CN110334596A - Invoice picture method of summary, electronic device and readable storage medium storing program for executing - Google Patents
Invoice picture method of summary, electronic device and readable storage medium storing program for executing Download PDFInfo
- Publication number
- CN110334596A CN110334596A CN201910462355.5A CN201910462355A CN110334596A CN 110334596 A CN110334596 A CN 110334596A CN 201910462355 A CN201910462355 A CN 201910462355A CN 110334596 A CN110334596 A CN 110334596A
- Authority
- CN
- China
- Prior art keywords
- invoice
- checked
- attribute
- picture
- information
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 230000000875 corresponding effect Effects 0.000 claims abstract description 96
- 238000013507 mapping Methods 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims description 46
- 230000002776 aggregation Effects 0.000 claims description 19
- 238000004220 aggregation Methods 0.000 claims description 19
- 230000008569 process Effects 0.000 claims description 10
- 238000013527 convolutional neural network Methods 0.000 claims description 7
- 238000003062 neural network model Methods 0.000 claims 1
- 238000005457 optimization Methods 0.000 abstract 1
- 238000012015 optical character recognition Methods 0.000 description 15
- 238000012545 processing Methods 0.000 description 14
- 238000010586 diagram Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000004308 accommodation Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 230000017105 transposition Effects 0.000 description 2
- 238000000151 deposition Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06V30/40—Document-oriented image-based pattern recognition
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Character Input (AREA)
Abstract
The present invention relates to process optimization techniques, a kind of invoice picture method of summary, electronic device and readable storage medium storing program for executing are provided, this method comprises: identifying the invoice type of each invoice picture to be summarized using model trained in advance；According to the mapping relations of predetermined invoice type and default invoice property location information, the default invoice property location information in each invoice picture is determined；OCR Text region is carried out to the default invoice attribute and corresponding property content that determine out position in each invoice picture, identifies the property content information in each invoice picture；The invoice attribute information to be checked of user's input is received, and invoice attribute information to be checked is matched with the property content information in each invoice picture；The invoice picture found out and matched, and show the invoice picture found out.The present invention realizes in multiple invoice pictures invoice needed for quickly positioning summarizes user out, improves work efficiency.
Description
Technical field
The present invention relates to field of computer technology more particularly to a kind of invoice picture methods of summary, electronic device and readable
Storage medium.
Background technique
Currently, user needs to find out a kind of invoice of particular community needed for meeting oneself in multiple invoice pictures to summarize
When checking, many pieces of page turning it can only check in multiple invoice pictures to search, it can not be quickly fixed in multiple invoice pictures
Position finds the invoice that oneself needs to pay close attention to, inefficiency.
Summary of the invention
The purpose of the present invention is to provide a kind of invoice picture method of summary, electronic device and readable storage medium storing program for executing, it is intended to
Invoice needed for quickly positioning summarizes user out in multiple invoice pictures.
To achieve the above object, the present invention provides a kind of electronic device, and the electronic device includes memory, processor,
The invoice picture aggregation system that can be run on the processor, the invoice picture aggregation system are stored on the memory
Following steps are realized when being executed by the processor:
After receiving multiple invoice pictures wait summarize, each hair to be summarized is identified using model trained in advance
The invoice type of ticket picture；
According to the mapping relations of predetermined invoice type and default invoice property location information, each hair is determined
Default invoice property location information in ticket picture；It include each default invoice attribute in the default invoice property location information
And the position of corresponding property content；
The knowledge of OCR text is carried out to the default invoice attribute and corresponding property content that determine out position in each invoice picture
Not, the corresponding property content information of each default invoice attribute in each invoice picture is identified；
The invoice attribute information to be checked of user's input is received, and the invoice attribute information to be checked is sent out with each
The corresponding property content information of each default invoice attribute is matched in ticket picture；
Invoice picture corresponding to the property content information to match with the invoice attribute information to be checked is found out, and is opened up
Show the invoice picture found out.
Preferably, in the invoice attribute information to be checked of the reception user input, and by the invoice attribute to be checked
Before the step of information property content information corresponding with default invoice attribute each in each invoice picture is matched, also
Include:
It shows preset information input interface to be checked, includes invoice category to be checked in the information input interface to be checked
Property options and invoice property content input item to be checked, for user inputted in the information input interface to be checked it is to be checked
Ask invoice attribute information；The invoice attribute information to be checked includes user in the to be checked of the information input interface to be checked
The invoice attribute to be checked selected in invoice Attributions selection item and user are in the to be checked of the information input interface to be checked
The invoice property content to be checked inputted in invoice property content input item.
Preferably, in the default invoice attribute to out position determining in each invoice picture and corresponding property content
OCR Text region is carried out, identifies the step of the corresponding property content information of each default invoice attribute in each invoice picture
After rapid, further includes:
One is established according to the corresponding property content information of default invoice attribute of each invoice picture identified to look into
Ask tables of data；Include the mapping relations between invoice picture, default invoice attribute and property content in the inquiry tables of data；
It is described receive user input invoice attribute information to be checked, and by the invoice attribute information to be checked with it is each
The step of corresponding property content information of each default invoice attribute is matched in invoice picture include:
The invoice attribute information to be checked for receiving user's input, according to be checked in the invoice attribute information to be checked
Invoice attribute and invoice property content to be checked are searched in the inquiry tables of data of foundation, are found out and the invoice to be checked
The invoice picture of invoice attribute to be checked and invoice property content to be checked mapping in attribute information.
Preferably, the model trained in advance is depth convolutional neural networks model, the model trained in advance
Training process is as follows:
A, the image pattern for being labeled with corresponding invoice type that invoice type prepares preset quantity is preset for each；
B, each is preset into the training subset and the second ratio that the corresponding image pattern of invoice type is divided into the first ratio
Verifying subset, the image pattern in each training subset is mixed to obtain training set, and will be in each verifying subset
Image pattern mixed be verified collection；
C, the training set training pattern is utilized；
D, using the recognition accuracy of the model of the verifying collection verifying training, if accuracy rate is more than or equal to pre-
If accuracy rate, then training terminates, alternatively, it is corresponding to increase each default invoice type if accuracy rate is less than default accuracy rate
Image pattern quantity, and re-execute the steps B, C, D.
In addition, to achieve the above object, the present invention also provides a kind of invoice picture method of summary, the invoice picture summarizes
Method includes:
After receiving multiple invoice pictures wait summarize, each hair to be summarized is identified using model trained in advance
The invoice type of ticket picture；
According to the mapping relations of predetermined invoice type and default invoice property location information, each hair is determined
Default invoice property location information in ticket picture；It include each default invoice attribute in the default invoice property location information
And the position of corresponding property content；
The knowledge of OCR text is carried out to the default invoice attribute and corresponding property content that determine out position in each invoice picture
Not, the corresponding property content information of each default invoice attribute in each invoice picture is identified；
The invoice attribute information to be checked of user's input is received, and the invoice attribute information to be checked is sent out with each
The corresponding property content information of each default invoice attribute is matched in ticket picture；
Invoice picture corresponding to the property content information to match with the invoice attribute information to be checked is found out, and is opened up
Show the invoice picture found out.
Preferably, in the invoice attribute information to be checked of the reception user input, and by the invoice attribute to be checked
Before the step of information property content information corresponding with default invoice attribute each in each invoice picture is matched, also
Include:
It shows preset information input interface to be checked, includes invoice category to be checked in the information input interface to be checked
Property options and invoice property content input item to be checked, for user inputted in the information input interface to be checked it is to be checked
Ask invoice attribute information；The invoice attribute information to be checked includes user in the to be checked of the information input interface to be checked
The invoice attribute to be checked selected in invoice Attributions selection item and user are in the to be checked of the information input interface to be checked
The invoice property content to be checked inputted in invoice property content input item.
Preferably, in the default invoice attribute to out position determining in each invoice picture and corresponding property content
OCR Text region is carried out, identifies the step of the corresponding property content information of each default invoice attribute in each invoice picture
After rapid, further includes:
One is established according to the corresponding property content information of default invoice attribute of each invoice picture identified to look into
Ask tables of data；Include the mapping relations between invoice picture, default invoice attribute and property content in the inquiry tables of data；
It is described receive user input invoice attribute information to be checked, and by the invoice attribute information to be checked with it is each
The step of corresponding property content information of each default invoice attribute is matched in invoice picture include:
The invoice attribute information to be checked for receiving user's input, according to be checked in the invoice attribute information to be checked
Invoice attribute and invoice property content to be checked are searched in the inquiry tables of data of foundation, are found out and the invoice to be checked
The invoice picture of invoice attribute to be checked and invoice property content to be checked mapping in attribute information.
Preferably, the model trained in advance is depth convolutional neural networks model, the model trained in advance
Training process is as follows:
A, the image pattern for being labeled with corresponding invoice type that invoice type prepares preset quantity is preset for each；
B, each is preset into the training subset and the second ratio that the corresponding image pattern of invoice type is divided into the first ratio
Verifying subset, the image pattern in each training subset is mixed to obtain training set, and will be in each verifying subset
Image pattern mixed be verified collection；
C, the training set training pattern is utilized；
D, using the recognition accuracy of the model of the verifying collection verifying training, if accuracy rate is more than or equal to pre-
If accuracy rate, then training terminates, alternatively, it is corresponding to increase each default invoice type if accuracy rate is less than default accuracy rate
Image pattern quantity, and re-execute the steps B, C, D.
Preferably, the default invoice attribute includes Business Name, company's industry, set of books.
Further, to achieve the above object, the present invention also provides a kind of computer readable storage medium, the computers
Readable storage medium storing program for executing is stored with invoice picture aggregation system, and the invoice picture aggregation system can be held by least one processor
Row, so that at least one described processor is executed such as the step of above-mentioned invoice picture method of summary.
Invoice picture method of summary, electronic device and readable storage medium storing program for executing proposed by the present invention pass through mould trained in advance
Type identifies the invoice type of each invoice picture to be summarized, and is determined in each invoice picture according to invoice type
Default invoice property location carries out default invoice attribute and corresponding property content that out position is determined in each invoice picture
OCR Text region；Receive user input invoice attribute information to be checked, and by the invoice attribute information to be checked with it is each
The corresponding property content information of each default invoice attribute is matched in invoice picture；It finds out and the invoice category to be checked
Property the invoice picture that matches of information, and show the invoice picture found out.Due to can in multiple invoice pictures to be summarized from
It is dynamic to be matched to invoice picture corresponding with the invoice attribute of inquiry needed for user, and user is showed, without user manually to every
One invoice picture page turning is searched, and is realized invoice needed for quickly positioning summarizes user out in multiple invoice pictures, is improved
Working efficiency.
Detailed description of the invention
Fig. 1 is the running environment schematic diagram of 10 preferred embodiment of invoice picture aggregation system of the present invention；
Fig. 2 is the flow diagram of one embodiment of invoice picture method of summary of the present invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that described herein, specific examples are only used to explain the present invention, not
For limiting the present invention.Based on the embodiments of the present invention, those of ordinary skill in the art are not before making creative work
Every other embodiment obtained is put, shall fall within the protection scope of the present invention.
It should be noted that the description for being related to " first ", " second " etc. in the present invention is used for description purposes only, and cannot
It is interpreted as its relative importance of indication or suggestion or implicitly indicates the quantity of indicated technical characteristic.Define as a result, " the
One ", the feature of " second " can explicitly or implicitly include at least one of the features.In addition, the skill between each embodiment
Art scheme can be combined with each other, but must be based on can be realized by those of ordinary skill in the art, when technical solution
Will be understood that the combination of this technical solution is not present in conjunction with there is conflicting or cannot achieve when, also not the present invention claims
Protection scope within.
The present invention provides a kind of invoice picture aggregation system.Referring to Fig. 1, be invoice picture aggregation system 10 of the present invention compared with
The running environment schematic diagram of good embodiment.
In the present embodiment, the invoice picture aggregation system 10 is installed and is run in electronic device 1.Electronics dress
Setting 1 may include, but be not limited only to, memory 11, processor 12 and display 13.Fig. 1 illustrates only the electricity with component 11-13
Sub-device 1, it should be understood that being not required for implementing all components shown, the implementation that can be substituted is more or less
Component.
The memory 11 is the readable computer storage medium of at least one type, and the memory 11 is in some implementations
It can be the internal storage unit of the electronic device 1, such as the hard disk or memory of the electronic device 1 in example.The memory
11 are also possible to the External memory equipment of the electronic device 1 in further embodiments, such as are equipped on the electronic device 1
Plug-in type hard disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card,
Flash card (Flash Card) etc..Further, the memory 11 can also both include the storage inside of the electronic device 1
Unit also includes External memory equipment.The memory 11 for store the application software for being installed on the electronic device 1 and respectively
Class data, such as the program code etc. of the invoice picture aggregation system 10.The memory 11 can be also used for temporarily depositing
Store up the data that has exported or will export.
The processor 12 can be in some embodiments a central processing unit (Central Processing Unit,
CPU), microprocessor or other data processing chips, for running the program code stored in the memory 11 or processing number
According to, such as execute the invoice picture aggregation system 10 etc..
The display 13 can be light-emitting diode display, liquid crystal display, touch-control liquid crystal display in some embodiments
And OLED (Organic Light-Emitting Diode, Organic Light Emitting Diode) touches device etc..The display 13 is used
In being shown in the information handled in the electronic device 1 and for showing visual user interface, such as each invoice
The invoice type of picture, the invoice picture matched etc..The component 11-13 of the electronic device 1 passes through the intercommunication of system bus phase
Letter.
Invoice picture aggregation system 10 includes that at least one is stored in the computer-readable instruction in the memory 11, should
At least one computer-readable instruction can be executed by the processor 12, to realize each embodiment of the application.
Wherein, following steps are realized when above-mentioned invoice picture aggregation system 10 is executed by the processor 12:
Step S1 is identified to be summarized after receiving multiple invoice pictures wait summarize using model trained in advance
The invoice type of each invoice picture；
Step S2 is determined according to the mapping relations of predetermined invoice type and default invoice property location information
Default invoice property location information in each invoice picture；Comprising each default in the default invoice property location information
The position of invoice attribute and corresponding property content；
Step S3 carries out default invoice attribute and corresponding property content that out position is determined in each invoice picture
OCR Text region identifies the corresponding property content information of each default invoice attribute in each invoice picture；
Step S4, receive user input invoice attribute information to be checked, and will the invoice attribute information to be checked and
The corresponding property content information of each default invoice attribute is matched in each invoice picture；
Step S5 finds out invoice figure corresponding to the property content information to match with the invoice attribute information to be checked
Piece, and show the invoice picture found out.
In the present embodiment, multiple invoice pictures to be summarized are received first.Such as receive user's sending includes to be summarized
The invoice summary request of multiple invoice pictures, for example, receive user (such as document typing personnel) by mobile phone, tablet computer, from
The invoice summary request for helping the terminals such as terminal device to send, such as reception user are whole in mobile phone, tablet computer, self-help terminal equipment
The invoice summary request sent in preassembled client in end, or user is received in mobile phone, tablet computer, self-aided terminal
The invoice summary request sent on browser in the terminals such as equipment.It, can after receiving multiple invoice pictures wait summarize
Treat multiple invoice pictures for summarizing carry out it is preset go hot-tempered processing, such as treat multiple the invoice pictures summarized and carry out Gaussian Blur
Processing, tentatively to remove the noise in multiple invoice pictures to be summarized, miscellaneous point interference.
It further,, can be to invoice picture if invoice Pictures location is not just after receiving multiple invoice pictures wait summarize
Carry out rotation processing.Specifically, can judge hair according to the position of seal in the depth-width ratio information and invoice picture of invoice picture
The transposition situation of ticket picture, and do overturning adjustment.For example, illustrating that invoice picture is high wide when the depth-width ratio of invoice picture is greater than 1
It is reverse, if seal position is on the left of invoice picture in invoice picture, rotated ninety degrees clockwise processing is done to invoice picture, if
Rotated ninety degrees counterclockwise then are done to invoice picture and are handled on the right side of invoice picture in seal position；When the depth-width ratio of invoice picture
When less than 1, illustrate that the high width of invoice picture does not overturn, if seal position is on the downside of invoice picture in invoice picture, to invoice figure
Piece rotates clockwise 180 degree of processing.
After receiving multiple invoice pictures wait summarize, to be summarized each is identified using preparatory trained model
The invoice type of invoice picture, such as food and drink invoice, traffic class invoice, accommodation invoice, outpatient service bill, bill of being hospitalized.Identification
Out after the invoice type of invoice picture, due to each attribute in all invoices of the same invoice type and property content is corresponded to
Position is all fixed and invariable, and therefore, can determine that in the invoice picture according to the invoice type of the invoice picture identified
The position of each invoice attribute and corresponding property content.Wherein, trained model is depth convolutional neural networks (example in advance
Such as, which can be to be chosen in the environment of CaffeNet based on depth convolutional neural networks SSD
(Single Shot MultiBox Detector) algorithm model, the training process of the model are as follows: being A, each default hair
Fare ticket type type (for example, default invoice type include outpatient service bill, bill of being hospitalized, insurance charge receipt, settle a claim out only according to etc.) it is quasi-
The image pattern for being labeled with corresponding invoice type of standby preset quantity (for example, 1000)；B, each is preset into invoice class
The corresponding image pattern of type is divided into the training subset of the first ratio (for example, 80%) and the verifying of the second ratio (for example, 20%)
Subset mixes the image pattern in each training subset to obtain training set, and by the image in each verifying subset
Sample is mixed to be verified collection；C, the training set training model is utilized；D, collect verifying instruction using the verifying
The recognition accuracy of the experienced model, if accuracy rate is more than or equal to default accuracy rate, training terminates, alternatively, if quasi-
True rate is less than default accuracy rate, then increases the quantity of each default corresponding image pattern of invoice type, and re-executes step
Rapid B, C, D.
Each default invoice attribute can be customized by the user, as user is customized need to often inquiring or more important
As default invoice attribute, all properties such as Business Name, company's industry, the taxpayer for being also possible to be defaulted as invoice know attribute
Alias, address, phone, bank of deposit and account etc..For example, if desired user opens up after invoice summarizes according to company's section or accounts
Show, then can preset default invoice attribute is company's section or accounts, then is identifying each invoice picture to be summarized
When the corresponding property content information of default invoice attribute, then " company's section " or " accounts " of each invoice picture is only identified
Property content information improves invoice and summarizes speed, user can quick locating query to other unrelated attributes then without identification
The invoice for needing to pay close attention to oneself.
It determines in invoice picture behind the position of each default invoice attribute and corresponding property content, it can be to the invoice figure
Determine that the corresponding property content of the default invoice attribute of position carries out OCR Text region in piece.For example, using predetermined
Character recognition model identifies the corresponding property content information of default invoice attribute that position is determined in the invoice picture.Its
In, which can be OCR optical character recognition engine, is also possible to be learnt in advance, train
Obtained character recognition model, such as time recurrent neural networks model (Long-Short Term Memory, LSTM) etc., herein
Without limitation.Specialized dictionary can also be pre-established, according to invoice common words (such as each Business Name that may relate to,
Number etc.) specialized dictionary is established, the default hair for identifying and determining position in the invoice picture is compared according to specialized dictionary
The corresponding property content information of ticket attribute, to save system resource.
Receive the invoice attribute information to be checked of user's input, it is possible to provide an information input interface to be checked, this is to be checked
Asking can be at this including invoice Attributions selection item to be checked and invoice property content input item to be checked, user in information input interface
It selects oneself to need to pay close attention to the invoice to be checked searched in invoice Attributions selection item to be checked in information input interface to be checked
Attribute, the invoice attribute to be checked are one in default invoice attribute.After user selects invoice attribute to be checked, it can be waited at this
It is inputted in invoice property content input item to be checked in query information input interface opposite with the invoice attribute to be checked of selection
The invoice property content to be checked answered, the invoice attribute to be checked as user selects is " Business Name ", then in invoice to be checked
Corresponding Business Name content (can be company name full name, or company name is referred to as) is inputted in property content input item,
Issuing inquiry instruction (such as clicking " inquiry " button in information input interface to be checked) can be from multiple invoice picture quickly
Find out the invoice picture to match with the invoice property content to be checked of user's input.
The present embodiment identifies the invoice type of each invoice picture to be summarized by model trained in advance, according to
Invoice type determines the default invoice property location in each invoice picture, to out position determining in each invoice picture
Default invoice attribute and corresponding property content carry out OCR Text region；The invoice attribute information to be checked of user's input is received,
And the invoice attribute information to be checked property content corresponding with default invoice attribute each in each invoice picture is believed
Breath is matched；The invoice picture to match with the invoice attribute information to be checked is found out, and shows the invoice picture found out.
Since invoice figure corresponding with the invoice attribute of inquiry needed for user can be automatically matched in multiple invoice pictures to be summarized
Piece, and user is showed, each invoice picture page turning is searched manually without user, is realized fast in multiple invoice pictures
Invoice needed for speed positioning summarizes user out, improves work efficiency.
In an optional embodiment, on the basis of the embodiment of above-mentioned Fig. 1,10 quilt of invoice picture aggregation system
When the processor 12 executes, following steps are also realized:
One is established according to the corresponding property content information of default invoice attribute of each invoice picture identified to look into
Ask tables of data；Include the mapping relations between invoice picture, default invoice attribute and property content in the inquiry tables of data.
In the present embodiment, according to the corresponding property content information of default invoice attribute of each invoice picture identified
Establish an inquiry tables of data；Comprising the corresponding each default invoice attribute of each invoice picture and right in the inquiry tables of data
The property content information answered.In this way, can be inputted according to user after the invoice attribute information to be checked for receiving user's input
Invoice attribute information to be checked is searched in the inquiry tables of data of foundation, finds out the invoice attribute to be checked with user's input
The invoice picture that information matches.The invoice picture presentation that will match to quickly is determined to realize according to user demand to user
Position invoice picture.
As shown in Fig. 2, Fig. 2 is the flow diagram of one embodiment of invoice picture method of summary of the present invention, the invoice picture
Method of summary the following steps are included:
Step S10 is identified to be summarized after receiving multiple invoice pictures wait summarize using model trained in advance
The invoice type of each invoice picture；
Step S20 is determined according to the mapping relations of predetermined invoice type and default invoice property location information
Default invoice property location information in each invoice picture；Comprising each default in the default invoice property location information
The position of invoice attribute and corresponding property content；
Step S30 carries out default invoice attribute and corresponding property content that out position is determined in each invoice picture
OCR Text region identifies the corresponding property content information of each default invoice attribute in each invoice picture；
Step S40, receive user input invoice attribute information to be checked, and will the invoice attribute information to be checked and
The corresponding property content information of each default invoice attribute is matched in each invoice picture；
Step S50 finds out invoice corresponding to the property content information to match with the invoice attribute information to be checked
Picture, and show the invoice picture found out.
In the present embodiment, multiple invoice pictures to be summarized are received first.Such as receive user's sending includes to be summarized
The invoice summary request of multiple invoice pictures, for example, receive user (such as document typing personnel) by mobile phone, tablet computer, from
The invoice summary request for helping the terminals such as terminal device to send, such as reception user are whole in mobile phone, tablet computer, self-help terminal equipment
The invoice summary request sent in preassembled client in end, or user is received in mobile phone, tablet computer, self-aided terminal
The invoice summary request sent on browser in the terminals such as equipment.It, can after receiving multiple invoice pictures wait summarize
Treat multiple invoice pictures for summarizing carry out it is preset go hot-tempered processing, such as treat multiple the invoice pictures summarized and carry out Gaussian Blur
Processing, tentatively to remove the noise in multiple invoice pictures to be summarized, miscellaneous point interference.
It further,, can be to invoice picture if invoice Pictures location is not just after receiving multiple invoice pictures wait summarize
Carry out rotation processing.Specifically, can judge hair according to the position of seal in the depth-width ratio information and invoice picture of invoice picture
The transposition situation of ticket picture, and do overturning adjustment.For example, illustrating that invoice picture is high wide when the depth-width ratio of invoice picture is greater than 1
It is reverse, if seal position is on the left of invoice picture in invoice picture, rotated ninety degrees clockwise processing is done to invoice picture, if
Rotated ninety degrees counterclockwise then are done to invoice picture and are handled on the right side of invoice picture in seal position；When the depth-width ratio of invoice picture
When less than 1, illustrate that the high width of invoice picture does not overturn, if seal position is on the downside of invoice picture in invoice picture, to invoice figure
Piece rotates clockwise 180 degree of processing.
After receiving multiple invoice pictures wait summarize, to be summarized each is identified using preparatory trained model
The invoice type of invoice picture, such as food and drink invoice, traffic class invoice, accommodation invoice, outpatient service bill, bill of being hospitalized.Identification
Out after the invoice type of invoice picture, due to each attribute in all invoices of the same invoice type and property content is corresponded to
Position is all fixed and invariable, and therefore, can determine that in the invoice picture according to the invoice type of the invoice picture identified
The position of each invoice attribute and corresponding property content.Wherein, trained model is depth convolutional neural networks (example in advance
Such as, which can be to be chosen in the environment of CaffeNet based on depth convolutional neural networks SSD
(Single Shot MultiBox Detector) algorithm model, the training process of the model are as follows: being A, each default hair
Fare ticket type type (for example, default invoice type include outpatient service bill, bill of being hospitalized, insurance charge receipt, settle a claim out only according to etc.) it is quasi-
The image pattern for being labeled with corresponding invoice type of standby preset quantity (for example, 1000)；B, each is preset into invoice class
The corresponding image pattern of type is divided into the training subset of the first ratio (for example, 80%) and the verifying of the second ratio (for example, 20%)
Subset mixes the image pattern in each training subset to obtain training set, and by the image in each verifying subset
Sample is mixed to be verified collection；C, the training set training model is utilized；D, collect verifying instruction using the verifying
The recognition accuracy of the experienced model, if accuracy rate is more than or equal to default accuracy rate, training terminates, alternatively, if quasi-
True rate is less than default accuracy rate, then increases the quantity of each default corresponding image pattern of invoice type, and re-executes step
Rapid B, C, D.
Each default invoice attribute can be customized by the user, as user is customized need to often inquiring or more important
As default invoice attribute, all properties such as Business Name, company's industry, the taxpayer for being also possible to be defaulted as invoice know attribute
Alias, address, phone, bank of deposit and account etc..For example, if desired user opens up after invoice summarizes according to company's section or accounts
Show, then can preset default invoice attribute is company's section or accounts, then is identifying each invoice picture to be summarized
When the corresponding property content information of default invoice attribute, then " company's section " or " accounts " of each invoice picture is only identified
Property content information improves invoice and summarizes speed, user can quick locating query to other unrelated attributes then without identification
The invoice for needing to pay close attention to oneself.
It determines in invoice picture behind the position of each default invoice attribute and corresponding property content, it can be to the invoice figure
Determine that the corresponding property content of the default invoice attribute of position carries out OCR Text region in piece.For example, using predetermined
Character recognition model identifies the corresponding property content information of default invoice attribute that position is determined in the invoice picture.Its
In, which can be OCR optical character recognition engine, is also possible to be learnt in advance, train
Obtained character recognition model, such as time recurrent neural networks model (Long-Short Term Memory, LSTM) etc., herein
Without limitation.Specialized dictionary can also be pre-established, according to invoice common words (such as each Business Name that may relate to,
Number etc.) specialized dictionary is established, the default hair for identifying and determining position in the invoice picture is compared according to specialized dictionary
The corresponding property content information of ticket attribute, to save system resource.
Receive the invoice attribute information to be checked of user's input, it is possible to provide an information input interface to be checked, this is to be checked
Asking can be at this including invoice Attributions selection item to be checked and invoice property content input item to be checked, user in information input interface
It selects oneself to need to pay close attention to the invoice to be checked searched in invoice Attributions selection item to be checked in information input interface to be checked
Attribute, the invoice attribute to be checked are one in default invoice attribute.After user selects invoice attribute to be checked, it can be waited at this
It is inputted in invoice property content input item to be checked in query information input interface opposite with the invoice attribute to be checked of selection
The invoice property content to be checked answered, the invoice attribute to be checked as user selects is " Business Name ", then in invoice to be checked
Corresponding Business Name content (can be company name full name, or company name is referred to as) is inputted in property content input item,
Issuing inquiry instruction (such as clicking " inquiry " button in information input interface to be checked) can be from multiple invoice picture quickly
Find out the invoice picture to match with the invoice property content to be checked of user's input.
The present embodiment identifies the invoice type of each invoice picture to be summarized by model trained in advance, according to
Invoice type determines the default invoice property location in each invoice picture, to out position determining in each invoice picture
Default invoice attribute and corresponding property content carry out OCR Text region；The invoice attribute information to be checked of user's input is received,
And the invoice attribute information to be checked property content corresponding with default invoice attribute each in each invoice picture is believed
Breath is matched；The invoice picture to match with the invoice attribute information to be checked is found out, and shows the invoice picture found out.
Since invoice figure corresponding with the invoice attribute of inquiry needed for user can be automatically matched in multiple invoice pictures to be summarized
Piece, and user is showed, each invoice picture page turning is searched manually without user, is realized fast in multiple invoice pictures
Invoice needed for speed positioning summarizes user out, improves work efficiency.
In an optional embodiment, on the basis of the above embodiments, this method further includes following steps:
One is established according to the corresponding property content information of default invoice attribute of each invoice picture identified to look into
Ask tables of data；Include the mapping relations between invoice picture, default invoice attribute and property content in the inquiry tables of data.
In the present embodiment, according to the corresponding property content information of default invoice attribute of each invoice picture identified
Establish an inquiry tables of data；Comprising the corresponding each default invoice attribute of each invoice picture and right in the inquiry tables of data
The property content information answered.In this way, can be inputted according to user after the invoice attribute information to be checked for receiving user's input
Invoice attribute information to be checked is searched in the inquiry tables of data of foundation, finds out the invoice attribute to be checked with user's input
The invoice picture that information matches.The invoice picture presentation that will match to quickly is determined to realize according to user demand to user
Position invoice picture.
In addition, the computer-readable recording medium storage has the present invention also provides a kind of computer readable storage medium
Invoice picture aggregation system, the invoice picture aggregation system can be executed by least one processor so that it is described at least one
Processor is executed such as the step of invoice picture method of summary in above-described embodiment, the step S10 of the invoice picture method of summary,
The specific implementation process such as S20, S30 are as described above, and details are not described herein.
It should be noted that, in this document, the terms "include", "comprise" or its any other variant are intended to non-row
His property includes, so that the process, method, article or the device that include a series of elements not only include those elements, and
And further include other elements that are not explicitly listed, or further include for this process, method, article or device institute it is intrinsic
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including being somebody's turn to do
There is also other identical elements in the process, method of element, article or device.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side
Method can be realized by means of software and necessary general hardware platform, naturally it is also possible to be realized by hardware, but very much
In the case of the former be more preferably embodiment.Based on this understanding, technical solution of the present invention is substantially in other words to existing
The part that technology contributes can be embodied in the form of software products, which is stored in a storage
In medium (such as ROM/RAM, magnetic disk, CD), including some instructions are used so that a terminal device (can be mobile phone, calculate
Machine, server, air conditioner or network equipment etc.) execute method described in each embodiment of the present invention.
Preferred embodiments of the present invention have been described above with reference to the accompanying drawings, not thereby limiting the scope of the invention.On
It is for illustration only to state serial number of the embodiment of the present invention, does not represent the advantages or disadvantages of the embodiments.It is patrolled in addition, though showing in flow charts
Sequence is collected, but in some cases, it can be with the steps shown or described are performed in an order that is different from the one herein.
Without departing from the scope and spirit of the invention, there are many variations to implement the present invention by those skilled in the art,
It can be used for another embodiment for example as the feature of one embodiment and obtain another embodiment.It is all to use technology of the invention
Made any modifications, equivalent replacements, and improvements within design, should all be within interest field of the invention.
Claims (10)
1. a kind of electronic device, which is characterized in that the electronic device includes memory, processor, is stored on the memory
There is the invoice picture aggregation system that can be run on the processor, the invoice picture aggregation system is executed by the processor
Shi Shixian following steps:
After receiving multiple invoice pictures wait summarize, each invoice figure to be summarized is identified using model trained in advance
The invoice type of piece；
According to the mapping relations of predetermined invoice type and default invoice property location information, each invoice figure is determined
Default invoice property location information in piece；Comprising each default invoice attribute and right in the default invoice property location information
Answer the position of property content；
OCR Text region is carried out to the default invoice attribute and corresponding property content that determine out position in each invoice picture,
Identify the corresponding property content information of each default invoice attribute in each invoice picture；
The invoice attribute information to be checked of user's input is received, and by the invoice attribute information to be checked and each invoice figure
The corresponding property content information of each default invoice attribute is matched in piece；
Invoice picture corresponding to the property content information to match with the invoice attribute information to be checked is found out, and shows and looks for
Invoice picture out.
2. electronic device as described in claim 1, which is characterized in that in the invoice attribute to be checked of the reception user input
Information, and will be in the invoice attribute information to be checked attribute corresponding with default invoice attribute each in each invoice picture
Before the step of appearance information is matched, further includes:
It shows preset information input interface to be checked, includes that invoice attribute to be checked selects in the information input interface to be checked
Item and invoice property content input item to be checked are selected, so that user inputs hair to be checked in the information input interface to be checked
Ticket attribute information；The invoice attribute information to be checked includes to be checked invoice of the user in the information input interface to be checked
The invoice to be checked of the invoice attribute to be checked that is selected in Attributions selection item and user in the information input interface to be checked
The invoice property content to be checked inputted in property content input item.
3. electronic device as claimed in claim 2, which is characterized in that described to out position determining in each invoice picture
Default invoice attribute and corresponding property content carry out OCR Text region, identify each default hair in each invoice picture
After the step of ticket attribute corresponding property content information, further includes:
An inquiry number is established according to the corresponding property content information of default invoice attribute of each invoice picture identified
According to table；Include the mapping relations between invoice picture, default invoice attribute and property content in the inquiry tables of data；
The invoice attribute information to be checked for receiving user's input, and the invoice attribute information to be checked is sent out with each
The step of corresponding property content information of each default invoice attribute is matched in ticket picture include:
The invoice attribute information to be checked for receiving user's input, according to the invoice to be checked in the invoice attribute information to be checked
Attribute and invoice property content to be checked are searched in the inquiry tables of data of foundation, are found out and the invoice attribute to be checked
The invoice picture of invoice attribute to be checked and invoice property content to be checked mapping in information.
4. electronic device as claimed in claim 1,2 or 3, which is characterized in that the model trained in advance is depth convolution
The training process of neural network model, the model trained in advance is as follows:
A, the image pattern for being labeled with corresponding invoice type that invoice type prepares preset quantity is preset for each；
B, each is preset into the corresponding image pattern of invoice type and is divided into the training subset of the first ratio and testing for the second ratio
Subset is demonstrate,proved, the image pattern in each training subset is mixed to obtain training set, and by the figure in each verifying subset
Decent is mixed to be verified collection；
C, the training set training pattern is utilized；
D, using the recognition accuracy of the model of the verifying collection verifying training, if accuracy rate is more than or equal to default standard
True rate, then training terminates, alternatively, increasing each default corresponding figure of invoice type if accuracy rate is less than default accuracy rate
Decent quantity, and it re-execute the steps B, C, D.
5. a kind of invoice picture method of summary, which is characterized in that the invoice picture method of summary includes:
After receiving multiple invoice pictures wait summarize, each invoice figure to be summarized is identified using model trained in advance
The invoice type of piece；
According to the mapping relations of predetermined invoice type and default invoice property location information, each invoice figure is determined
Default invoice property location information in piece；Comprising each default invoice attribute and right in the default invoice property location information
Answer the position of property content；
OCR Text region is carried out to the default invoice attribute and corresponding property content that determine out position in each invoice picture,
Identify the corresponding property content information of each default invoice attribute in each invoice picture；
The invoice attribute information to be checked of user's input is received, and by the invoice attribute information to be checked and each invoice figure
The corresponding property content information of each default invoice attribute is matched in piece；
Invoice picture corresponding to the property content information to match with the invoice attribute information to be checked is found out, and shows and looks for
Invoice picture out.
6. invoice picture method of summary as claimed in claim 5, which is characterized in that in the to be checked of the reception user input
Invoice attribute information, and the invoice attribute information to be checked is corresponding with default invoice attribute each in each invoice picture
Property content information the step of being matched before, further includes:
It shows preset information input interface to be checked, includes that invoice attribute to be checked selects in the information input interface to be checked
Item and invoice property content input item to be checked are selected, so that user inputs hair to be checked in the information input interface to be checked
Ticket attribute information；The invoice attribute information to be checked includes to be checked invoice of the user in the information input interface to be checked
The invoice to be checked of the invoice attribute to be checked that is selected in Attributions selection item and user in the information input interface to be checked
The invoice property content to be checked inputted in property content input item.
7. invoice picture method of summary as claimed in claim 6, which is characterized in that described to true in each invoice picture
The default invoice attribute and corresponding property content for making position carry out OCR Text region, identify each in each invoice picture
After the step of a default invoice attribute corresponding property content information, further includes:
An inquiry number is established according to the corresponding property content information of default invoice attribute of each invoice picture identified
According to table；Include the mapping relations between invoice picture, default invoice attribute and property content in the inquiry tables of data；
The invoice attribute information to be checked for receiving user's input, and the invoice attribute information to be checked is sent out with each
The step of corresponding property content information of each default invoice attribute is matched in ticket picture include:
The invoice attribute information to be checked for receiving user's input, according to the invoice to be checked in the invoice attribute information to be checked
Attribute and invoice property content to be checked are searched in the inquiry tables of data of foundation, are found out and the invoice attribute to be checked
The invoice picture of invoice attribute to be checked and invoice property content to be checked mapping in information.
8. the invoice picture method of summary as described in claim 5,6 or 7, which is characterized in that the model trained in advance is
The training process of depth convolutional neural networks model, the model trained in advance is as follows:
A, the image pattern for being labeled with corresponding invoice type that invoice type prepares preset quantity is preset for each；
B, each is preset into the corresponding image pattern of invoice type and is divided into the training subset of the first ratio and testing for the second ratio
Subset is demonstrate,proved, the image pattern in each training subset is mixed to obtain training set, and by the figure in each verifying subset
Decent is mixed to be verified collection；
C, the training set training pattern is utilized；
D, using the recognition accuracy of the model of the verifying collection verifying training, if accuracy rate is more than or equal to default standard
True rate, then training terminates, alternatively, increasing each default corresponding figure of invoice type if accuracy rate is less than default accuracy rate
Decent quantity, and it re-execute the steps B, C, D.
9. the invoice picture method of summary as described in claim 5,6 or 7, which is characterized in that the default invoice attribute includes
Business Name, company's industry, set of books.
10. a kind of computer readable storage medium, which is characterized in that be stored with invoice figure on the computer readable storage medium
Piece aggregation system is realized as described in any one of claim 5 to 9 when the invoice picture aggregation system is executed by processor
The step of invoice picture method of summary.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910462355.5A CN110334596B (en) | 2019-05-30 | 2019-05-30 | Invoice picture summarizing method, electronic device and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910462355.5A CN110334596B (en) | 2019-05-30 | 2019-05-30 | Invoice picture summarizing method, electronic device and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110334596A true CN110334596A (en) | 2019-10-15 |
| CN110334596B CN110334596B (en) | 2024-02-02 |
Family
ID=68140562
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910462355.5A Active CN110334596B (en) | 2019-05-30 | 2019-05-30 | Invoice picture summarizing method, electronic device and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110334596B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112434689A (en) * | 2020-12-01 | 2021-03-02 | 天冕信息技术(深圳)有限公司 | Method, device and equipment for identifying information in picture and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107766809A (en) * | 2017-10-09 | 2018-03-06 | 平安科技（深圳）有限公司 | Electronic installation, billing information recognition methods and computer-readable recording medium |
| 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 |
| CN109308476A (en) * | 2018-09-06 | 2019-02-05 | 邬国锐 | Billing information processing method, system and computer readable storage medium |
| CN109359127A (en) * | 2018-09-07 | 2019-02-19 | 彩讯科技股份有限公司 | A kind of querying method of electronic invoice, device, equipment and storage medium |
| CN109815949A (en) * | 2018-12-20 | 2019-05-28 | 航天信息股份有限公司 | Invoice publicity method and system neural network based |
-
2019
- 2019-05-30 CN CN201910462355.5A patent/CN110334596B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107766809A (en) * | 2017-10-09 | 2018-03-06 | 平安科技（深圳）有限公司 | Electronic installation, billing information recognition methods and computer-readable recording medium |
| 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 |
| CN109308476A (en) * | 2018-09-06 | 2019-02-05 | 邬国锐 | Billing information processing method, system and computer readable storage medium |
| CN109359127A (en) * | 2018-09-07 | 2019-02-19 | 彩讯科技股份有限公司 | A kind of querying method of electronic invoice, device, equipment and storage medium |
| CN109815949A (en) * | 2018-12-20 | 2019-05-28 | 航天信息股份有限公司 | Invoice publicity method and system neural network based |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112434689A (en) * | 2020-12-01 | 2021-03-02 | 天冕信息技术(深圳)有限公司 | Method, device and equipment for identifying information in picture and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110334596B (en) | 2024-02-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10528626B2 (en) | Document processing | |
| US20080244442A1 (en) | Techniques to share information between application programs | |
| WO2019024496A1 (en) | Enterprise recommendation method and application server | |
| CN103902535B (en) | Obtain the method, apparatus and system of associational word | |
| WO2019062193A1 (en) | Information display method and device | |
| US10929461B2 (en) | Automatic detection and transfer of relevant image data to content collections | |
| WO2017180072A1 (en) | Content based search and retrieval of trademark images | |
| CN109299334B (en) | Data processing method and device of knowledge graph | |
| CN108279987A (en) | The method for edition management and device of application program | |
| CN110598107A (en) | Management method of query system and computer storage medium | |
| CN105335515A (en) | Information recommendation method and information recommendation device | |
| WO2019222083A1 (en) | Action indicators for search operation output elements | |
| CN110084658A (en) | The matched method and apparatus of article | |
| CN107992523A (en) | The function choosing-item lookup method and terminal device of mobile application | |
| WO2021150632A1 (en) | Systems, methods, and interfaces for transaction aggregation, management, and visualization | |
| CN104426838A (en) | Internet cache scheduling method and system | |
| CN106528570A (en) | Recommendation method and device | |
| US20130346405A1 (en) | Systems and methods for managing data items using structured tags | |
| US20080170792A1 (en) | Apparatus and Method for Identifying Marker | |
| CN110334596A (en) | Invoice picture method of summary, electronic device and readable storage medium storing program for executing | |
| CN107748772A (en) | A kind of brand recognition method and device | |
| CN113343109A (en) | List recommendation method, computing device and computer storage medium | |
| CN109948040A (en) | Storage, recommended method and the system of object information, equipment and storage medium | |
| CN103473290B (en) | The processing method and processing device of the attribute data of point of interest | |
| CN110134867A (en) | Corporation information query method and Related product |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |