CN108171127A - A kind of invoice automatic identifying method based on deep learning - Google Patents
A kind of invoice automatic identifying method based on deep learning Download PDFInfo
- Publication number
- CN108171127A CN108171127A CN201711332643.6A CN201711332643A CN108171127A CN 108171127 A CN108171127 A CN 108171127A CN 201711332643 A CN201711332643 A CN 201711332643A CN 108171127 A CN108171127 A CN 108171127A
- Authority
- CN
- China
- Prior art keywords
- invoice
- character
- image
- deep learning
- alexnet
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/40—Document-oriented image-based pattern recognition
- G06V30/41—Analysis of document content
- G06V30/412—Layout analysis of documents structured with printed lines or input boxes, e.g. business forms or tables
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/24—Aligning, centring, orientation detection or correction of the image
- G06V10/243—Aligning, centring, orientation detection or correction of the image by compensating for image skew or non-uniform image deformations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
-
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of invoice automatic identifying method based on deep learning, and this method first passes through the image of acquisition invoice；Character segmentation is carried out to the invoice image of acquisition again；Character classification and grouping are carried out to the character cut again；Then to the character being grouped, AlexNet deep learning network trainings are imported, the AlexNet networks trained complete invoice and identify part.The method of the present invention completes quick obtaining and the preservation of invoice image, completes the positional dissection of invoice amount, identifier etc., improves the discrimination of character.
Description
Technical field
The present invention relates to image domains, more particularly, to a kind of invoice automatic identifying method based on deep learning.
Background technology
Invoice handles intricate operation and dull, heavy workload, and record data are more, and artificial record needs to consume a large amount of people
Power material resources., can be quick using numerical identification technology, accurately extract the information of invoice.During so as to greatly reduce invoice processing
Between, it uses manpower and material resources sparingly, and improve the accuracy of invoice information record.
At present, number identification mainly has, the technologies such as template matches, characteristic matching, neural network.Template matching method is that handle is treated
It surveys masterplate and seeks similarity one by one with ready standard masterplate, take the highest module of similarity.The algorithm to seek template and sample
It is consistent, it is no to know that treatment effect is undesirable, and it is long to expend the time.Characteristic matching method compares template matching algorithm, improves identification
Speed, and with more robustness, but discrimination is the selection of feature.Neural network algorithm is hot research in recent years, identification
Effect is preferable, and when requiring, trained sample is enough, and the training time is long.
Invention content
The present invention provides a kind of invoice automatic identifying method based on deep learning, and this method overcomes existing invoice record speed
Degree is slow, it is error-prone the defects of.
In order to reach above-mentioned technique effect, technical scheme is as follows：
A kind of invoice automatic identifying method based on deep learning, which is characterized in that include the following steps：
S1：Acquire the image of invoice；
S2：Character segmentation is carried out to the invoice image of acquisition；
S3：Character classification and grouping are carried out to the character cut；
S4：To the character being grouped, AlexNet deep learning network trainings are imported；
S5：The AlexNet networks trained complete invoice and identify part.
Further, the detailed process of the step S1 is：
Invoice is shot using high photographing instrument, the characteristics of clearly demarcated is compared using high photographing instrument black pedestal and white invoice,
It is extracted using minimized profile and completes invoice locations of contours, then using affine transformation, completed the extraction of invoice image, do not carried on the back
The invoice image of scape.
Further, the detailed process of the step S2 is：
It is dispatched first using the inclination of Hough line change detection invoice horizontal line, by rotation to picture into line tilt correction.
Secondly invoice number subregion is obtained using template matching method, Morphological scale-space and utilization then is done to the subregion of acquisition
Gray Projection method cuts character, finally by image of the obtained single character normalization for 256 × 256 sizes.
Further, the detailed process of the step S3 is：
To establish the file of kinds of characters, digital figure mixed and disorderly under cutting is manually grouped, and therefrom random
Training sample is selected, test sample is prepared for CNN network trainings.
Further, the detailed process of the step S4 is：
To the AlexNet deep learnings network training, specially：AlexNet shares eight layers, first five layer is convolution
Layer, latter three layers are full articulamentums, and there are 10 to export for the output of the last one full articulamentum, respectively 0~9；AlexNet models
Parameter, batch sizes be 256, iteration 1000 times；Learning rate are initialized as 0.001, using step algorithms, often
500 times iterative attenuation is primary, and Momentum values are that 0.9, weight decay are 0.0005, and every 500 iteration export one
snapshot。
Compared with prior art, the advantageous effect of technical solution of the present invention is：
The present invention first passes through the image of acquisition invoice；Character segmentation is carried out to the invoice image of acquisition again；Again to having cut
Character carry out character classification with grouping；Then to the character being grouped, AlexNet deep learning network trainings, training are imported
Complete AlexNet networks complete invoice and identify part.The method of the present invention completes quick obtaining and the preservation of invoice image, completes hair
The positional dissection of the ticket amount of money, identifier etc. improves the discrimination of character.
Description of the drawings
Fig. 1 is the flow chart of the method for the present invention；
Fig. 2-1 high photographing instruments obtain invoice image；
The invoice image that Fig. 2-2 shootings obtain；
Fig. 2-3 is the design sketch for the invoice that Hough transformation adjusts slightly-inclined；
Fig. 3 invoice image preprocessing flows；
Fig. 3-3 Hough transformations adjust the invoice of slightly-inclined；
Fig. 3-3 dollar sign templates；
Fig. 3-4 amount of money locating effects；
The classification of Fig. 4 characters and grouping (for character 0,5)；
Fig. 5 AlexNet network structures；
Fig. 6 recognition effect figures.
Specific embodiment
The attached figures are only used for illustrative purposes and cannot be understood as limitating the patent；
In order to more preferably illustrate the present embodiment, the certain components of attached drawing have omission, zoom in or out, and do not represent actual product
Size；
To those skilled in the art, it is to be appreciated that certain known features and its explanation, which may be omitted, in attached drawing
's.
The following further describes the technical solution of the present invention with reference to the accompanying drawings and examples
Embodiment 1
First, master-plan
The present invention overall design philosophy be：The acquisition of invoice image；Traditional Method Character segmentation is carried out to invoice image；It cuts
The character cut is classified and is grouped；By the character being grouped, importing AlexNet network structures are trained；Last profit
With the complete AlexNet networks of training, invoice identification is completed.System overall procedure is as shown in Figure 1.
2nd, the acquisition of invoice image
This experiment shoots invoice using high photographing instrument.It is clearly demarcated using high photographing instrument black pedestal and white invoice comparison
The characteristics of, it is extracted using minimized profile and completes invoice locations of contours.Again using affine transformation, the extraction of invoice image is completed, is obtained
Obtain the invoice image of no background.
It can be seen that, invoice is clearly demarcated with background contrast from Fig. 2-1, using the minimized profile abstraction function in opencv libraries into
Row positions, the purple frame at invoice edge in effect such as figure.Accurate positioning, and record the coordinate on four vertex of invoice.
Invoice putting under high photographing instrument is not level, but there are an angles.The affine transformation side used herein
Method first establishes the invoice matrix of a horizontal direction.It is pushed up again by 4 vertex and newly-established horizontal invoice space that have just positioned
Point calculates projective transformation matrix.Using the transformation matrix, complete to tilt conversion of the invoice image to horizontal invoice image.Effect
As shown in Fig. 2-2.
3rd, Traditional Method Character segmentation
Invoice image preprocessing flow as shown in figure 3-1, including image slant correction, numeric area positioning, Character segmentation.
Image slant correction：
From Fig. 2-3, it can be seen that, obtained invoice image still has slight inclination, needs that image is carried out to tilt school
Just.Mainly using Hough transformation algorithm.Self-adaption binaryzation, then the Sobel_img sides using Y-direction are first carried out to figure
Edge detection algorithm removes the straight line of vertical direction in figure, the horizontal line of retention level.Morphologic corrosion expansion is recycled, is dashed forward
Go out horizontal line.The horizontal line in figure can be obtained using Hough transform method, as shown in figure 3-2.It is tilted further according to the equation calculation of horizontal line
Angle.Finally invoice image is rotated, completes image slant correction.
Numeric area positions：
It is positioned using template matches and obtains invoice number subregion.For identifying the amount of money, because invoice size determines, respectively
Information position is relatively fixed, and first takes one piece of approximate region of the amount of money to be identified, and the regional location and size are determined by designer, can
Reduce the time of template matches positioning.There is the symbol of RMB " " in the front end of amount of money number, takes the template (figure of the symbol
Shown in 3-3) it is positioned.The subregion of amount of money number is obtained, effect is as shown in Figure 3-4.
Character segmentation：
After obtaining the character subregion of Fig. 3-5, Character segmentation is carried out to subregion.Mainly do Morphological scale-space and profit
Character is cut with Gray Projection method.Finally by image of the obtained single character normalization for 256 × 256 sizes.At morphology
It manages and is expanded for corrosion, it is possible to reduce the possibility that character is connected.Gray Projection method is recycled, first carries out the projection of Y-direction, further according to
Projection result searches character edge, and Character segmentation is carried out according to edge.The projection cutting of X-direction is similarly.The list that will be obtained recently
A character normalization is the image of 256 × 256 sizes.
4th, character is classified and is grouped
The file of kinds of characters is established, arranges the character picture cut.It is each to randomly choose again from every a kind of data
200 data are as training group, and 20 data are as test group.Training picture and test pictures are normalized to 256 × 256 points
Resolution, binding tab is 0,1,2 respectively ..., 9.Part training sample picture is as shown in Figure 4.
5th, training AlexNet networks
AlexNet networks are as shown in Figure 5.The demand identified according to invoice number is by the output nerve of AlexNet graders
Member is reduced to 10 from 1000 and is respectively used to represent number 0~9.The parameter of AlexNet models, batch sizes are 256, repeatedly
Generation 1000 times.Learning rate are initialized as 0.001, and using step algorithms, every 500 iterative attenuations are primary.Momentum
It is 0.0005 to be worth for 0.9, weight decay, and every 500 iteration export a snapshot.By the character of Fig. 4 well cuttings into
Row training, obtains sorter network.200 test samples are tested in experiment.Entire recognizer in VS2013 and
It programs and realizes in the environment of OpenCV2.4.10, CPU frequency 2.0MHz, memory 2GB, 64 bit manipulation systems.It is taken the photograph using CMOS
As head, shot under the environment of light source abundance.Experimental result is as shown in table 1.According to experimental result, the average identification of invoice number
Rate is up to 99%.
6th, invoice identification is completed
Using trained AlexNet networks, we are with regard to that can complete the identification of character.Effect is received as shown in fig. 6, identifying
Tax people's identifier, the amount of money, the amount of tax to be paid.The tax rate in figure is obtained by the amount of tax to be paid/amount of money, and the purpose done so is that the tax rate is one
Equable numerical value, if finding tax rate mistake, i.e., there is identification mistake in the amount of tax to be paid or the amount of money.Manually data can be corrected.
The same or similar label correspond to the same or similar components；
Position relationship is used for only for illustration described in attached drawing, it is impossible to be interpreted as the limitation to this patent；
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be pair
The restriction of embodiments of the present invention.For those of ordinary skill in the art, may be used also on the basis of the above description
To make other variations or changes in different ways.There is no necessity and possibility to exhaust all the enbodiments.It is all this
All any modification, equivalent and improvement made within the spirit and principle of invention etc., should be included in the claims in the present invention
Protection domain within.
Claims (5)
1. a kind of invoice automatic identifying method based on deep learning, which is characterized in that include the following steps：
S1：Acquire the image of invoice；
S2：Character segmentation is carried out to the invoice image of acquisition；
S3：Character classification and grouping are carried out to the character cut；
S4：To the character being grouped, AlexNet deep learning network trainings are imported；
S5：The AlexNet networks trained complete invoice and identify part.
2. the invoice automatic identifying method according to claim 1 based on deep learning, which is characterized in that the step S1
Detailed process be：
Invoice is shot using high photographing instrument, the characteristics of clearly demarcated is compared using high photographing instrument black pedestal and white invoice, uses
Invoice locations of contours is completed in minimized profile extraction, then using affine transformation, is completed the extraction of invoice image, obtained no background
Invoice image.
3. the invoice automatic identifying method according to claim 2 based on deep learning, which is characterized in that the step S2
Detailed process be：
It is dispatched first using the inclination of Hough line change detection invoice horizontal line, by rotation to picture into line tilt correction.Secondly
Invoice number subregion is obtained using template matching method, Morphological scale-space is then done to the subregion of acquisition and utilizes gray scale
Sciagraphy cuts character, finally by image of the obtained single character normalization for 256 × 256 sizes.
4. the invoice automatic identifying method according to claim 3 based on deep learning, which is characterized in that the step S3
Detailed process be：
To establish the file of kinds of characters, digital figure mixed and disorderly under cutting is manually grouped, and is therefrom selected at random
Training sample, test sample are prepared for CNN network trainings.
5. the invoice automatic identifying method according to claim 4 based on deep learning, which is characterized in that the step S4
Detailed process be：
To the AlexNet deep learnings network training, specially：AlexNet shares eight layers, first five layer is convolutional layer, after
Three layers are full articulamentums, and there are 10 to export for the output of the last one full articulamentum, respectively 0~9；The ginseng of AlexNet models
Number, batch sizes are 256, iteration 1000 times；Learning rate are initialized as 0.001, using step algorithms, every 500 times
Iterative attenuation is primary, and Momentum values are that 0.9, weight decay are 0.0005, and every 500 iteration export one
snapshot。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711332643.6A CN108171127A (en) | 2017-12-13 | 2017-12-13 | A kind of invoice automatic identifying method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711332643.6A CN108171127A (en) | 2017-12-13 | 2017-12-13 | A kind of invoice automatic identifying method based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108171127A true CN108171127A (en) | 2018-06-15 |
Family
ID=62525189
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711332643.6A Pending CN108171127A (en) | 2017-12-13 | 2017-12-13 | A kind of invoice automatic identifying method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108171127A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108830236A (en) * | 2018-06-21 | 2018-11-16 | 电子科技大学 | A kind of recognition methods again of the pedestrian based on depth characteristic |
| CN108921166A (en) * | 2018-06-22 | 2018-11-30 | 深源恒际科技有限公司 | Medical bill class text detection recognition method and system based on deep neural network |
| CN109002768A (en) * | 2018-06-22 | 2018-12-14 | 深源恒际科技有限公司 | Medical bill class text extraction method based on the identification of neural network text detection |
| CN109214382A (en) * | 2018-07-16 | 2019-01-15 | 顺丰科技有限公司 | A kind of billing information recognizer, equipment and storage medium based on CRNN |
| CN109241894A (en) * | 2018-08-28 | 2019-01-18 | 南京安链数据科技有限公司 | A kind of specific aim ticket contents identifying system and method based on form locating and deep learning |
| CN109657665A (en) * | 2018-10-31 | 2019-04-19 | 广东工业大学 | A kind of invoice batch automatic recognition system based on deep learning |
| CN109858480A (en) * | 2019-01-08 | 2019-06-07 | 北京全路通信信号研究设计院集团有限公司 | Digital instrument identification method |
| CN109948617A (en) * | 2019-03-29 | 2019-06-28 | 南京邮电大学 | A kind of invoice image position method |
| CN109977957A (en) * | 2019-03-04 | 2019-07-05 | 苏宁易购集团股份有限公司 | A kind of invoice recognition methods and system based on deep learning |
| CN110909733A (en) * | 2019-10-28 | 2020-03-24 | 世纪保众(北京)网络科技有限公司 | Template positioning method and device based on OCR picture recognition and computer equipment |
| CN111127673A (en) * | 2019-12-27 | 2020-05-08 | 航天信息股份有限公司 | Invoice self-service authentication method and system supporting high-speed scanner |
| CN111462388A (en) * | 2020-03-19 | 2020-07-28 | 广州市玄武无线科技股份有限公司 | Bill inspection method and device, terminal equipment and storage medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102737242A (en) * | 2012-06-12 | 2012-10-17 | 丰豪盈彩（北京）科技有限公司 | Automatic bill recognition method and system applied to mobile terminal |
| CN103745213A (en) * | 2014-02-28 | 2014-04-23 | 中国人民解放军63680部队 | Optical character recognition method based on LVQ neural network |
| CN104751194A (en) * | 2015-04-27 | 2015-07-01 | 陈包容 | Processing method and processing device for financial expense reimbursement |
| CN105678292A (en) * | 2015-12-30 | 2016-06-15 | 成都数联铭品科技有限公司 | Complex optical text sequence identification system based on convolution and recurrent neural network |
| CN105678300A (en) * | 2015-12-30 | 2016-06-15 | 成都数联铭品科技有限公司 | Complex image and text sequence identification method |
| CN105956626A (en) * | 2016-05-12 | 2016-09-21 | 成都新舟锐视科技有限公司 | Deep learning based vehicle license plate position insensitive vehicle license plate recognition method |
| US20160283787A1 (en) * | 2008-01-18 | 2016-09-29 | Mitek Systems, Inc. | Systems and methods for mobile image capture and content processing of driver's licenses |
| US9471833B1 (en) * | 2012-04-03 | 2016-10-18 | Intuit Inc. | Character recognition using images at different angles |
| CN107194400A (en) * | 2017-05-31 | 2017-09-22 | 北京天宇星空科技有限公司 | A kind of finance reimbursement unanimous vote is according to picture recognition processing method |
| CN107358232A (en) * | 2017-06-28 | 2017-11-17 | 中山大学新华学院 | Invoice recognition methods and identification and management system based on plug-in unit |
-
2017
- 2017-12-13 CN CN201711332643.6A patent/CN108171127A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160283787A1 (en) * | 2008-01-18 | 2016-09-29 | Mitek Systems, Inc. | Systems and methods for mobile image capture and content processing of driver's licenses |
| US9471833B1 (en) * | 2012-04-03 | 2016-10-18 | Intuit Inc. | Character recognition using images at different angles |
| CN102737242A (en) * | 2012-06-12 | 2012-10-17 | 丰豪盈彩（北京）科技有限公司 | Automatic bill recognition method and system applied to mobile terminal |
| CN103745213A (en) * | 2014-02-28 | 2014-04-23 | 中国人民解放军63680部队 | Optical character recognition method based on LVQ neural network |
| CN104751194A (en) * | 2015-04-27 | 2015-07-01 | 陈包容 | Processing method and processing device for financial expense reimbursement |
| CN105678292A (en) * | 2015-12-30 | 2016-06-15 | 成都数联铭品科技有限公司 | Complex optical text sequence identification system based on convolution and recurrent neural network |
| CN105678300A (en) * | 2015-12-30 | 2016-06-15 | 成都数联铭品科技有限公司 | Complex image and text sequence identification method |
| CN105956626A (en) * | 2016-05-12 | 2016-09-21 | 成都新舟锐视科技有限公司 | Deep learning based vehicle license plate position insensitive vehicle license plate recognition method |
| CN107194400A (en) * | 2017-05-31 | 2017-09-22 | 北京天宇星空科技有限公司 | A kind of finance reimbursement unanimous vote is according to picture recognition processing method |
| CN107358232A (en) * | 2017-06-28 | 2017-11-17 | 中山大学新华学院 | Invoice recognition methods and identification and management system based on plug-in unit |
Non-Patent Citations (3)
| Title |
|---|
| EBRAHIM FARAHBAKHSH 等: "Improving Persian Digit Recognition by Combining Data Augmentation and AlexNet", 《10TH IRANIAN CONFERENCE ON MACHINE VISION AND IMAGE PROCESSING》 * |
| 刘峰: "一种改进的自适应增值税发票字符识别方法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 杜刚: "银行票据识别系统的研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108830236A (en) * | 2018-06-21 | 2018-11-16 | 电子科技大学 | A kind of recognition methods again of the pedestrian based on depth characteristic |
| CN108921166A (en) * | 2018-06-22 | 2018-11-30 | 深源恒际科技有限公司 | Medical bill class text detection recognition method and system based on deep neural network |
| CN109002768A (en) * | 2018-06-22 | 2018-12-14 | 深源恒际科技有限公司 | Medical bill class text extraction method based on the identification of neural network text detection |
| CN109214382A (en) * | 2018-07-16 | 2019-01-15 | 顺丰科技有限公司 | A kind of billing information recognizer, equipment and storage medium based on CRNN |
| CN109241894B (en) * | 2018-08-28 | 2022-04-08 | 南京安链数据科技有限公司 | Bill content identification system and method based on form positioning and deep learning |
| CN109241894A (en) * | 2018-08-28 | 2019-01-18 | 南京安链数据科技有限公司 | A kind of specific aim ticket contents identifying system and method based on form locating and deep learning |
| CN109657665A (en) * | 2018-10-31 | 2019-04-19 | 广东工业大学 | A kind of invoice batch automatic recognition system based on deep learning |
| CN109657665B (en) * | 2018-10-31 | 2023-01-20 | 广东工业大学 | Invoice batch automatic identification system based on deep learning |
| CN109858480A (en) * | 2019-01-08 | 2019-06-07 | 北京全路通信信号研究设计院集团有限公司 | Digital instrument identification method |
| CN109977957A (en) * | 2019-03-04 | 2019-07-05 | 苏宁易购集团股份有限公司 | A kind of invoice recognition methods and system based on deep learning |
| CN109948617A (en) * | 2019-03-29 | 2019-06-28 | 南京邮电大学 | A kind of invoice image position method |
| CN110909733A (en) * | 2019-10-28 | 2020-03-24 | 世纪保众(北京)网络科技有限公司 | Template positioning method and device based on OCR picture recognition and computer equipment |
| CN111127673A (en) * | 2019-12-27 | 2020-05-08 | 航天信息股份有限公司 | Invoice self-service authentication method and system supporting high-speed scanner |
| CN111462388A (en) * | 2020-03-19 | 2020-07-28 | 广州市玄武无线科技股份有限公司 | Bill inspection method and device, terminal equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108171127A (en) | A kind of invoice automatic identifying method based on deep learning | |
| CN110569878B (en) | Photograph background similarity clustering method based on convolutional neural network and computer | |
| CN111325203B (en) | American license plate recognition method and system based on image correction | |
| US11210797B2 (en) | Systems, methods, and devices for image matching and object recognition in images using textures | |
| CN112686812B (en) | Bank card inclination correction detection method and device, readable storage medium and terminal | |
| CN104700099B (en) | The method and apparatus for recognizing traffic sign | |
| CN105512657B (en) | Character identifying method and equipment | |
| CN109241808A (en) | Two dimensional code auto-screening method in a kind of fashion images | |
| CN111783757A (en) | OCR technology-based identification card recognition method in complex scene | |
| CN102024144A (en) | Container number identification method | |
| CN109271991A (en) | A kind of detection method of license plate based on deep learning | |
| CN103984920B (en) | Three-dimensional face identification method based on sparse representation and multiple feature points | |
| CN108564092A (en) | Sunflower disease recognition method based on SIFT feature extraction algorithm | |
| CN110413816A (en) | Colored sketches picture search | |
| CN108681735A (en) | Optical character recognition method based on convolutional neural networks deep learning model | |
| CN106897681A (en) | A kind of remote sensing images comparative analysis method and system | |
| CN110390228A (en) | The recognition methods of traffic sign picture, device and storage medium neural network based | |
| CN108154132A (en) | Method, system and equipment for extracting characters of identity card and storage medium | |
| CN107784321A (en) | Numeral paints this method for quickly identifying, system and computer-readable recording medium | |
| CN111340022A (en) | Identity card information identification method and device, computer equipment and storage medium | |
| CN105335760A (en) | Image number character recognition method | |
| CN110348289A (en) | A kind of finger vein identification method based on binary map | |
| CN105069403B (en) | A kind of three-dimensional human ear identification based on block statistics feature and the classification of dictionary learning rarefaction representation | |
| CN109977899A (en) | A kind of training, reasoning and the method and system for increasing New raxa of article identification | |
| CN110458019B (en) | Water surface target detection method for eliminating reflection interference under scarce cognitive sample condition |
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 |
Application publication date: 20180615 |
|
| RJ01 | Rejection of invention patent application after publication |