CN110135346A - Identity card automatic identifying method and system based on deep learning - Google Patents
Identity card automatic identifying method and system based on deep learning Download PDFInfo
- Publication number
- CN110135346A CN110135346A CN201910406426.XA CN201910406426A CN110135346A CN 110135346 A CN110135346 A CN 110135346A CN 201910406426 A CN201910406426 A CN 201910406426A CN 110135346 A CN110135346 A CN 110135346A
- Authority
- CN
- China
- Prior art keywords
- textbox
- picture
- textboxs
- identity card
- convolutional neural
- 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
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/146—Aligning or centring of the image pick-up or image-field
- G06V30/1475—Inclination or skew detection or correction of characters or of image to be recognised
- G06V30/1478—Inclination or skew detection or correction of characters or of image to be recognised of characters or characters lines
-
- 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)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses a kind of identity card automatic identifying method and system based on deep learning, wherein, this method, it include: when getting input picture, it is detected using the feature that identity card picture of first full convolutional neural networks for input carries out face and national emblem, the rotation information of picture is obtained, and is corrected；Text detection is carried out to postrotational picture using second full convolutional neural networks, obtains multiple textboxs of true text region；Slant Rectify is carried out to textbox, in horizontal direction by the angle adjustment of multiple textboxs；The friendship in the textbox of all horizontal directions between adjacent textbox and ratio are calculated, so as to connect in multiple textboxs with a line；Text region is carried out to the textbox after connection using third full convolutional neural networks, with the character in the textbox after the connection of identification.Technical solution of the present invention can automatic identification identity demonstrate,prove information, and can be improved the accuracy rate of identity card identification, there is preferable tolerance.
Description
Technical field
The present invention relates to field of information processing more particularly to a kind of identity card automatic identifying method based on deep learning,
System, computer equipment and storage medium.
Background technique
Currently, most of existing ID Card Recognition System and then is extracted in identity card based on the analysis being laid out to picture
Character area, the method anti-interference ability is poor, can not preferably analyze text in picture for complicated picture background
Position.Existing character recognition method carries out Text region, Text region to the character area that detected based on artificial feature
The lower of accuracy rate can not be determined in picture with the presence or absence of ID card information, be unfavorable for the identification operation of identity card.
In view of this, it is necessary to which current identity card identification technology is further improved in proposition.
Summary of the invention
To solve an above-mentioned at least technical problem, the main object of the present invention is to provide a kind of identity based on deep learning
Demonstrate,prove automatic identifying method, system, computer equipment and storage medium.
To achieve the above object, first technical solution that the present invention uses are as follows: a kind of body based on deep learning is provided
Part card automatic identifying method, comprising:
When getting input picture, face is carried out using identity card picture of first full convolutional neural networks to input
It is detected with the feature of national emblem, obtains the rotation information of picture, and correct；
Text detection is carried out to postrotational picture using second full convolutional neural networks, obtains true text location
Multiple textboxs in domain；
Slant Rectify is carried out to textbox, in horizontal direction by the angle adjustment of multiple textboxs；
The friendship in the textbox of all horizontal directions between adjacent textbox and ratio are calculated, so that in the multiple of a line
Textbox connection；
Text region is carried out according to tilt angle to the textbox after connection using the full convolutional neural networks of third, to know
The character in textbox after other connection.
Wherein, described to carry out face and national emblem using identity card picture of first full convolutional neural networks for input
Feature detection, obtains the rotation information of picture, and correct, specifically includes:
Using first full convolutional neural networks for the face and national emblem of the four direction on the identity card picture of input
Characteristic indication, obtain position and the bearing data of characteristic indication；
Become a full member according to picture of the bearing data of characteristic indication to input；
Identity card area image is separated from entire identity card picture according to the position data of characteristic indication, and is sent into the
Two full convolutional neural networks.
Wherein, described that Slant Rectify is carried out to textbox, it specifically includes:
Affine transformation is carried out to the textbox of second full convolutional neural networks output.
Wherein, the friendship in the textbox for calculating all horizontal directions between adjacent textbox and ratio, so that in same
Multiple textboxs of a line connect, and specifically include:
All textboxs are calculated according to the registration in Y-axis；
If registration is greater than given threshold, lateral connection is carried out to two adjacent textboxs.
Wherein, it is described lateral connection carried out to adjacent two textboxs after, comprising:
The width that the textbox of the second width is selected from all textboxs is used as setting width,
According to the width of other textboxs of the width adjustment of setting textbox.
To achieve the above object, second technical solution that the present invention uses are as follows: a kind of body based on deep learning is provided
Part card automatic recognition system, comprising:
Feature detection module, for get input picture when, using first full convolutional neural networks for input
Identity card picture carry out the feature detection of face and national emblem, obtain the rotation information of picture, and correct
Text detection module, for carrying out text detection to postrotational picture using second full convolutional neural networks,
Obtain multiple textboxs of true text region；
The angle adjustment of multiple textboxs is in water for carrying out Slant Rectify to textbox by textbox rectification module
Square to；
Link block, friendship and ratio in the textbox for calculating all horizontal directions between adjacent textbox, so that place
It is connected in multiple textboxs of same a line；
Text region module, for carrying out text knowledge to the textbox after connection using the full convolutional neural networks of third
Not, with the character in the textbox after the connection of identification.
To achieve the above object, the third technical solution that the present invention uses are as follows: a kind of computer equipment is provided, including is deposited
Reservoir, processor and storage on a memory and the computer program that can run on a processor, described in the processor execution
The step of above method is realized when computer program.
To achieve the above object, the 4th technical solution that the present invention uses are as follows: a kind of computer-readable storage medium is provided
The step of matter is stored thereon with computer program, and the computer program realizes above-mentioned method when being executed by processor.
Technical solution of the present invention use get input picture when, first with first full convolutional neural networks for
The identity card picture of input carries out the feature detection of face and national emblem, obtains the rotation information of picture, and correct；Then the is utilized
Two full convolutional neural networks carry out text detection to postrotational picture, obtain multiple texts of true text region
Frame；And Slant Rectify is carried out to textbox, in horizontal direction by the angle adjustment of multiple textboxs；Then calculate all water
Square to textbox in friendship between adjacent textbox and ratio, so as to be connected in multiple textboxs with a line；Last benefit
Text region is carried out to the textbox after connection with third full convolutional neural networks, in the textbox after the connection of identification
Character through the above steps being capable of automatic automatic identification identity card information.In addition, due to this programme by textbox into
Row is corrected and hands over and compare processing, can be improved the tolerance of picture, guarantees the accuracy of identity card identification.
Detailed description of the invention
Fig. 1 is the method flow diagram of identity card automatic identifying method of the one embodiment of the invention based on deep learning；
Fig. 2 is the block diagram of identity card automatic recognition system of the one embodiment of the invention based on deep learning；
Fig. 3 is the internal structure chart of one embodiment of the invention computer equipment.
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiment is only a part of the embodiments of the present invention, instead of all the embodiments.Base
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts it is all its
His embodiment, shall fall within the protection scope of the present invention.
It is to be appreciated that the description of " first ", " second " involved in the present invention etc. is used for description purposes only, and should not be understood as
Its relative importance of indication or suggestion or the quantity for implicitly indicating indicated technical characteristic.Define as a result, " first ",
The feature of " second " can explicitly or implicitly include at least one of the features.In addition, the technical side between each embodiment
Case can be combined with each other, but must be based on can be realized by those of ordinary skill in the art, when the combination of technical solution
Conflicting or cannot achieve when occur will be understood that the combination of this technical solution is not present, also not the present invention claims guarantor
Within the scope of shield.
Fig. 1 is please referred to, Fig. 1 is the method stream of identity card automatic identifying method of the one embodiment of the invention based on deep learning
Cheng Tu.In embodiments of the present invention, it is somebody's turn to do the identity card automatic identifying method based on deep learning, comprising:
Step S10, when getting input picture, using first full convolutional neural networks to the identity card picture of input
The feature detection for carrying out face and national emblem, obtains the rotation information of picture, and correct；
Step S20, text detection is carried out to postrotational picture using second full convolutional neural networks, obtains true text
Multiple textboxs of word region；
Step S30, Slant Rectify is carried out to textbox, in horizontal direction by the angle adjustment of multiple textboxs；
Step S40, the friendship in the textbox of all horizontal directions between adjacent textbox and ratio are calculated, so that in same
Capable multiple textboxs connection；
Step S50, Text region is carried out to the textbox after connection using third full convolutional neural networks, with identification
The character in textbox after connection.
In the present embodiment, detection identification can be carried out to ID card information included in input picture using this method.
Specifically, carrying out face using identity card picture of first full convolutional neural networks to input when getting input picture
It is detected with the feature of national emblem, obtains the rotation information of picture, and correct；Then using second full convolutional neural networks to rotation
Picture afterwards carries out text detection, determines the character area in input picture.It is pre-defined on second full convolutional neural networks
There is a large amount of potential textbox.During text detection, constantly the position of textbox is adjusted, to obtain true text
Multiple textboxs of word region.The position of multiple textboxs will appear inclination, be unfavorable for the connection of subsequent textbox.It will
After the angle adjustment horizontally of textbox, the recognition accuracy of full convolutional neural networks can be greatly improved.Existing text
Usually there is textbox leak detection in word detection, detect not congruent problem, for this purpose, this programme uses the board-like feature of identity card and leads to
It crosses the character area that using the height friendship of the textbox of horizontal direction and Determination goes out to be attached, will be greater than given threshold
Textbox carries out lateral connection, and by after the certain pixel of connection result horizontal expansion, can batch identification, improve recognition efficiency.
Finally, Text region is carried out to the textbox after connection using the full convolutional neural networks of third, with the text after the connection of identification
Character in word frame.The full convolutional neural networks of above-mentioned third are polytypic neural network model made of training in advance.
Text can be identified by the polytypic neural network model, achieve the purpose that automatic identification identity demonstrate,proves information.It should
CTC (Connectionist temporal classification) computation model is used in polytypic neural network model
Loss, to improve the accuracy of Text region.
Technical solution of the present invention is used when getting input picture, first with first full convolutional neural networks to defeated
The identity card picture entered carries out the feature detection of face and national emblem, obtains the rotation information of picture, and correct；Then second is utilized
A full convolutional neural networks carry out text detection to postrotational picture, obtain multiple textboxs of true text region；
And Slant Rectify is carried out to textbox, in horizontal direction by the angle adjustment of multiple textboxs；Then calculate all levels
Friendship and ratio in the textbox in direction between adjacent textbox, so as to be connected in multiple textboxs with a line；Finally utilize
The full convolutional neural networks of third carry out Text region to the textbox after connection, with the word in the textbox after the connection of identification
Symbol through the above steps being capable of automatic automatic identification identity card information.In addition, since this programme is by carrying out textbox
It corrects and hands over and compare processing, can be improved the tolerance of picture, guarantee the accuracy of identity card identification.
In a specific embodiment, the identity card picture for utilizing first full convolutional neural networks for input
The feature detection for carrying out face and national emblem, obtains the rotation information of picture, and correct, specifically includes:
Using first full convolutional neural networks for the face and national emblem of the four direction on the identity card picture of input
Characteristic indication, obtain position and the bearing data of characteristic indication；
Become a full member according to picture of the bearing data of characteristic indication to input；
Identity card area image is separated from entire identity card picture according to the position data of characteristic indication, and is sent into the
Two full convolutional neural networks.
In the present embodiment, first full convolutional neural networks is constructed, for detecting the full volume of identity card face and national emblem
First step of the product neural network as whole system, he is compared to for other methods, there is detection speed, more accurate faster
Image rotation information output.Specifically, first full convolutional neural networks can detect identity for the identity card picture of input
The face of four direction (i.e. upper and lower, left and right) and/or the characteristic indication of national emblem on picture are demonstrate,proved, if having recognized Zuoren face (body
Part card front), then it represents that this identity card picture has rotated to the left 90 °, and national emblem (identity card reverse side) is similarly.Therefore this
One full convolutional network can detecte and/or the position and direction of national emblem characteristic indication.It, can be with according to the directional information of mark
The identity card picture of input is become a full member.According to the location information of mark, we can proportionally by identity card region from
It is stripped out in entire picture, keeps the picture for being sent into second full convolutional neural networks more accurate.By first full convolution
After neural network, the available image become a full member and only remain identity card region is sent into second full convolution and detects neural network.
Foregoing description is the processing step to the characteristic indication information for detecting response.In the flag information for not detecting response
When, the input of original image is maintained, and be sent into second full convolution and detect neural network.
It is described that Slant Rectify is carried out to textbox in a specific embodiment, it specifically includes:
Affine transformation is carried out according to tilt angle to the textbox of second full convolutional neural networks output.
The present embodiment, by the textbox to second full convolutional neural networks output and according to the tilt angle of detection
Affine transformation can will adjust horizontally with directive textbox.
Friendship in a specific embodiment, in the textbox for calculating all horizontal directions between adjacent textbox
And compare, so as to be connected in multiple textboxs with a line, specifically include:
All textboxs are calculated according to the registration in Y-axis；
If registration is greater than given threshold, lateral connection is carried out to two adjacent textboxs.
Further, it is described lateral connection carried out to adjacent two textboxs after, comprising:
The width that the textbox of the second width is selected from all textboxs is used as setting width,
According to the width of other textboxs of the width adjustment of setting textbox.
In the present embodiment, in order to improve recognition efficiency, multiple textboxs can be identified simultaneously in Text region, by upper
Processing is stated, each textbox can be adjusted to the same size, it being capable of batch progress identifying processing.
Referring to figure 2., Fig. 2 is the module side of identity card automatic recognition system of the one embodiment of the invention based on deep learning
Block diagram.In the embodiment of the present invention, it is somebody's turn to do the identity card automatic recognition system based on deep learning, comprising:
Feature detection module 10, for get input picture when, using first full convolutional neural networks for defeated
The identity card picture entered carries out the feature detection of face and national emblem, obtains the rotation information of picture, and correct；
Text detection module 20, for carrying out text inspection to postrotational picture using second full convolutional neural networks
It surveys, obtains multiple textboxs of true text region；
Textbox rectification module 30, for being in by the angle adjustment of multiple textboxs to textbox progress Slant Rectify
Horizontal direction；
Link block 40, friendship and ratio in the textbox for calculating all horizontal directions between adjacent textbox, so that
Multiple textboxs connection in same a line；
Text region module 50, for carrying out text knowledge to the textbox after connection using second full convolutional neural networks
Not, with the character in the textbox after the connection of identification.
In the present embodiment, when getting input picture, feature detection module 10 utilizes first full convolutional neural networks
The feature detection that face and national emblem are carried out for the identity card picture of input, obtains the rotation information of picture, and correct.Text inspection
Module 20 is surveyed, text detection is carried out to postrotational picture using second full convolutional neural networks, is determined in input picture
Character area.Pre-defining on second full convolutional neural networks has a large amount of potential textbox.During text detection,
Constantly the position of textbox is adjusted, to obtain multiple textboxs of true text region.The position of multiple textboxs
It sets and will appear inclination, be unfavorable for the connection of subsequent textbox.It, can be by the angle of textbox by textbox rectification module 30
Adjustment horizontally, greatly improves the recognition accuracy of full convolutional neural networks.Usually there is text in existing text detection
Frame leak detection detects not congruent problem, for this purpose, this programme is by link block 40, using identity card board-like feature and pass through
It is handed over using the height of the textbox of horizontal direction and the character area of Determination out is attached, will be greater than the text of given threshold
Word frame carries out lateral connection, and by after the certain pixel of connection result horizontal expansion, can batch identification, improve recognition efficiency.Most
Afterwards, by Text region module 50, Text region is carried out to the textbox after connection using third full convolutional neural networks, with
The character in textbox after the connection of identification.The full convolutional neural networks of above-mentioned third are more classification made of training in advance
Neural network model.Text can be identified by the polytypic neural network model, reach automatic identification identity
Demonstrate,prove the purpose of information.
In one embodiment, further include feature detection module 10, be specifically used for: utilizing first full convolutional neural networks pair
In the face of the four direction on the identity card picture of input and the characteristic indication of national emblem, the position and direction of characteristic indication are obtained
Data；
Become a full member according to picture of the bearing data of characteristic indication to input；
Identity card area image is separated from entire identity card picture according to the position data of characteristic indication, and is sent into the
Two full convolutional neural networks
Further, the textbox rectification module 30, is specifically used for: to the text of second full convolutional neural networks output
Word frame carries out affine transformation according to tilt angle.
Further, the link block 40, is specifically used for:
All textboxs are calculated according to the registration in Y-axis；
If registration is greater than given threshold, lateral connection is carried out to two adjacent textboxs.
Further, the link block 40, is also used to:
The width that the textbox of the second width is selected from all textboxs is used as setting width,
According to the width of other textboxs of the width adjustment of setting textbox.
Specifically, feature detection module 10, text detection module 20 and Text region module 50 are by server come real
It is existing.The server uses deep learning service arrangement scheme, specifically: using Tensorflow-Serving is trained spy
Detection, text detection, identification model offer infrastructure service frame are provided, and Tensorflow-Serving service is packaged into one
Docker mirror image；Packed Docker mirror image is placed in the server for needing to provide service, and loads corresponding feature
The Web application framework Flask of detection, text detection, identification model, the lightweight used develops api interface, and passes through high-performance
Gunicorn http server service is externally provided.Its advantage is as follows: 1. can be moved using Tensorflow-Serving
The heat of state updates model algorithm, does not need to stop service；2. using Tensorflow-Serving can to the short time simultaneously on
The picture of biography carries out batch packing processing, and returns to recognition result, improves service performance；3. service is packaged in a Docker
The container environment that in mirror image, can be convenient the transplanting of service, and be relatively isolated improves the compatibility of deployment, accomplishes light weight
Grade, the effect of rapid deployment；4. process can flexibly be arranged according to the performance of server using the method for service of Gunicon
(workers), the quantity of thread, the workers quantity of recommendation are current CPU several * 2+1.And it can customize log
The function of content, facilitates inquiry and export；5. all models all use image processor (GPU) operation to calculate, increase model
Parallel speed.Further, when feature detection module 10 carries out feature detection, Gunicorn is by external detection port
Acquired picture is sent into the Tensorflow-Serving wrapped up by Docker, and Tensorflow-Serving is according to port
With the corresponding relationship of model name, picture is sent into special based on carrying out in the trained feature detection model of deep neural network
Sign detection.When text detection module 20 carries out text detection, picture acquired in external detection port is sent by Gunicorn
In the Tensorflow-Serving wrapped up by Docker, Tensorflow-Serving is corresponding with model name according to port
Picture is sent into and is based on carrying out text detection in the trained text detection model of deep neural network by relationship.Text region mould
When block 50 carries out Text region, the character area image of identity card acquired in external identification port is sent into quilt by Gunicorn
In the Tensorflow-Serving of Docker package.Tensorflow-Serving is closed according to port is corresponding with model name
Identity card character area image is sent into based on progress text knowledge in the trained Text region model of deep neural network by system
Not.
Referring to figure 3., Fig. 3 is the internal structure chart of one embodiment of the invention computer equipment.In one embodiment, the meter
Calculating machine equipment includes processor, memory and the network interface connected by system bus.Wherein, the processing of the computer equipment
Device is for providing calculating and control ability.The memory of the computer equipment includes non-volatile memory medium, built-in storage.It should
Non-volatile memory medium is stored with operating system, computer program and database.The built-in storage is non-volatile memories Jie
The operation of operating system and computer program in matter provides environment.The network interface of the computer equipment is used for and external end
End passes through network connection communication.When the computer program is executed by processor with realize a kind of identity card based on deep learning from
Dynamic recognition methods.
It will be understood by those skilled in the art that structure shown in Fig. 3, only part relevant to application scheme is tied
The block diagram of structure does not constitute the restriction for the computer equipment being applied thereon to application scheme, specific computer equipment
It may include perhaps combining certain components or with different component layouts than more or fewer components as shown in the figure.
In one embodiment, a kind of computer equipment is provided, including memory, processor and storage are on a memory
And the computer program that can be run on a processor, processor are realized when executing computer program in above each embodiment of the method
The step of.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
The step in above each embodiment of the method is realized when machine program is executed by processor.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the computer program can be stored in a non-volatile computer
In read/write memory medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein,
To any reference of memory, storage, database or other media used in each embodiment provided herein,
Including non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
The above description is only a preferred embodiment of the present invention, is not intended to limit the scope of the invention, all at this
Under the inventive concept of invention, using equivalent structure transformation made by description of the invention and accompanying drawing content, or directly/use indirectly
It is included in other related technical areas in scope of patent protection of the invention.
Claims (8)
1. a kind of identity card automatic identifying method based on deep learning, which is characterized in that the identity based on deep learning
Demonstrate,prove automatic identifying method, comprising:
When getting input picture, face and state are carried out using identity card picture of first full convolutional neural networks to input
The feature of emblem detects, and obtains the rotation information of picture, and correct；
Text detection is carried out to postrotational picture using second full convolutional neural networks, obtains true text region
Multiple textboxs；
Slant Rectify is carried out to textbox, in horizontal direction by the angle adjustment of multiple textboxs；
The friendship in the textbox of all horizontal directions between adjacent textbox and ratio are calculated, so that in multiple texts with a line
Frame connection；
Text region is carried out to the textbox after connection using third full convolutional neural networks, to identify the textbox after connection
In character.
2. the identity card automatic identifying method based on deep learning as described in claim 1, which is characterized in that described to utilize
The feature that one full convolutional neural networks carries out face and national emblem for the identity card picture of input detects, and obtains the rotation of picture
Information, and correct, it specifically includes:
Using first full convolutional neural networks for the face of the four direction on the identity card picture of input and the spy of national emblem
Sign mark, obtains position and the bearing data of characteristic indication；
Become a full member according to picture of the bearing data of characteristic indication to input；
Identity card area image is separated from entire identity card picture according to the position data of characteristic indication, and is sent into second
Full convolutional neural networks.
3. the identity card automatic identifying method based on deep learning as described in claim 1, which is characterized in that described to text
Frame carries out Slant Rectify, specifically includes:
Affine transformation is carried out according to tilt angle to the textbox of second full convolutional neural networks output.
4. the identity card automatic identifying method based on deep learning as described in claim 1, which is characterized in that the calculating institute
There are the friendship in the textbox of horizontal direction between adjacent textbox and ratio, so as to connect in multiple textboxs with a line, tool
Body includes:
All textboxs are calculated according to the registration in Y-axis；
If registration is greater than given threshold, lateral connection is carried out to two adjacent textboxs.
5. the identity card automatic identifying method based on deep learning as claimed in claim 4, which is characterized in that described to adjacent
Two textboxs carry out lateral connection after, comprising:
The width that the textbox of the second width is selected from all textboxs is used as setting width,
According to the width of other textboxs of the width adjustment of setting textbox.
6. a kind of identity card automatic recognition system based on deep learning, which is characterized in that the identity based on deep learning
Demonstrate,prove automatic recognition system, comprising:
Feature detection module, for get input picture when, using first full convolutional neural networks for the body of input
Part card picture carries out the feature detection of face and national emblem, obtains the rotation information of picture, and correct
Text detection module is obtained for carrying out text detection to postrotational picture using second full convolutional neural networks
Multiple textboxs of true text region；
The angle adjustment of multiple textboxs is in level side for carrying out Slant Rectify to textbox by textbox rectification module
To；
Link block, friendship and ratio in the textbox for calculating all horizontal directions between adjacent textbox, so that in same
Multiple textboxs of a line connect；
Text region module, for carrying out Text region to the textbox after connection using the full convolutional neural networks of third, with
The character in textbox after the connection of identification.
7. a kind of computer equipment including memory, processor and stores the meter that can be run on a memory and on a processor
Calculation machine program, which is characterized in that the processor realizes any one of claims 1 to 5 institute when executing the computer program
The step of stating method.
8. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program quilt
The step of processor realizes method described in any one of claims 1 to 5 when executing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910406426.XA CN110135346A (en) | 2019-05-16 | 2019-05-16 | Identity card automatic identifying method and system based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910406426.XA CN110135346A (en) | 2019-05-16 | 2019-05-16 | Identity card automatic identifying method and system based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110135346A true CN110135346A (en) | 2019-08-16 |
Family
ID=67574468
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910406426.XA Pending CN110135346A (en) | 2019-05-16 | 2019-05-16 | Identity card automatic identifying method and system based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110135346A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110689658A (en) * | 2019-10-08 | 2020-01-14 | 北京邮电大学 | Taxi bill identification method and system based on deep learning |
| CN111046883A (en) * | 2019-12-05 | 2020-04-21 | 吉林大学 | Intelligent evaluation method and system based on ancient coin image |
| CN111950554A (en) * | 2020-08-17 | 2020-11-17 | 深圳市丰巢网络技术有限公司 | Identification card identification method, device, equipment and storage medium |
| CN112183250A (en) * | 2020-09-14 | 2021-01-05 | 北京三快在线科技有限公司 | Character recognition method and device, storage medium and electronic equipment |
| CN112418158A (en) * | 2020-02-11 | 2021-02-26 | 支付宝实验室(新加坡)有限公司 | System suitable for detecting identity card and device and processing method associated with same |
| CN112766255A (en) * | 2021-01-19 | 2021-05-07 | 上海微盟企业发展有限公司 | Optical character recognition method, device, equipment and storage medium |
| CN112949523A (en) * | 2021-03-11 | 2021-06-11 | 兴业银行股份有限公司 | Method and system for extracting key information from identity card image picture |
| CN113392827A (en) * | 2021-06-22 | 2021-09-14 | 平安健康保险股份有限公司 | Character recognition method, device, equipment and medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7313251B2 (en) * | 1993-11-18 | 2007-12-25 | Digimarc Corporation | Method and system for managing and controlling electronic media |
| CN104680161A (en) * | 2015-01-09 | 2015-06-03 | 安徽清新互联信息科技有限公司 | Digit recognition method for identification cards |
| CN106682629A (en) * | 2016-12-30 | 2017-05-17 | 佳都新太科技股份有限公司 | Identification number identification algorithm in complicated background |
| CN107346420A (en) * | 2017-06-19 | 2017-11-14 | 中国科学院信息工程研究所 | Text detection localization method under a kind of natural scene based on deep learning |
| CN107665354A (en) * | 2017-09-19 | 2018-02-06 | 北京小米移动软件有限公司 | Identify the method and device of identity card |
-
2019
- 2019-05-16 CN CN201910406426.XA patent/CN110135346A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7313251B2 (en) * | 1993-11-18 | 2007-12-25 | Digimarc Corporation | Method and system for managing and controlling electronic media |
| CN104680161A (en) * | 2015-01-09 | 2015-06-03 | 安徽清新互联信息科技有限公司 | Digit recognition method for identification cards |
| CN106682629A (en) * | 2016-12-30 | 2017-05-17 | 佳都新太科技股份有限公司 | Identification number identification algorithm in complicated background |
| CN107346420A (en) * | 2017-06-19 | 2017-11-14 | 中国科学院信息工程研究所 | Text detection localization method under a kind of natural scene based on deep learning |
| CN107665354A (en) * | 2017-09-19 | 2018-02-06 | 北京小米移动软件有限公司 | Identify the method and device of identity card |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110689658A (en) * | 2019-10-08 | 2020-01-14 | 北京邮电大学 | Taxi bill identification method and system based on deep learning |
| CN111046883A (en) * | 2019-12-05 | 2020-04-21 | 吉林大学 | Intelligent evaluation method and system based on ancient coin image |
| CN111046883B (en) * | 2019-12-05 | 2022-08-23 | 吉林大学 | Intelligent assessment method and system based on ancient coin image |
| CN112418158A (en) * | 2020-02-11 | 2021-02-26 | 支付宝实验室(新加坡)有限公司 | System suitable for detecting identity card and device and processing method associated with same |
| CN111950554A (en) * | 2020-08-17 | 2020-11-17 | 深圳市丰巢网络技术有限公司 | Identification card identification method, device, equipment and storage medium |
| CN112183250A (en) * | 2020-09-14 | 2021-01-05 | 北京三快在线科技有限公司 | Character recognition method and device, storage medium and electronic equipment |
| CN112766255A (en) * | 2021-01-19 | 2021-05-07 | 上海微盟企业发展有限公司 | Optical character recognition method, device, equipment and storage medium |
| CN112949523A (en) * | 2021-03-11 | 2021-06-11 | 兴业银行股份有限公司 | Method and system for extracting key information from identity card image picture |
| CN113392827A (en) * | 2021-06-22 | 2021-09-14 | 平安健康保险股份有限公司 | Character recognition method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110135346A (en) | Identity card automatic identifying method and system based on deep learning | |
| CN110796082B (en) | Nameplate text detection method and device, computer equipment and storage medium | |
| CN111242126A (en) | Irregular text correction method and device, computer equipment and storage medium | |
| CN110674712A (en) | Interactive behavior recognition method and device, computer equipment and storage medium | |
| CN109389030A (en) | Facial feature points detection method, apparatus, computer equipment and storage medium | |
| CN108447061B (en) | Commodity information processing method and device, computer equipment and storage medium | |
| CN110751149B (en) | Target object labeling method, device, computer equipment and storage medium | |
| CN109285105A (en) | Method of detecting watermarks, device, computer equipment and storage medium | |
| CN108304243B (en) | Interface generation method and device, computer equipment and storage medium | |
| CN110147787A (en) | Bank's card number automatic identifying method and system based on deep learning | |
| CN109711419A (en) | Image processing method, device, computer equipment and storage medium | |
| CN111950422B (en) | Drawing identification method and related device | |
| CN112418278A (en) | Multi-class object detection method, terminal device and storage medium | |
| CN104537367B (en) | A kind of method of calibration of VIN codes | |
| CN109685013A (en) | The detection method and device of header key point in human body attitude identification | |
| CN111144358A (en) | Vehicle quality certificate verification method and device, computer equipment and storage medium | |
| CN116416626B (en) | Method, device, equipment and storage medium for acquiring circular seal data | |
| CN116543091B (en) | Visualization method, system, computer equipment and storage medium for power transmission line | |
| CN112580499A (en) | Text recognition method, device, equipment and storage medium | |
| CN112348116A (en) | Target detection method and device using spatial context and computer equipment | |
| CN109711381A (en) | Target identification method, device and the computer equipment of remote sensing images | |
| CN112818985A (en) | Text detection method, device, system and medium based on segmentation | |
| CN112766246A (en) | Document title identification method, system, terminal and medium based on deep learning | |
| CN114386504A (en) | Engineering drawing character recognition method | |
| CN111553431A (en) | Picture definition detection method and device, computer equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190816 |