-identified draft is shot by the inspector through the acquisition device, the image containing the to-be-identified draft is acquired, the image containing the to-be-identified draft is sent to the electronic device through the self-used device, and the electronic device can receive the image containing the to-be-identified draft.
Wherein, the draft is a written unconditional payment order issued by one person to another, requiring the other party (the person receiving the order) to pay a certain amount to someone or the designated person or the ticket holder, either on demand or periodically or at a determinable future time.
S102: identifying a text sequence in the image; inputting the text sequence into a pre-trained recognition model, and acquiring each audit information and the type of each audit information contained in the text sequence output by the recognition model.
In order to accurately identify draft information, the electronic device may identify the text sequence in the image first, specifically, the electronic device may locally store a model of the pre-trained identified text sequence, and the electronic device may input the received image into the model of the pre-trained identified text sequence, and obtain an output of the model of the identified text sequence, where the output is the text sequence in the image. The text sequence is an ordered set of text components in the image.
In order to accurately identify draft information, the electronic equipment locally stores a pre-trained identification model, after recognizing a text sequence in an image, the electronic equipment can input the text sequence into the pre-trained identification model and acquire the output of the identification model, wherein the output of the identification model is the audit information and the type of the audit information contained in the text sequence.
Table 1 shows the types of information included in a draft provided in an embodiment of the present application.
TABLE 1
Wherein the first column in table 1 is the type of information contained in the draft and the second column in table 1 is an explanation of the type of the first column. As can be seen from table 1, doc_time refers to document TITLE; draft_no refers to DRAFT number; draft_date refers to the DATE of the DRAFT; a draft_issue_place point of sale; payment_term refers to the TERM of the draft period; draw_under_clauise refers to the terms of the ticket; draft_amount_word refers to the DRAFT AMOUNT; draft_AMOUNT_FOGURE refers to DRAFT AMOUNT capitalization; drawee_name refers to the NAME of the payer; drawee_add refers to the payer address; payname refers to the PAYEE NAME; payee_add refers to the PAYEE address; ISSUER_NAME refers to the NAME of the person who has issued; ISSUER_ADD refers to the address of the person; drawnder_issue_bank refers to the issuing-issuing BANK; drawnder LC NO refers to the ticket-letter of the credit number; drawnder LC DATE refers to the ticket-issuing DATE; ENDORSE refers to endorsement; CONTRACT_NO refers to the CONTRACT number; CONTRACT_DATE refers to the CONTRACT DATE; lc_no refers to a letter number; lc_date refers to the DATE of the issuance; OTHER references are made to OTHER content.
S103: and outputting the type of each audit information.
After each audit information and the type of each audit information included in the text sequence are obtained, the electronic device may output each audit information and the type of each audit information. Specifically, the audit information and the type of the audit information can be directly displayed on a display interface of the electronic device, and the audit information and the type of the audit information can be sent to the device used by the audit personnel during working, so that the audit by the audit personnel is facilitated, and the audit efficiency is improved.
The audit information is part of information contained in the draft. That is, the information for auditing among the information contained in the draft is audit information.
It should be noted that, the characters included in the character sequence are characters in the draft to be identified, so that each audit information and each audit information type included in the character sequence are each audit information and each audit information type in the draft to be identified. That is to say, by adopting the method provided by the embodiment of the application, each audit information and the type of each audit information in the draft to be identified can be identified.
According to the embodiment of the application, the electronic equipment identifies the text sequence in the image containing the draft to be identified through the identification model, and obtains each audit information and the type of each audit information in the draft, so that each audit information and the type of each audit information for audit can be rapidly and accurately screened, further, the audit of subsequent audit personnel is facilitated, and the audit efficiency is improved.
Example 2:
in order to accurately identify the text sequence in the image, in the embodiment of the present application, the identifying the text sequence in the image includes:
determining each sub-text sequence contained in the image and the coordinates of the circumscribed rectangle of each sub-text sequence in the image through an OCR technology;
and according to the coordinates of the circumscribed rectangle of each sub-character sequence in the image, arranging characters in each sub-character sequence to generate a character sequence.
The character recognition is a process that the electronic equipment checks characters printed on paper and then uses a character recognition method to translate the characters into computer characters; the text data is scanned, and then the image is analyzed and processed to obtain the text and layout information.
To accurately identify the text sequences in the image, the electronic device may identify the image based on OCR technology, determine the coordinates of each sub-text sequence contained in the image, and the circumscribed rectangle of each sub-text sequence in the image. Wherein, adjacent characters in the image are characters in the same sub-character sequence. Specifically, since the draft template is fixed, that is, what type of information is filled in each region of the draft is fixed, the electronic device can recognize each text in the region by OCR technology for each preset region, each text shown in the region is a corresponding sub-text sequence of the region, and coordinates of the region in the image are coordinates of circumscribed rectangles of the corresponding sub-text sequence in the image. Specifically, the coordinates of the circumscribed rectangle in the image may be the coordinates of the center point of the circumscribed rectangle in the image and the length and width of the circumscribed rectangle.
After determining each sub-text sequence contained in the image and coordinates of the circumscribed rectangle of each sub-text sequence in the image, the electronic device may arrange each sub-text sequence according to the coordinates of the circumscribed rectangle of each sub-text sequence in the image to generate a text sequence. Specifically, the electronic device may sequentially arrange the characters in the sub-character sequence according to the size of the abscissa of the circumscribed rectangle of the sub-character sequence in the image, and if the abscissas are the same, sequentially arrange the characters in the sub-character sequence according to the size of the ordinate of the circumscribed rectangle of the sub-character sequence in the image.
In the embodiment of the application, the electronic equipment identifies each audit information and the type of each audit information in the image containing the draft to be identified by combining the identification model with the OCR technology, so that the audit is conveniently carried out by an audit person. Compared with the traditional offline manual bill auditing mode, the bill auditing method has higher accuracy and recall rate, and can improve the auditing efficiency during bill auditing.
Example 3:
in order to determine the coordinates of each sub-text sequence and the circumscribed rectangle of each sub-text sequence in the image, in the embodiments of the present application, the determining the coordinates of each sub-text sequence and the circumscribed rectangle of each sub-text sequence in the image by the text recognition OCR technology includes:
determining each character in the image and the coordinates of the circumscribed rectangle of each character in the image by an OCR technology;
determining the distance between the circumscribed rectangles of each character, and determining the characters with the distance smaller than the preset distance as characters in the same sub-character sequence;
and determining the coordinates of the circumscribed rectangle of the sub-text sequence in the image according to the coordinates of the circumscribed rectangle of each text in the sub-text sequence aiming at each sub-text sequence.
In order to determine each sub-text sequence and the coordinates of the circumscribed rectangle of each sub-text sequence in the image, the electronic device may determine each text in the received image and the coordinates of the circumscribed rectangle of each text in the image by using an OCR technology, and specifically, how the electronic device determines each text in the received image and the coordinates of the circumscribed rectangle of each text in the image by using the OCR technology are the prior art, which is not described herein. The coordinates of the circumscribed rectangle of the character in the image can be the coordinates of the center point of the circumscribed rectangle of the character in the image, the height and width of the circumscribed rectangle, or the coordinates of the points of the upper left corner and the lower right corner of the circumscribed rectangle of the character in the image.
After determining the coordinates of each character and the circumscribed rectangle of each character in the image, the electronic equipment can determine the distance between the circumscribed rectangles of each character, and if the coordinates of the circumscribed rectangle of each character in the image are the coordinates of the center point of the circumscribed rectangle of each character in the image and the height and width of the circumscribed rectangle, the electronic equipment can determine the distance between the center points of the circumscribed rectangles of two characters in the image as the distance between the circumscribed rectangles of the two characters; if the coordinates of the circumscribed rectangles of the characters in the image are the coordinates of the points of the upper left corner and the lower right corner of the circumscribed rectangles of the characters in the image, the electronic equipment can determine the distance between the points of the upper left corner of the circumscribed rectangles of the two characters as the distance between the circumscribed rectangles of the two characters when determining the distance between the circumscribed rectangles of the two characters. Specifically, the coordinates of two circumscribed rectangles are known, and how to determine the distance between the two circumscribed rectangles is the prior art, which is not described herein.
After determining the distance between the circumscribed rectangles of each word, the electronic device may determine the word having a distance between the circumscribed rectangles less than the preset distance as a word in the same sub-word sequence. After determining each sub-text sequence, the electronic device may determine, for each sub-text sequence, coordinates of the circumscribed rectangle of each text in the sub-text sequence in the image according to the coordinates of the circumscribed rectangle of the text.
If the coordinates of the circumscribed rectangles of the characters in the image are the coordinates of the center points of the circumscribed rectangles of the characters in the image and the height and width of the circumscribed rectangles, when the electronic equipment determines the coordinates of the circumscribed rectangles of a certain sub-character sequence in the image, the electronic equipment can determine the average value of the horizontal coordinates of the center points of the circumscribed rectangles of each character contained in the sub-character sequence in the image, wherein the average value is the horizontal coordinates of the center points of the circumscribed rectangles of the sub-character sequence, namely the horizontal coordinates of the circumscribed rectangles of the sub-character sequence, and the electronic equipment can determine the average value of the vertical coordinates of the center points of the circumscribed rectangles of each character contained in the sub-character sequence in the image, namely the vertical coordinates of the center points of the circumscribed rectangles of the sub-character sequence; the electronic equipment can acquire a first character with the smallest abscissa of the circumscribed rectangle in each character contained in the sub-character sequence; obtaining a second character with the largest abscissa of the circumscribed rectangle in each character contained in the sub character sequence; determining the sum of the deviation of the abscissa of the circumscribed rectangle of the first character and the second character, half of the width of the circumscribed rectangle of the first character and half of the width of the circumscribed rectangle of the second character as the width of the circumscribed rectangle of the sub character sequence; the electronic equipment can acquire a third character with the smallest ordinate of the circumscribed rectangle in each character contained in the sub character sequence; obtaining a fourth character with the largest ordinate of the circumscribed rectangle in each character contained in the sub character sequence; and determining the sum of the deviation of the ordinate of the circumscribed rectangle of the third character and the fourth character, half of the width of the circumscribed rectangle of the third character and half of the width of the circumscribed rectangle of the fourth character as the width of the circumscribed rectangle of the sub character sequence.
If the coordinates of the circumscribed rectangle of the characters in the image are the coordinates of the points of the left upper corner and the right lower corner of the circumscribed rectangle of the characters in the image, when the electronic equipment determines the coordinates of the circumscribed rectangle of a certain sub-character sequence in the image, the electronic equipment can determine the minimum abscissa of the points of the left upper corner of the circumscribed rectangle of each character in the sub-character sequence as the abscissa of the points of the left upper corner of the circumscribed rectangle of the sub-character sequence in the image, and if the abscissas of the points of the left upper corner of the circumscribed rectangle of a plurality of characters in the sub-character sequence are minimum, the maximum ordinate of the points of the left upper corner of the smallest abscissas of the plurality of abscissas is determined as the ordinate of the points of the left upper corner of the circumscribed rectangle of the sub-character sequence in the image; and determining the minimum ordinate of the point at the lower right corner of the circumscribed rectangle of each character in the sub-character sequence as the ordinate of the point at the lower right corner of the circumscribed rectangle of the sub-character sequence in the image, and if the ordinate of the point at the lower right corner of the circumscribed rectangle of a plurality of characters in the sub-character sequence is minimum, determining the maximum abscissa of the points at the lower right corner of the plurality of characters by the electronic equipment as the abscissa of the point at the lower right corner of the circumscribed rectangle of the sub-character sequence in the image.
Example 4:
in order to accurately determine the text sequence, based on the above embodiments, in the embodiment of the present application, the arranging the text in each sub-text sequence according to the coordinates of the circumscribed rectangle of each sub-text sequence in the image to generate the text sequence includes:
identifying the sub-text sequences in the same row according to the distance between the ordinate of the circumscribed rectangles of the sub-text sequences, and sequentially arranging the sub-text sequences in the same row according to the size of the abscissa of the circumscribed rectangles of the sub-text sequences in the same row to obtain the sub-text sequences in each row;
and splicing the sub-text sequences of each row in sequence according to the size sequence of the ordinate to obtain the text sequence.
After determining each sub-text sequence in the image, the electronic device may identify sub-text sequences located in the same line according to a distance between the ordinate of the circumscribed rectangle of each sub-text sequence in the image, and specifically, the electronic device may determine a sub-text sequence with a distance of 0 between the ordinate of the circumscribed rectangle of the sub-text sequences as a sub-text sequence located in the same line. After determining the sub-text sequences located in the same row, the electronic device may sequentially arrange the sub-text sequences located in the same row according to the size of the abscissa of the circumscribed rectangle of the sub-text sequences located in the same row, to obtain the sub-text sequences of each row.
After the sub-text sequences of each row are obtained, the electronic device can splice the sub-text sequences of each row in sequence according to the size sequence of the ordinate, so as to obtain the text sequences.
In order to accurately identify the sub-text sequences located in the same row, based on the above embodiments, in the embodiments of the present application, identifying the sub-text sequences located in the same row according to the distance between the ordinate of the circumscribed rectangle of the sub-text sequences includes:
and determining the sub-text sequences with the distance between the ordinate of the circumscribed rectangle smaller than the preset value as the sub-text sequences positioned in the same row.
In order to accurately identify the sub-text sequences located in the same row, the electronic device locally stores a preset value, and the electronic device can determine the sub-text sequences with the distance between the ordinate coordinates of the circumscribed rectangles smaller than the preset value as the sub-text sequences located in the same row.
Specifically, if the coordinates of the circumscribed rectangle of the sub-text sequence in the image are the coordinates of the center point of the circumscribed rectangle of the sub-text sequence in the image, the electronic device may determine the sub-text sequence with the distance between the ordinate of the center point of the circumscribed rectangle of the sub-text sequence smaller than the preset value as the sub-text sequence located in the same row. If the coordinates of the outer rectangle of the sub-word sequence in the image are the coordinates of the points of the left upper corner and the right lower corner of the outer rectangle of the sub-word sequence in the image, the electronic device can determine the sub-word sequence with the distance between the points of the left upper corner of the outer rectangle of the sub-word sequence smaller than a preset value as the sub-word sequence positioned in the same row, and the electronic device can determine the sub-word sequence with the distance between the points of the right lower corner of the outer rectangle of the sub-word sequence smaller than the preset value as the sub-word sequence positioned in the same row.
Fig. 2 is a schematic diagram of a detailed process for identifying draft information according to an embodiment of the present application, where the process includes the following steps:
s201: an image is received that includes a draft to be identified.
S202: and determining each sub-character sequence contained in the image and the coordinates of the circumscribed rectangle of each sub-character sequence in the image through an OCR technology.
S203: and identifying the sub-text sequences positioned in the same row according to the distance between the ordinate of the circumscribed rectangles of the sub-text sequences.
S204: and according to the size of the abscissa of the circumscribed rectangle of the sub-text sequences positioned in the same row, the sub-text sequences positioned in the same row are sequentially arranged to obtain the sub-text sequences of each row.
S205: and splicing the sub-text sequences of each row in sequence according to the size sequence of the ordinate to obtain the text sequence.
S206: inputting the text sequence into a recognition model which is trained in advance, and acquiring each audit information and the type of each audit information contained in the text sequence output by the recognition model.
S207: outputting each audit information and the type of each audit information.
In order to accurately identify the sub-text sequences located in the same line, based on the above embodiments, in the embodiments of the present application, the preset value is determined by:
For each sub-text sequence, determining a first height of the sub-text sequence according to coordinates of an external rectangle of the sub-text sequence in the image; determining a second height of the other sub-text sequences according to coordinates of the circumscribed rectangles of the other sub-text sequences in the image; and determining the preset numerical values corresponding to the sub-text sequences and the other sub-text sequences according to the sum value of the first height and the second height.
In order to accurately identify the sub-text sequences located in the same row, the electronic device may determine, for each sub-text sequence, a first height of the sub-text sequence according to coordinates of an external rectangle of the sub-text sequence in the image, and if the coordinates of the external rectangle of the sub-text sequence in the image are coordinates of points of an upper left corner and a lower right corner of the external rectangle of the sub-text sequence in the image, the electronic device may determine a deviation of ordinate coordinates of the upper left corner and the lower right corner of the external rectangle of the sub-text sequence as the first height of the sub-text sequence. The electronic device may further determine the second height of the other sub-text sequences according to coordinates of the circumscribed rectangle of the other sub-text sequences in the image, and specifically how to determine the height of a certain sub-text sequence is described in the embodiment of the present application and is not described herein.
After determining the first height of the sub-text sequence and the second heights of other sub-text sequences, the electronic device may determine a sum of the first height and the second height, and determine a preset value corresponding to the sub-text sequence included in the draft according to the sum. Specifically, the electronic device may determine a product of the sum and a preset value, where the preset value is a value greater than 0 and less than 1; and determining the product as a preset value corresponding to the sub-text sequence and the other sub-text sequences, wherein the preset value can be 1/4.
Example 5:
in order to accurately identify draft information, in the embodiments of the present application, the identification model is a CRF model.
The identification model provided by the embodiment of the application can be a CRF model.
The CRF has strong reasoning capability, can train and infer by using complex, overlapped and non-independent features, can fully utilize the context information as the features, and can also add other external features arbitrarily, so that the information which can be acquired by the recognition model is very rich. In addition, the CRF model has certain advantages in combining various characteristics, so that the problem of label bias is effectively avoided. The CRF model is a differential probability model that is commonly used to label or analyze sequence data, such as natural language text or biological sequences. The conditional random field is an undirected graph model, vertexes in the undirected graph represent random variables, connection lines among the vertexes represent the dependency relationship among the random variables, in the conditional random field, the distribution of the random variables is conditional probability, and given observation values are the random variables.
The embodiment of the application combines the bill characteristics of the draft and adopts the CRF model to identify the draft information. For the image containing the draft to be checked, the image is firstly recognized by OCR technology to obtain a character sequence, and then each audit information and the type of each audit information in the character sequence are obtained by a trained CRF model, so that a good recognition effect is obtained on the basis of not needing massive training samples.
In order to accurately identify draft information, in the embodiments of the present application, the identification model is trained by:
any sample sequence in a sample set is obtained, and each standard audit information and the standard type of each standard audit information contained in the sample sequence are obtained;
inputting the sample sequence into an original recognition model, and acquiring each output audit information and the output type of each output audit information contained in the sample sequence;
and training the original recognition model according to the standard audit information, the output audit information, the standard type and the output type.
In order to realize training of the recognition model, a sample set for training is stored in the embodiment of the application, and a sample sequence in the sample set comprises different characters. In order to facilitate training of the original recognition model, the sample set stores, for each sample sequence, each standard audit information and a standard type of each standard audit information contained in the sample sequence.
In the embodiment of the application, after any sample sequence in a sample set and each standard audit information and standard type of each standard audit information contained in the sample sequence are acquired, each standard audit information and standard type of each standard audit information contained in the sample sequence and the sample sequence are input into an original identification model, and the original identification model outputs each output audit information and output type of each output audit information contained in the sample sequence. The business personnel define a set of independent information type according to the basic information of the draft and the audit information of the audit personnel, so that the follow-up manual marking is facilitated, and further the training of the identification model is performed.
After the original recognition model determines each output audit information and the output type of each output audit information contained in the sample sequence, training the original recognition model according to each standard audit information and the standard type of each standard audit information in the sample sequence, and each output audit information and the output type of each output audit information output by the original recognition model.
The original recognition model is trained in the mode, and when preset conditions are met, the trained original recognition model is obtained. The preset condition may be that the number of each output audit information and each output audit information output type obtained after the sample sequence in the sample set is trained by the original recognition model, and each standard audit information standard type is greater than the set number; or the iteration number of training the original recognition model reaches the set maximum iteration number. In particular, embodiments of the present application are not limited in this regard.
The application is equivalent to determining a set of independent draft type and system for determining audit information according to the bill layout characteristics and basic contents of the draft.
Example 6:
fig. 3 is a schematic structural diagram of a draft information identifying apparatus according to an embodiment of the present application, where the apparatus includes:
a receiving module 301, configured to receive an image including a draft to be identified;
a processing module 302, configured to identify a text sequence in the image; inputting the text sequence into a pre-trained recognition model, and acquiring each audit information and the type of each audit information contained in the text sequence output by the recognition model;
and the output module 303 is configured to output each audit information and the type of each audit information.
In a possible implementation manner, the processing module 302 is specifically configured to determine, by using OCR technology, coordinates of each sub-text sequence included in the image and a rectangle circumscribing each sub-text sequence in the image; and according to the coordinates of the circumscribed rectangle of each sub-character sequence in the image, arranging characters in each sub-character sequence to generate a character sequence.
In a possible implementation manner, the processing module 302 is specifically configured to determine, by OCR technology, coordinates of each text in the image and a rectangle circumscribing each text in the image; determining the distance between the circumscribed rectangles of each character, and determining the characters with the distance smaller than the preset distance as characters in the same sub-character sequence; and determining the coordinates of the circumscribed rectangle of the sub-text sequence in the image according to the coordinates of the circumscribed rectangle of each text in the sub-text sequence aiming at each sub-text sequence.
In a possible implementation manner, the processing module 302 is specifically configured to identify the sub-text sequences located in the same row according to a distance between the ordinate of the circumscribed rectangles of the sub-text sequences, and sequentially arrange the sub-text sequences located in the same row according to the size of the abscissa of the circumscribed rectangles of the sub-text sequences located in the same row, so as to obtain the sub-text sequences of each row; and splicing the sub-text sequences of each row in sequence according to the size sequence of the ordinate to obtain the text sequence.
In a possible implementation manner, the processing module 302 is specifically configured to determine, as the sub-text sequences located in the same row, the sub-text sequences with the distance between the ordinate of the circumscribed rectangle being smaller than the preset value.
In a possible implementation manner, the processing module 302 is further configured to determine a first height of the sub-text sequence according to coordinates of a circumscribed rectangle of the sub-text sequence in the image; determining a second height of the other sub-text sequences according to coordinates of the circumscribed rectangles of the other sub-text sequences in the image; and determining a preset value according to the sum value of the first height and the second height.
In a possible implementation manner, the processing module 302 is further configured to obtain any one of the sample sequences in the sample set, and each standard audit information and a standard type of each standard audit information included in the sample sequence; inputting the sample sequence into an original recognition model, and acquiring each output audit information and the output type of each output audit information contained in the sample sequence; and training the original recognition model according to the standard audit information, the output audit information, the standard type and the output type.
Example 7:
fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present application, and on the basis of the foregoing embodiments, the embodiment of the present application further provides an electronic device, as shown in fig. 4, including: the processor 401, the communication interface 402, the memory 403 and the communication bus 404, wherein the processor 401, the communication interface 402 and the memory 403 complete communication with each other through the communication bus 404;
the memory 403 has stored therein a computer program which, when executed by the processor 401, causes the processor 401 to perform the steps of:
Receiving an image containing a draft to be identified;
identifying a text sequence in the image; inputting the text sequence into a pre-trained recognition model, and acquiring each audit information and the type of each audit information contained in the text sequence output by the recognition model;
and outputting the type of each audit information.
Further, the processor 401 is specifically configured to determine, by OCR, coordinates of each sub-text sequence included in the image and a rectangle circumscribing each sub-text sequence in the image;
and according to the coordinates of the circumscribed rectangle of each sub-character sequence in the image, arranging characters in each sub-character sequence to generate a character sequence.
Further, the processor 401 is specifically configured to determine, by OCR, coordinates of each text in the image and a rectangle circumscribing each text in the image;
determining the distance between the circumscribed rectangles of each character, and determining the characters with the distance smaller than the preset distance as characters in the same sub-character sequence;
and determining the coordinates of the circumscribed rectangle of the sub-text sequence in the image according to the coordinates of the circumscribed rectangle of each text in the sub-text sequence aiming at each sub-text sequence.
Further, the processor 401 is specifically configured to identify the sub-text sequences located in the same row according to a distance between the ordinate of the circumscribed rectangles of the sub-text sequences, and sequentially arrange the sub-text sequences located in the same row according to the size of the abscissa of the circumscribed rectangles of the sub-text sequences located in the same row, so as to obtain the sub-text sequences of each row;
and splicing the sub-text sequences of each row in sequence according to the size sequence of the ordinate to obtain the text sequence.
Further, the processor 401 is specifically configured to determine the sub-text sequences with the distance between the ordinate of the circumscribed rectangle smaller than the preset value as the sub-text sequences located in the same row.
Further, the processor 401 is further configured to determine a first height of the sub-text sequence according to coordinates of a circumscribed rectangle of the sub-text sequence in the image;
determining a second height of the other sub-text sequences according to coordinates of the circumscribed rectangles of the other sub-text sequences in the image;
and determining a preset value according to the sum value of the first height and the second height.
Further, the processor 401 is further configured to obtain any sample sequence in a sample set, and each standard audit information and a standard type of each standard audit information included in the sample sequence;
Inputting the sample sequence into an original recognition model, and acquiring each output audit information and the output type of each output audit information contained in the sample sequence;
and training the original recognition model according to the standard audit information, the output audit information, the standard type and the output type.
The communication bus mentioned by the server may be a peripheral component interconnect standard (Peripheral Component Interconnect, PCI) bus or an extended industry standard architecture (Extended Industry Standard Architecture, EISA) bus, etc. The communication bus may be classified as an address bus, a data bus, a control bus, or the like. For ease of illustration, the figures are shown with only one bold line, but not with only one bus or one type of bus.
The communication interface is used for communication between the electronic device and other devices.
The Memory may include random access Memory (Random Access Memory, RAM) or may include Non-Volatile Memory (NVM), such as at least one disk Memory. Optionally, the memory may also be at least one memory device located remotely from the aforementioned processor.
The processor may be a general-purpose processor, including a central processing unit, a network processor (Network Processor, NP), etc.; but also digital instruction processors (Digital Signal Processing, DSP), application specific integrated circuits, field programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
Example 8:
on the basis of the above embodiments, the embodiments of the present invention further provide a computer readable storage medium having stored therein a computer program executable by an electronic device, which when run on the electronic device, causes the electronic device to perform the steps of:
the memory has stored therein a computer program which, when executed by the processor, causes the processor to perform the steps of:
receiving an image containing a draft to be identified;
identifying a text sequence in the image; inputting the text sequence into a pre-trained recognition model, and acquiring each audit information and the type of each audit information contained in the text sequence output by the recognition model;
And outputting the type of each audit information.
In one possible implementation, the identifying the text sequence in the image includes:
determining each sub-text sequence contained in the image and the coordinates of the circumscribed rectangle of each sub-text sequence in the image through an OCR technology;
and according to the coordinates of the circumscribed rectangle of each sub-character sequence in the image, arranging characters in each sub-character sequence to generate a character sequence.
In one possible implementation manner, the determining, by OCR technology, coordinates of each sub-text sequence in the image and a rectangle circumscribing each sub-text sequence in the image includes:
determining each character in the image and the coordinates of the circumscribed rectangle of each character in the image by an OCR technology;
determining the distance between the circumscribed rectangles of each character, and determining the characters with the distance smaller than the preset distance as characters in the same sub-character sequence;
and determining the coordinates of the circumscribed rectangle of the sub-text sequence in the image according to the coordinates of the circumscribed rectangle of each text in the sub-text sequence aiming at each sub-text sequence.
In one possible implementation manner, the arranging the characters in each sub-character sequence according to the coordinates of the circumscribed rectangle of each sub-character sequence in the image to generate the character sequence includes:
Identifying the sub-text sequences in the same row according to the distance between the ordinate of the circumscribed rectangles of the sub-text sequences, and sequentially arranging the sub-text sequences in the same row according to the size of the abscissa of the circumscribed rectangles of the sub-text sequences in the same row to obtain the sub-text sequences in each row;
and splicing the sub-text sequences of each row in sequence according to the size sequence of the ordinate to obtain the text sequence.
In one possible implementation manner, the identifying the sub-text sequences located in the same row according to the distance between the ordinate of the circumscribed rectangle of the sub-text sequences includes:
and determining the sub-text sequences with the distance between the ordinate of the circumscribed rectangle smaller than the preset value as the sub-text sequences positioned in the same row.
In one possible embodiment, the preset value is determined by:
determining a first height of the sub-text sequence according to coordinates of the circumscribed rectangle of the sub-text sequence in the image;
determining a second height of the other sub-text sequences according to coordinates of the circumscribed rectangles of the other sub-text sequences in the image;
and determining a preset value according to the sum value of the first height and the second height.
In one possible embodiment, the identification model is a CRF model.
In one possible embodiment, the recognition model is trained by:
any sample sequence in a sample set is obtained, and each standard audit information and the standard type of each standard audit information contained in the sample sequence are obtained;
inputting the sample sequence into an original recognition model, and acquiring each output audit information and the output type of each output audit information contained in the sample sequence;
and training the original recognition model according to the standard audit information, the output audit information, the standard type and the output type.
Example 9:
the embodiment of the application also provides a computer program product which is executed by a computer to realize the draft information identification method of any method embodiment applied to electronic equipment.
In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof, and may be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions which, when loaded and executed on a computer, produce, in whole or in part, a process or function in accordance with embodiments of the present application.
According to the embodiment of the application, the electronic equipment identifies the text sequence in the image containing the draft to be identified through the identification model, and obtains each audit information and the type of each audit information in the draft, so that each audit information and the type of each audit information for audit can be rapidly and accurately screened, further, the audit of subsequent audit personnel is facilitated, and the audit efficiency is improved.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks. It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims (18)

1. A method of identifying draft information, the method comprising:
receiving an image containing a draft to be identified;
identifying a text sequence in the image; inputting the text sequence into a pre-trained recognition model, and acquiring each audit information and the type of each audit information contained in the text sequence output by the recognition model;
and outputting the type of each audit information.
2. The method of claim 1, wherein the identifying the sequence of words in the image comprises:
determining each sub-text sequence contained in the image and coordinates of an external rectangle of each sub-text sequence in the image through a text recognition OCR technology;
and according to the coordinates of the circumscribed rectangle of each sub-character sequence in the image, arranging characters in each sub-character sequence to generate a character sequence.
3. The method of claim 2, wherein determining the coordinates of each sub-text sequence in the image and the bounding rectangle of each sub-text sequence in the image by text recognition OCR comprises:
determining each character in the image and the coordinates of the circumscribed rectangle of each character in the image by an OCR technology;
Determining the distance between the circumscribed rectangles of each character, and determining the characters with the distance smaller than the preset distance as characters in the same sub-character sequence;
and determining the coordinates of the circumscribed rectangle of the sub-text sequence in the image according to the coordinates of the circumscribed rectangle of each text in the sub-text sequence aiming at each sub-text sequence.
4. The method of claim 1, wherein the arranging the text in each sub-text sequence according to the coordinates of the circumscribed rectangle of each sub-text sequence in the image to generate the text sequence comprises:
identifying the sub-text sequences in the same row according to the distance between the ordinate of the circumscribed rectangles of the sub-text sequences, and sequentially arranging the sub-text sequences in the same row according to the size of the abscissa of the circumscribed rectangles of the sub-text sequences in the same row to obtain the sub-text sequences in each row;
and splicing the sub-text sequences of each row in sequence according to the size sequence of the ordinate to obtain the text sequence.
5. The method of claim 4, wherein the identifying the sub-text sequences located in the same row based on the distance between the ordinate of the circumscribed rectangle of the sub-text sequences comprises:
And determining the sub-text sequences with the distance between the ordinate of the circumscribed rectangle smaller than the preset value as the sub-text sequences positioned in the same row.
6. The method of claim 5, wherein the predetermined value is determined by:
determining a first height of the sub-text sequence according to coordinates of the circumscribed rectangle of the sub-text sequence in the image;
determining a second height of the other sub-text sequences according to coordinates of the circumscribed rectangles of the other sub-text sequences in the image;
and determining a preset value according to the sum value of the first height and the second height.
7. The method of claim 1, wherein the identification model is a conditional random field CRF model.
8. The method of claim 1, wherein the recognition model is trained by:
any sample sequence in a sample set is obtained, and each standard audit information and the standard type of each standard audit information contained in the sample sequence are obtained;
inputting the sample sequence into an original recognition model, and acquiring each output audit information and the output type of each output audit information contained in the sample sequence;
And training the original recognition model according to the standard audit information, the output audit information, the standard type and the output type.
9. A draft information identification apparatus, the apparatus comprising:
the receiving module is used for receiving the image containing the draft to be identified;
the processing module is used for identifying the text sequence in the image; inputting the text sequence into a pre-trained recognition model, and acquiring each audit information and the type of each audit information contained in the text sequence output by the recognition model;
and the output module is used for outputting each audit information and the type of each audit information.
10. The apparatus according to claim 9, wherein the processing module is specifically configured to determine, by a character recognition OCR technique, coordinates of each sub-character sequence included in the image and a circumscribed rectangle of each sub-character sequence in the image; and according to the coordinates of the circumscribed rectangle of each sub-character sequence in the image, arranging characters in each sub-character sequence to generate a character sequence.
11. The apparatus according to claim 10, wherein the processing module is configured to determine, by OCR technology, coordinates of each text in the image and a circumscribed rectangle of each text in the image; determining the distance between the circumscribed rectangles of each character, and determining the characters with the distance smaller than the preset distance as characters in the same sub-character sequence; and determining the coordinates of the circumscribed rectangle of the sub-text sequence in the image according to the coordinates of the circumscribed rectangle of each text in the sub-text sequence aiming at each sub-text sequence.
12. The apparatus of claim 9, wherein the processing module is specifically configured to identify the sub-text sequences located in the same row according to a distance between the ordinate of the circumscribed rectangles of the sub-text sequences, and sequentially arrange the sub-text sequences located in the same row according to the size of the abscissa of the circumscribed rectangles of the sub-text sequences located in the same row, to obtain the sub-text sequences of each row; and splicing the sub-text sequences of each row in sequence according to the size sequence of the ordinate to obtain the text sequence.
13. The apparatus according to claim 12, wherein the processing module is specifically configured to determine the sub-word sequences with the distance between the ordinate axes of the circumscribed rectangle smaller than the preset value as the sub-word sequences located in the same row.
14. The apparatus of claim 13, wherein the processing module is further configured to determine the first height of the sequence of sub-words based on coordinates of a circumscribed rectangle of the sequence of sub-words in the image; determining a second height of the other sub-text sequences according to coordinates of the circumscribed rectangles of the other sub-text sequences in the image; and determining a preset value according to the sum value of the first height and the second height.
15. The apparatus of claim 9, wherein the processing module is further configured to obtain any sample sequence in a sample set, and each standard audit information and a standard type of each standard audit information included in the sample sequence; inputting the sample sequence into an original recognition model, and acquiring each output audit information and the output type of each output audit information contained in the sample sequence; and training the original recognition model according to the standard audit information, the output audit information, the standard type and the output type.
16. An electronic device comprising at least a processor and a memory, the processor being adapted to implement the steps of the draft information identification method according to any of the preceding claims 1-8 when executing a computer program stored in the memory.
17. A computer readable storage medium, characterized in that it stores a computer program which, when executed by a processor, implements the steps of the draft information identification method according to any of the preceding claims 1-8.
18. A computer program product, the computer program product comprising: computer program code for causing a computer to perform the steps of the draft information identification method according to any of the preceding claims 1-8 when said computer program code is run on a computer.
CN202310911695.8A 2023-07-24 2023-07-24 Draft information identification method, device, equipment and medium Pending CN116935398A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310911695.8A CN116935398A (en) 2023-07-24 2023-07-24 Draft information identification method, device, equipment and medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310911695.8A CN116935398A (en) 2023-07-24 2023-07-24 Draft information identification method, device, equipment and medium

Publications (1)

Publication Number Publication Date
CN116935398A true CN116935398A (en) 2023-10-24

Family

ID=88391890

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310911695.8A Pending CN116935398A (en) 2023-07-24 2023-07-24 Draft information identification method, device, equipment and medium

Country Status (1)

Country Link
CN (1) CN116935398A (en)

Similar Documents

Publication Publication Date Title
CN110163478B (en) Risk examination method and device for contract clauses
RU2695489C1 (en) Identification of fields on an image using artificial intelligence
CN107545422B (en) Cashing detection method and device
CN110648214B (en) Method and device for determining abnormal account
JPH07110841A (en) Method and apparatus for processing of securities
CN112395996A (en) Financial bill OCR recognition and image processing method, system and readable storage medium
US9436937B1 (en) Highlight-based bill processing
US11966890B2 (en) Bill identification method, device, electronic device and computer-readable storage medium
CN111914729A (en) Voucher association method and device, computer equipment and storage medium
Sabab et al. eExpense: a smart approach to track everyday expense
Fieberg et al. Machine learning in accounting research
CN113887199A (en) Financial text checking system based on knowledge base
Wang et al. A document analysis deep learning regression model for initial coin offerings success prediction
CN112365352A (en) Anti-cash-out method and device based on graph neural network
US20230040388A1 (en) Computerized method of training a computer executed model for recognizing numerical quantities
CN111428725A (en) Data structuring processing method and device and electronic equipment
CN116935398A (en) Draft information identification method, device, equipment and medium
US20230113578A1 (en) Transaction and ownership information document extraction
KR20150097837A (en) Pattern generating system for statement data generation and account recommendation and providing method thereof
CN114549177A (en) Insurance letter examination method, device, system and computer readable storage medium
CN113159924A (en) Method and device for determining trusted client object
CN113052679A (en) Model training method, prediction method and device based on multi-view learning and electronic equipment
KR20120124928A (en) Mathematical idenity based double entry jounalizing and accounting processing method
CN111797608A (en) Credit data checking method and device
Burke et al. Using a Large Language Model for Accounting Topic Classification

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination