CN109858420A - A kind of bill processing system and processing method - Google Patents
A kind of bill processing system and processing method Download PDFInfo
- Publication number
- CN109858420A CN109858420A CN201910067174.2A CN201910067174A CN109858420A CN 109858420 A CN109858420 A CN 109858420A CN 201910067174 A CN201910067174 A CN 201910067174A CN 109858420 A CN109858420 A CN 109858420A
- Authority
- CN
- China
- Prior art keywords
- bill
- information
- text
- module
- examination
- 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
Links
Landscapes
- Character Input (AREA)
Abstract
The present invention provides a kind of bill processing system and processing method.This method comprises: extracting crucial text information from the image information of bill using the identification model based on deep neural network；Validity based on the crucial text information examination bill；Comparison based on the crucial text information and electronics bottom bill obtains the full dose information of required bill.Bill processing method of the invention can extract the key message of bill based on deep neural network model, effectively verify bill and obtain the full dose information of bill, bill processing system of the invention further can improve bank slip recognition efficiency using cloud platform.
Description
Technical field
The present invention relates to Financial Billing informationization technology field more particularly to a kind of bill processing system and processing methods.
Background technique
Bill is many kinds of, and in addition to the explicitly provided note form of country, there are many from pallet, different fonts for various regions
Financial Billing is also very common in real financial system, and such as electronic invoice, train ticket, boarding card, all kinds of machines dismiss ticket etc..?
In existing financial transaction, in order to which literal message in kind is converted to the character stream in information system, generally require to expend a large amount of
Manpower and material resources.Furthermore it is ensured that billing information accuracy also consumes a large amount of energy, generally required if there is small mistake
More manpower and material resources are put into be checked.
Currently, the usual way of bank slip recognition is to handle various bill in a manner of making template, however, with bill kind
Class, template increase, and production template will cause extra workload, and often exist between true bill and standard form
Gap, to cause the mistake for positioning and identifying.
Therefore, it is necessary to be improved to the prior art, to provide more efficient, universality stronger Financial Billing identification system
System and recognition methods.
Summary of the invention
Therefore, it is an object of the invention to overcome the defect of the above-mentioned prior art, a kind of Financial Billing processing system is provided
And processing method.
According to the first aspect of the invention, a kind of bill processing method is provided.This method comprises:
Step 1: extracting crucial text information from the image information of bill using the identification model based on deep neural network；
Step 2: the validity based on the crucial text information examination bill；
Step 3: the comparison based on the crucial text information and electronics bottom bill obtains the full dose information of required bill.
In one embodiment, step 1 includes following sub-step:
Step 12: using it is described based on the identification model of deep neural network to the character area of the image information of bill into
The choosing of row cutting frame, obtains text filed；
Step 13: to the text filed carry out Text region, obtaining content of text；
Step 14: the crucial text information of bill is extracted according to the content of text.
In one embodiment, step 12 includes following sub-step:
The image information for carrying out character area cutting will be needed to be input to the identification model based on deep neural network,
Export one group of rectangle anchor, wherein this group of rectangle anchor has different positions, size, direction and deflection angle；
For this group of rectangle anchor, different text boxes is combined into based on position, the close degree of angle and range information group, into
And it obtains described text filed.
According to the second aspect of the invention, a kind of bill processing system is provided.The system includes:
First bank slip recognition subsystem: for utilizing image information of the identification model based on deep neural network from bill
Extract crucial text information；
First examination module: for the validity based on the crucial text information examination bill and it is based on the key
The comparison of text information and electronics bottom bill obtains the full dose information of required bill.
First model library, for storing the identification model, so that the first bank slip recognition subsystem extracts the pass
Key text information
In one embodiment, the first bank slip recognition subsystem, the first examination module and first model
Warehouse compartment is in client, and the system also includes server end, the server end includes the second bank slip recognition subsystem and second
Model library, in which:
The second bank slip recognition subsystem is used to extract crucial text from the image information of bill using the identification model
This information；
Second model library extracts institute for storing the identification model, to provide the second bank slip recognition subsystem
State crucial text information.
In one embodiment, the client further includes scheduler module, wherein the scheduler module is for receiving bill
Image information, and according to the selection of the occupation condition of the client of monitoring the image information of bill is sent to described
First bank slip recognition subsystem processes are sent to the second bank slip recognition subsystem processes.
In one embodiment, the server end further includes the second examination module, and the second examination module is used for root
According to the validity of the crucial Textual information examination bill and the comparison based on the crucial text information and electronics bottom bill
The full dose information of bill needed for obtaining.
In one embodiment, first model library and second model library are synchronized more using scheduled strategy
Newly, to guarantee that the identification model of first model library keeps consistent with the identification model that second model library stores.
In one embodiment, the client further includes memory module and interactive display module, and the memory module is used
In the recognition result of storage Financial Billing, the interactive display module is used to show the recognition result of Financial Billing to user；Institute
Stating server end further includes bill pool module, for storing the relevant information of Financial Billing.
Compared with the prior art, the advantages of the present invention are as follows: the bill processing method and processing system of offer can be realized
Bill mechanized classification, effectively veritification bill, carry out bill classification, data inputting, validation verification without financial staff by hand
Deng；Processing different type Financial Billing is improved based on the Financial Billing processing system that artificial intelligence, cloud computing technology construct
Validity, safety and universality, and at the same time improving the efficiency of processing Financial Billing, simplifying financial process.
Detailed description of the invention
The following drawings only makees schematical description and interpretation to the present invention, is not intended to limit the scope of the present invention, in which:
Fig. 1 shows the schematic diagram of bill processing system according to an embodiment of the invention；
Fig. 2 shows the process flows of bank slip recognition subsystem in bill processing system according to an embodiment of the invention
Figure；
Fig. 3 shows the flow chart of bill processing method according to an embodiment of the invention.
Specific embodiment
It is logical below in conjunction with attached drawing in order to keep the purpose of the present invention, technical solution, design method and advantage more clear
Crossing specific embodiment, the present invention is described in more detail.It should be appreciated that specific embodiment described herein is only used for explaining
The present invention is not intended to limit the present invention.
According to one embodiment of present invention, a kind of bill processing system (by taking Financial Billing as an example) is provided, referring to Fig. 1
Shown, which generally comprises 200 two parts of client 100 and cloud platform.
Client 100 is mountable on the computer of user, mobile phone or receipts ticket machine.Cloud platform 200 is to run beyond the clouds
Service group, for example, cloud platform 200 is made of the Cloud Server and networking component for being equipped with cloud platform server software,
The functions such as analysis processing and storage for realizing financial data.
Client 100 include scheduler module 101, bank slip recognition subsystem 102, model library 103, display interactive module 104,
Check module 105 and memory module 106.Cloud platform 200 includes cloud model library 201, cloud examination module 202, cloud bill
Recognition subsystem 203 and bill pool module 204.
In the embodiment in figure 1, bank slip recognition subsystem includes two kinds of implementations of client and cloud platform, that is, is divided into visitor
Family end standalone version and Distributed Services version, are labeled as bank slip recognition subsystem in client, are labeled as cloud bill in cloud platform
Recognition subsystem.In one embodiment, it can be selected to use client according to the occupancy situation of client and/or the resource of cloud platform
The bank slip recognition subsystem or cloud bank slip recognition subsystem at end.
Each functions of modules of client 100 and cloud platform 200 will be hereafter introduced respectively.
One, the function of each module of client
Scheduler module 101 is the core control and scheduler module of client, with bank slip recognition subsystem 102, examination mould
Block 105, display interactive module 104 and memory module 106 have communication connection.Scheduler module 101 is appointed for Financial Billing identification
Recognition result is supplied to user etc. by reception, distribution and the storage of control recognition result of business.
For example, scheduler module 101 can from receive ticket terminal receive bill image data, receive ticket terminal can be scanner,
Mobile phone mobile terminal (the bill picture for being stored in mobile phone photo album) or other kinds of receipts ticket terminal.User can be by receiving ticket terminal
Application program initiation carries out the task requests of examination identification (or structuring processing) to Financial Billing and submits corresponding image
Data (such as the bill picture of scanning).Received image data is sent to bank slip recognition subsystem 102 and carried out by scheduler module 101
Handle and receive the recognition result of bank slip recognition subsystem 102.
In one embodiment, scheduler module 101 is also used to the resource service condition of monitor client 100, in client
Inadequate resource or when occupying over scheduled alarm level, the cloud bill that bank slip recognition task is uploaded to cloud platform 200 is known
Small pin for the case system 203 carries out distributed treatment.
For example, scheduler module 101 can detect the CPU usage, memory usage or monitoring the past period of client
Interior client to the processing speed etc. of bill, when one or more in these monitor values meet scheduled threshold value (for example,
CPU usage, memory usage are greater than a certain threshold value, and processing speed is lower than a certain threshold value), then scheduler module 101 is by bill
Image data is sent to cloud bank slip recognition subsystem 203 and is handled, for example, client to be switched to the upload of picture information
Bank slip recognition task is transferred to cloud to handle by mode, to utilize high concurrent, the High Availabitity characteristic in cloud.In this way, energy
The resource service condition and user for enough taking into account client require, and the distributed treatment performance in cloud are made full use of, to improve user
Experience.
Bank slip recognition subsystem 102 and scheduler module 101 and model library 103 have communication connection.Bank slip recognition subsystem
102 for converting the image information of bill, being identified, to extract the crucial text information needed, and recognition result is returned
Back to scheduler module 101, the key message of extraction is sent to examination module 105 by scheduler module 101 and carries out validity examination
Deng.
Checking module 105 and model library 103, scheduler module 101 and cloud examination module 202 has communication connection.?
In one embodiment, examination module 105 is used to for the key message of bill being sent to cloud examination module 202, is checked by cloud
Module 202 verifies the validity of bill and pulls the billing information of full dose.Examination module 105 can obtain ticket from scheduler module 101
According to key message, cloud examination module 202 can extract full dose based on the data in the key message of bill and electronics bottom account library
Billing information, and by examination result return to scheduler module 101.
Validity examination for bill, can pass through the inspection of bill consecutive numbers, double sign inspection, sola bill amount of money upper limit inspection
And keyword inspection etc. determines whether it is effective bill.For example, keyword inspection is closed by being compared in key word library
It is the specified not allowable content information of Financial Billing internally controlling requirement in key character library.
Examination result includes effective bill and invalid bill, for example, this Financial Billing is if occurred without double sign
Effective bill, otherwise the bill of this is invalid bill, and for another example, if it exceeds individual amount of money upper limit, then this bill is invalid
Bill.It in practical applications, whether can be effective bill according to a criterion or the examination of multinomial criterion.
The billing information for pulling full dose refers to, after the key message for extracting bill, according to the electronics bottom having built up
Account library identifies ticket information entirely, extracts comprehensive information from the account library of electronics bottom by key message, to have ensured letter
The comprehensive of breath, accuracy and validity.By this method, can be obtained merely with key message it is required it is comprehensive,
Accurate billing information, and improve processing speed.
Model library 103 is used to provide identification model, and the cloud model library with cloud platform for bank slip recognition subsystem 102
201 synchronized updates, to guarantee that client is consistent with model used in cloud platform.Model library 103 and cloud model library 201 have
It is bi-directionally connected, synchronized update request can be triggered by client, can also be triggered by cloud platform.
For example, identification model is trained neural network model, using the identification model, the image letter of bill is inputted
Breath, can obtain the key message of bill.The detailed process for extracting bill key message will introduce below.
In one embodiment, the synchronization between model library 103 and cloud model library 201 uses periodic triggers and event
Property the mode that combines of triggering.For example, by update side, actively triggering is synchronized after model modification in cloud model library 201
It updates, while the settable update cycle is periodically updated, in the case of updating failure to avoid event, client and cloud
The inconsistent problem of the model of platform.When the identification model of client and cloud platform is inconsistent, it is preferable that with the mould of cloud platform
Subject to type.
Showing that interactive module 104 is used for is the user of client according to the bill examination result from scheduler module 101
Show progress, billing information and the validity of bill etc. of bill structuring.
The bill pool module 204 of memory module 106 and scheduler module 101 and cloud platform has communication connection, memory module
106 function include data be locally stored and it is synchronous etc. with cloud data.Data synchronization can both be triggered by client,
It can be triggered by cloud platform, equally may be configured as event triggering and the mode that periodic triggers combine.
Two, the function of each module of cloud platform
Bill pool module 204 is related to bill for storing bill images file, billing information, bill source of magnanimity etc.
Digital information etc., consulted for the later period and store function be provided.
Cloud model library 201 is for storing bill images and trained identification model etc. through precipitating in bill pool.Know
Other model can be updated based on the newest data set of acquisition, to improve the accuracy and speed of identification, the model of client 100
Library 103 can synchronize cloud model library 201.
Cloud bank slip recognition subsystem 203 and cloud examination module 202 and cloud model library 201 have communication connection, use
In the key message based on the identification model extraction Financial Billing in cloud model library 201, and key message is sent to cloud
Check module 202.Since identification process carries out beyond the clouds, distributed way can be used and carry out bank slip recognition, the identification to carrying
The execution of the high concurrent, High Availabitity of task, to solve the problems, such as that processed offline resource consumption is big, concurrency is low.
Cloud examination module 202 compares the key message of bill and electronics bottom account library data, to verify bill
Validity and the full dose data for pulling bill provide unique, effective and comprehensive guarantee of billing information.For example, electronics bottom account
Library is timely, complete, accurate invoice electronic information base, such as the full ticket information of invoice that taxpayer issues.
By description above it is found that the bank slip recognition subsystem 102 of client and cloud bank slip recognition subsystem 203 are used
In the key message (crucial word content) for extracting bill based on the identification model in model library, module 105 and cloud are checked
Examination module 202 checks the validity of bill for the key message based on extraction and pulls full dose information of bill etc..
In one embodiment, the bank slip recognition subsystem 102 of client 100 is cut including preprocessing module, character area
Sub-module, Text region module and post-processing module, the treatment process of each module are shown in Figure 2, comprising the following steps:
Step S210 presorts to bill according to bill images information.
Preprocessing module receives the bill images information of scheduler module 101, carries out document field detection and bill classification, will
Bill is presorted by different type.
For example, presorting as VAT invoice, aircraft stroke list, train ticket, ticket kind known to boarding card etc., other are customized
The bill of template is classified by other types.Specific category really classified by post-processing module according to ticket contents by rule.
Step S220 carries out the choosing of cutting frame to the character area in bill images.
Character area cutting module carries out frame choosing for the word segment in image, in one embodiment, carries out text
The process of region cutting includes:
Step S221 determines the rectangle anchor for positioning text block based on neural network model.
In one embodiment, rectangle is obtained using the combination of depth convolutional neural networks and two-way shot and long term memory models
Anchor.For example, the image for needing to do character area cutting, which is input to depth convolutional neural networks, extracts characteristics of image, then through double
The feature that space correlation is extracted to shot and long term Memory Neural Networks, by the calculating of the built-up pattern, exporting one group of the ratio of width to height is
The rectangle anchor of 1:4, this group of rectangle anchor have the information such as different location, size, direction and deflection angle, utilize this group of rectangle anchor
In the subsequent position that can be used for that text block is marked to occur.
Step S222 determines text box according to rectangle anchor
The combination that rectangle anchor is carried out according to the position of rectangle anchor, size, direction and deflection angle information, for example, passing through position
It sets and searches for size in close region with angle and be combined towards consistent rectangle anchor, combined rectangle anchor ultimately forms text
The rectangle anchor group for being closer (such as distance is less than a certain predetermined threshold) is combined into a text box by the selection range in region, will
Rectangle anchor group of the distance farther out (as distance is greater than a certain threshold value) is combined into different text boxes.It in this way, can be effectively
The text of different zones is divided, combined text box has the information such as position, size, direction and deflection angle, according to these letters
Breath, can come out some region of image interception, and further become horizontal text image according to angle adjustment, obtain text
Area image, to carry out the text identification of next step.
Step S230 exports content of text to the text filed carry out Text region through cutting.
In one embodiment, the technology of use is depth convolutional neural networks, two-way shot and long term memory models and dynamic
Text filed carry out Text region of the law of planning to acquisition.Input is text filed image, and output is content of text.
For example, text filed image is extracted characteristics of image by convolutional neural networks, then remember through two-way shot and long term
Neural network extracts the feature of space correlation and predicts the text probability distribution of each position, finally by beam search (boundling
Search) etc. dynamic programming methods export content of text.
Step S240 extracts the key message of bill according to content of text.
It, may be rear to locate comprising other redundancies except many key elements after above-mentioned identification content of text
It manages according to the content of text of identification in module, carries out the filtering and extraction of key element information.
In one embodiment, the technology of use is the attention mechanism (attention of deep neural network
Mechanism), the realization of attention mechanism is to be exported by retaining the encoder of deep neural network to the intermediate of list entries
As a result, then train a model come to these input carry out selectivity study and model output when by output sequence with
Be associated, i.e., the generating probability of each single item in output sequence depend on selected in list entries which.The party
The input of method is that step S230 obtains all text flows, by for example comprising the deep neural network of RNN, output is that rejecting is superfluous
The crucial text information of remaining information.For example, the crucial text information of output includes the title of bill, code, number, date, gold
The information such as volume.
In conclusion the bank slip recognition subsystem that the embodiment of the present invention provides can identify the key message of bill, have
Effect solves the problems, such as bank slip recognition, classification and the information extraction of the unconventional format under template way.
Fig. 3 shows the flow chart of Financial Billing processing method according to an embodiment of the invention.The following steps are included:
Step S310 receives the task requests and corresponding bill images data of billing information processing.
User is sent out by scanner, mobile phone mobile terminal or the other kinds of client-side program for receiving ticket terminal to client
It rises and the task requests of structuring processing is carried out to bill and submit corresponding image data, such as born by the scheduler module of client
Duty receives task requests and image data.
Step S320 determines the bank slip recognition subsystem for assigning the task to client according to the resource service condition of monitoring
Or the cloud bank slip recognition subsystem of cloud platform.
After the scheduler module of client receives task requests, determined according to the system resource condition of monitoring by task point
The bank slip recognition subsystem or cloud bank slip recognition subsystem of dispensing local, to take into account the speed and efficiency of bill processing.
Step S330, corresponding bank slip recognition subsystem extract the key message of bill using identification model.
The bank slip recognition subsystem of client or the cloud bill subsystem of cloud platform handle bill images, mention
The key message (embodiment for the bank slip recognition subsystem that detailed process is seen above) for evidence of getting tickets.The key message of bill is for example
Including code, number, time and amount of money etc..
In the embodiment handled by the bank slip recognition subsystem of client, the bill key message result of extraction is returned
Back to scheduler module.
In the embodiment by cloud bill subsystem processes, the bill key message of extraction can be transmitted directly to cloud and look into
It tests module or can return to the scheduler module to client also to carry out subsequent processing.
Step S340 checks the validity of bill and the billing information of full dose according to the key message of bill.
In one embodiment, when being handled by the bank slip recognition subsystem of client, scheduler module is by bill
Key message is sent to examination module, checks module according to the key message of bill and checks the effective of module polls bill to cloud
Property and pull the billing information of full dose, and the billing information for checking result and full dose is back to scheduler module.Pass through this side
Formula can make full use of the storage and distributed computing function of cloud platform, improve the speed of bill processing.
In yet another embodiment, it is handled when by cloud bank slip recognition subsystem, after the key message for obtaining bill,
Key message is directly sent to the validity of cloud examination module polls bill and pulls the billing information of full dose, then, by
The examination module that the billing information for checking result and full dose is returned to client by module is checked in cloud, and by the examination of client
Module is sent to scheduler module.In a further embodiment, it after examination module beyond the clouds obtains examination result, also can return to
To the bank slip recognition subsystem in cloud, and examination result is sent to the scheduling mould of client by the bank slip recognition subsystem in cloud
Block.
Step S350 shows examination result to user.
Scheduler module checks result storage to local memory module for what is received, meanwhile, it is sent to interaction display mould
Block is to show examination result and request confirmation to user.
It should be understood that in the above embodiments, client and cloud platform further include for carrying out the logical of information exchange
Module is interrogated, wired or wireless connection can be used, cloud platform, which can be, is integrated with this based on what existing cloud computing platform was developed
The cloud platform of invention module.In addition, the present invention using model can be based on a large amount of historical data carry out data mining and
The neural network model that machine learning training obtains, is also possible to the model obtained by other modeling patterns, as long as being based on
Input can obtain the output of needs.It will also be appreciated that do not shown in attached drawing have between each module it is all
Possible connection, for example, examination module and memory module, it is also possible to there is communication connection or bank slip recognition subsystem and look into
Communication connection may also be had by testing between module, the key message of extraction is transmitted directly to examination mould by bank slip recognition subsystem
Block is handled.
It should be noted that processing system of the invention, client can also store electronics bottom bill, be independently completed finance
The examination identification work of bill simultaneously shows examination as a result, alternatively, only by cloud platform storage electronics bottom bill and bill pool to user
Information, to make full use of the storage and computing function of cloud platform.In addition, the processing task that cloud platform can receive multiple client is asked
It asks.
In conclusion combining client the present invention provides the method and system for carrying out automatically structuring processing to bill
End and cloud platform are identified and are checked to bill data, more smart according to trained neural network model acquisition in ping procedure
It is true as a result, solving the heavy billing information typing work problem of financial staff；Solves current Financial Billing OCR system pair
The restricted problem of specific template；Solve the problems, such as that current Financial Billing data input is easy error, review examination complexity；It solves
High-precision identifying system calculates that consumption is high, and be delayed long problem in low configuration client.Method of the invention can be user
Wisdom property tax total solution is provided.
It should be noted that, although each step is described according to particular order above, it is not intended that must press
Each step is executed according to above-mentioned particular order, in fact, some in these steps can concurrently execute, or even is changed suitable
Sequence, as long as can be realized required function.
The present invention can be system, method and/or computer program product.Computer program product may include computer
Readable storage medium storing program for executing, containing for making processor realize the computer-readable program instructions of various aspects of the invention.
Computer readable storage medium can be to maintain and store the tangible device of the instruction used by instruction execution equipment.
Computer readable storage medium for example can include but is not limited to storage device electric, magnetic storage apparatus, light storage device, electromagnetism and deposit
Store up equipment, semiconductor memory apparatus or above-mentioned any appropriate combination.The more specific example of computer readable storage medium
Sub (non exhaustive list) include: portable computer diskette, hard disk, random access memory (RAM), read-only memory (ROM),
Erasable programmable read only memory (EPROM or flash memory), static random access memory (SRAM), Portable compressed disk are read-only
Memory (CD-ROM), memory stick, floppy disk, mechanical coding equipment, is for example stored thereon with instruction at digital versatile disc (DVD)
Punch card or groove internal projection structure and above-mentioned any appropriate combination.
Various embodiments of the present invention are described above, above description is exemplary, and non-exclusive, and
It is not limited to disclosed each embodiment.Without departing from the scope and spirit of illustrated each embodiment, for this skill
Many modifications and changes are obvious for the those of ordinary skill in art field.The selection of term used herein, purport
In principle, the practical application or to the technological improvement in market for best explaining each embodiment, or make the art its
Its those of ordinary skill can understand each embodiment disclosed herein.
Claims (10)
1. a kind of bill processing method, comprising the following steps:
Step 1: extracting crucial text information from the image information of bill using the identification model based on deep neural network；
Step 2: the validity based on the crucial text information examination bill；
Step 3: the comparison based on the crucial text information and electronics bottom bill obtains the full dose information of required bill.
2. according to the method described in claim 1, wherein, step 1 includes following sub-step:
Step 12: being cut using described based on character area of the identification model of deep neural network to the image information of bill
Divide frame choosing, obtains text filed；
Step 13: to the text filed carry out Text region, obtaining content of text；
Step 14: the crucial text information of bill is extracted according to the content of text.
3. system according to claim 2, wherein step 12 includes following sub-step:
The image information for carrying out character area cutting will be needed to be input to the identification model based on deep neural network, output
One group of rectangle anchor, wherein this group of rectangle anchor has different positions, size, direction and deflection angle；
For this group of rectangle anchor, different text boxes is combined into based on position, the close degree of angle and range information group, and then obtain
It obtains described text filed.
4. a kind of bill processing system, comprising:
First bank slip recognition subsystem: for being extracted using the identification model based on deep neural network from the image information of bill
Crucial text information；
First examination module: for the validity based on the crucial text information examination bill and based on the crucial text
The comparison of information and electronics bottom bill obtains the full dose information of required bill.
First model library, for storing the identification model, so that the first bank slip recognition subsystem extracts the crucial text
This information.
5. system according to claim 4, which is characterized in that the first bank slip recognition subsystem, the first examination mould
Block and first model library are located at client, and the system also includes server end, the server end includes the second bill
Recognition subsystem and the second model library, in which:
The second bank slip recognition subsystem is used to extract crucial text envelope from the image information of bill using the identification model
Breath；
Second model library extracts the pass for storing the identification model, to provide the second bank slip recognition subsystem
Key text information.
6. system according to claim 5, which is characterized in that the client further includes scheduler module, wherein the tune
Degree module is used to receive the image information of bill, and is selected according to the occupation condition of the client of monitoring by bill
Image information is sent to the first bank slip recognition subsystem processes or is sent to the second bank slip recognition subsystem processes.
7. system according to claim 5, which is characterized in that the server end further includes the second examination module, described
Second examination module is used for according to the validity of the crucial Textual information examination bill and based on the crucial text information
The full dose information of required bill is obtained with the comparison of electronics bottom bill.
8. system according to claim 5, which is characterized in that first model library and second model library are using pre-
Fixed strategy synchronizes update, to guarantee that the identification model of first model library is kept and second model library storage
Identification model is consistent.
9. system according to claim 5, which is characterized in that the client further includes memory module and interaction display mould
Block, the memory module are used to store the recognition result of Financial Billing, and the interactive display module is used to show finance to user
The recognition result of bill；The server end further includes bill pool module, for storing the relevant information of Financial Billing.
10. a kind of computer readable storage medium, is stored thereon with computer program, wherein when the program is executed by processor
The step of realizing any one of -3 the method according to claim 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910067174.2A CN109858420A (en) | 2019-01-24 | 2019-01-24 | A kind of bill processing system and processing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910067174.2A CN109858420A (en) | 2019-01-24 | 2019-01-24 | A kind of bill processing system and processing method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109858420A true CN109858420A (en) | 2019-06-07 |
Family
ID=66895882
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910067174.2A Pending CN109858420A (en) | 2019-01-24 | 2019-01-24 | A kind of bill processing system and processing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109858420A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110472524A (en) * | 2019-07-25 | 2019-11-19 | 广东工业大学 | Invoice information management method, system and readable medium based on deep learning |
| CN110717484A (en) * | 2019-10-11 | 2020-01-21 | 支付宝(杭州)信息技术有限公司 | Image processing method and system |
| CN110956166A (en) * | 2019-12-02 | 2020-04-03 | 中国银行股份有限公司 | Bill marking method and device |
| CN111666885A (en) * | 2020-06-08 | 2020-09-15 | 成都知识视觉科技有限公司 | Template construction and matching method for medical document structured knowledge extraction |
| CN111709339A (en) * | 2020-06-09 | 2020-09-25 | 北京百度网讯科技有限公司 | Bill image recognition method, device, equipment and storage medium |
| CN112364645A (en) * | 2020-10-29 | 2021-02-12 | 浪潮通用软件有限公司 | Method and equipment for automatically auditing ERP financial system business documents |
| CN113449698A (en) * | 2021-08-30 | 2021-09-28 | 湖南文盾信息技术有限公司 | Automatic paper document input method, system, device and storage medium |
| CN117611363A (en) * | 2023-10-25 | 2024-02-27 | 浙江爱信诺航天信息技术有限公司 | Online verification method and medium for certificates |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105809164A (en) * | 2016-03-11 | 2016-07-27 | 北京旷视科技有限公司 | Character identification method and device |
| US9754176B2 (en) * | 2015-09-07 | 2017-09-05 | Abbyy Production Llc | Method and system for data extraction from images of semi-structured documents |
| CN107368831A (en) * | 2017-07-19 | 2017-11-21 | 中国人民解放军国防科学技术大学 | English words and digit recognition method in a kind of natural scene image |
| CN206975668U (en) * | 2017-05-31 | 2018-02-06 | 北京空间飞行器总体设计部 | A kind of finance reimbursement unanimous vote is according to intelligent checks system |
| CN107766894A (en) * | 2017-11-03 | 2018-03-06 | 吉林大学 | Remote sensing images spatial term method based on notice mechanism and deep learning |
| CN107798299A (en) * | 2017-10-09 | 2018-03-13 | 平安科技（深圳）有限公司 | Billing information recognition methods, electronic installation and readable storage medium storing program for executing |
| CN108229303A (en) * | 2017-11-14 | 2018-06-29 | 北京市商汤科技开发有限公司 | Detection identification and the detection identification training method of network and device, equipment, medium |
| CN108446621A (en) * | 2018-03-14 | 2018-08-24 | 平安科技（深圳）有限公司 | Bank slip recognition method, server and computer readable storage medium |
| CN108664897A (en) * | 2018-04-18 | 2018-10-16 | 平安科技（深圳）有限公司 | Bank slip recognition method, apparatus and storage medium |
| CN108717543A (en) * | 2018-05-14 | 2018-10-30 | 北京市商汤科技开发有限公司 | A kind of invoice recognition methods and device, computer storage media |
| CN109255084A (en) * | 2018-08-28 | 2019-01-22 | 腾讯科技（深圳）有限公司 | Electronic bill querying method, device, storage medium and computer equipment |
-
2019
- 2019-01-24 CN CN201910067174.2A patent/CN109858420A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9754176B2 (en) * | 2015-09-07 | 2017-09-05 | Abbyy Production Llc | Method and system for data extraction from images of semi-structured documents |
| CN105809164A (en) * | 2016-03-11 | 2016-07-27 | 北京旷视科技有限公司 | Character identification method and device |
| CN206975668U (en) * | 2017-05-31 | 2018-02-06 | 北京空间飞行器总体设计部 | A kind of finance reimbursement unanimous vote is according to intelligent checks system |
| CN107368831A (en) * | 2017-07-19 | 2017-11-21 | 中国人民解放军国防科学技术大学 | English words and digit recognition method in a kind of natural scene image |
| CN107798299A (en) * | 2017-10-09 | 2018-03-13 | 平安科技（深圳）有限公司 | Billing information recognition methods, electronic installation and readable storage medium storing program for executing |
| CN107766894A (en) * | 2017-11-03 | 2018-03-06 | 吉林大学 | Remote sensing images spatial term method based on notice mechanism and deep learning |
| CN108229303A (en) * | 2017-11-14 | 2018-06-29 | 北京市商汤科技开发有限公司 | Detection identification and the detection identification training method of network and device, equipment, medium |
| CN108446621A (en) * | 2018-03-14 | 2018-08-24 | 平安科技（深圳）有限公司 | Bank slip recognition method, server and computer readable storage medium |
| CN108664897A (en) * | 2018-04-18 | 2018-10-16 | 平安科技（深圳）有限公司 | Bank slip recognition method, apparatus and storage medium |
| CN108717543A (en) * | 2018-05-14 | 2018-10-30 | 北京市商汤科技开发有限公司 | A kind of invoice recognition methods and device, computer storage media |
| CN109255084A (en) * | 2018-08-28 | 2019-01-22 | 腾讯科技（深圳）有限公司 | Electronic bill querying method, device, storage medium and computer equipment |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110472524A (en) * | 2019-07-25 | 2019-11-19 | 广东工业大学 | Invoice information management method, system and readable medium based on deep learning |
| CN110472524B (en) * | 2019-07-25 | 2022-09-13 | 广东工业大学 | Invoice information management method and system based on deep learning and readable medium |
| CN110717484A (en) * | 2019-10-11 | 2020-01-21 | 支付宝(杭州)信息技术有限公司 | Image processing method and system |
| TWI793418B (en) * | 2019-10-11 | 2023-02-21 | 大陸商支付寶（杭州）信息技術有限公司 | Image processing method and system |
| CN110956166A (en) * | 2019-12-02 | 2020-04-03 | 中国银行股份有限公司 | Bill marking method and device |
| CN111666885A (en) * | 2020-06-08 | 2020-09-15 | 成都知识视觉科技有限公司 | Template construction and matching method for medical document structured knowledge extraction |
| CN111709339A (en) * | 2020-06-09 | 2020-09-25 | 北京百度网讯科技有限公司 | Bill image recognition method, device, equipment and storage medium |
| CN111709339B (en) * | 2020-06-09 | 2023-09-19 | 北京百度网讯科技有限公司 | Bill image recognition method, device, equipment and storage medium |
| US11854246B2 (en) | 2020-06-09 | 2023-12-26 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Method, apparatus, device and storage medium for recognizing bill image |
| CN112364645A (en) * | 2020-10-29 | 2021-02-12 | 浪潮通用软件有限公司 | Method and equipment for automatically auditing ERP financial system business documents |
| CN113449698A (en) * | 2021-08-30 | 2021-09-28 | 湖南文盾信息技术有限公司 | Automatic paper document input method, system, device and storage medium |
| CN117611363A (en) * | 2023-10-25 | 2024-02-27 | 浙江爱信诺航天信息技术有限公司 | Online verification method and medium for certificates |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109858420A (en) | A kind of bill processing system and processing method | |
| CN109887153A (en) | A kind of property tax processing method and processing system | |
| CN110135427B (en) | Method, apparatus, device and medium for recognizing characters in image | |
| CN109840519A (en) | A kind of adaptive intelligent form recognition input device and its application method | |
| CN107067044A (en) | A kind of finance reimbursement unanimous vote is according to intelligent checks system | |
| CN102737242A (en) | Automatic bill recognition method and system applied to mobile terminal | |
| CN109858414A (en) | A kind of invoice piecemeal detection method | |
| CN110178139A (en) | Use the system and method for the character recognition of the full convolutional neural networks with attention mechanism | |
| CN110188714A (en) | A kind of method, system and storage medium for realizing financial management under chat scenario | |
| CN109858373A (en) | A kind of invoice identification verification method and system based on deep learning | |
| CN112418812A (en) | Distributed full-link automatic intelligent clearance system, method and storage medium | |
| CN111581193A (en) | Data processing method, device, computer system and storage medium | |
| CN116797195A (en) | Work order processing method, apparatus, computer device, and computer readable storage medium | |
| CN113706291A (en) | Fraud risk prediction method, device, equipment and storage medium | |
| CN112069893A (en) | Bill processing method and device, electronic equipment and storage medium | |
| CN113673528B (en) | Text processing method, text processing device, electronic equipment and readable storage medium | |
| CN111105549A (en) | Optical character recognition method, device and computer storage medium | |
| CN116343237A (en) | Bill identification method based on deep learning and knowledge graph | |
| CN111414889B (en) | Financial statement identification method and device based on character identification | |
| CN115809796B (en) | Project intelligent dispatching method and system based on user portrait | |
| CN116823422A (en) | Form data processing method and device | |
| CN112232036A (en) | Reimbursement bill generation method, electronic device and computer-readable storage medium | |
| CN109697224B (en) | Bill message processing method, device and storage medium | |
| CN111104853A (en) | Image information input method and device, electronic equipment and storage medium | |
| US20220172301A1 (en) | System and method for clustering an electronic document that includes transaction evidence |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190607 |