CN109766825A - Handwritten signature identifying system neural network based - Google Patents
Handwritten signature identifying system neural network based Download PDFInfo
- Publication number
- CN109766825A CN109766825A CN201910015205.XA CN201910015205A CN109766825A CN 109766825 A CN109766825 A CN 109766825A CN 201910015205 A CN201910015205 A CN 201910015205A CN 109766825 A CN109766825 A CN 109766825A
- Authority
- CN
- China
- Prior art keywords
- handwritten signature
- module
- data
- acquisition module
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Collating Specific Patterns (AREA)
Abstract
The invention discloses handwritten signature identifying systems neural network based, in view of practical field, present invention seek to address that nowadays authentication when ensure handwritten signature identification reliability and real-time the problems such as.System includes: handwritten signature acquisition module;CPU module;Display module;Interface chip module.A large amount of practical handwritten signature data are acquired by the handwritten signature acquisition module and transfer data to memory module, then the data in memory module are imported in the CPU module through the interface chip module, the CPU module has the training data of importing using the convolutional neural networks algorithm model put up, unsupervised successive ignition preservation model after training, real time data, which is passed to, quickly can accurately identify whether the handwritten signature newly inputted is signed by signer in this model, and recognition result is shown on the display module, achieve the effect that identify the handwritten signature true and false in real time.
Description
Technical field
The present invention relates to handwritten signature identifying systems neural network based, belong to artificial intelligence image recognition technology neck
Domain more particularly to functional form handwritten signature identifying system.
Background technique
It is the problems such as in order to solve social recurrent leakage of personal information instantly, especially at full speed in current science and technology
The information age of development also becomes necessary and important for the certification of personally identifiable information.In daily life, bank takes
Money, unit discrepancy, hotel's disengaging and online trading etc. require correct identity information;And under Internet era, it is public
Information security is particularly important, such as system, the administration of justice, e-finance, national security etc. require more accurate identity mirror
It is fixed.In this regard, personal biology characteristics identification technology also becomes research hotspot, wherein handwritten signature it is more stable as human body and
It is easier to the biological behavior characteristic obtained, the important function of oneself is commonly used and played always in various fields, but
It is that its simplicity is also it by infringement notch, is often utilized by criminal.Therefore, in order to meet mankind's present society
Personal for protection and public cybersecurity demand, develops more safe and simple, effective and high practicability identity information
Authentication mode has huge realistic meaning and practical value.
Traditional handwritten signature identification only identifies " being ", and be not " to or not ", if wanting to be accurately identified
" to or not ", then need the sample the physical features such as the image of handwritten signature, the order of strokes observed in calligraphy, fluency and pressure and its actual signature
This information compares, to identify the true and false of handwritten signature.This identification technology taken time and effort is not able to satisfy nowadays real
The requirement that existing handwritten signature quickly accurately identifies.
Summary of the invention
In view of the above-mentioned problems, being asked in reliability and the real-time etc. for carrying out ensuring handwritten signature identification when authentication
Topic, the technical problem to be solved in the present invention is to provide handwritten signature identifying systems neural network based.
The technical solution adopted by the present invention to solve the technical problems is: handwritten signature identification neural network based system
System, comprising: handwritten signature acquisition module, CPU module, display module and interface chip module, wherein
Handwritten signature acquisition module, for acquiring a large amount of handwritten signature data;
CPU module, for building and in the training and identification of data for network model;
Display module shows true and false result later for identification;
Interface chip module, for the data-linkage between handwritten signature acquisition module and CPU module.
Further, the handwritten signature acquisition module includes acquisition module and memory module, wherein
The acquisition module is using electromagnetical type touch screen, for collected handwritten signature data to be passed through center control
Unit processed is transferred to the memory module of storage handwritten signature data;
The memory module, for storing a large amount of practical handwritten signature image datas and being transmitted to CPU module.
Further, the CPU module is building and to not based on convolutional neural networks algorithm progress Extended Model
It is trained and tests with data set, wherein
The data set includes practical handwritten signature data set and public individual Chinese character data set and draws according to the ratio of 3:1
It is divided into experiment used test collection and training set;
The practical hand-written data collection is obtained by the handwritten signature acquisition module, and the common data sets derive from
The data set for the simulation handwritten signature that HWDB data set is spliced;
The training set may originate from practical hand-written data collection and HWDB common data sets, do the network with and without supervision respectively
Training, test set can only practical hand-written data collection described in source, and be with training dataset without the brand-new of duplicate the same category
Data.
Further, the display module is using UFB display screen, wherein
The UFB display screen, for the true and false visual result after identification to be shown, and picture quality with higher.
The invention has the benefit that acquiring a large amount of practical handwritten signature data by handwritten signature acquisition module and will count
According to memory module is transferred to, then the data in memory module are imported in CPU module through interface chip module, CPU module is adopted
The successive ignition training with and without supervision is carried out with training data of the convolutional neural networks algorithm model put up to importing
Preservation model afterwards, real time data, which is passed to, quickly can accurately identify whether the handwritten signature newly inputted is signer in this model
I is signed, and recognition result is shown to the effect for being finally reached identify the handwritten signature true and false in real time on the display module.
Detailed description of the invention
Fig. 1 is the overall structure diagram of the specific embodiment of the invention.
Fig. 2 is the structural schematic diagram of the handwritten signature acquisition module of the specific embodiment of the invention.
Fig. 3 is the structural schematic diagram of the CPU module of the specific embodiment of the invention.
Fig. 4 is the neural network algorithm execution flow chart of the specific embodiment of the invention.
Specific implementation method
In order to make the objectives, technical solutions and advantages of the present invention clearer, below by shown in the accompanying drawings specific
Embodiment describes the present invention.However, it should be understood that these descriptions are merely illustrative, and it is not intended to limit model of the invention
It encloses.In the following description, descriptions of well-known structures and technologies are omitted, so as not to unnecessarily obscure the concept of the present invention.
It further needs exist for illustrating, only the parts related to the present invention are shown for ease of description, in attached drawing rather than entire infrastructure.
In the present embodiment, as shown in Figure 1, the specific embodiment of the invention uses following technical scheme: it includes hand-written
Signature acquisition module 1, for acquiring a large amount of handwritten signature data;CPU module 2, for network model build and data
Training and identification;Display module 3 shows true and false result later for identification;Interface chip module 4 is acquired for handwritten signature
Data-linkage between module and CPU module;Interface chip module 5, for the data rank between CPU module and display module
It connects.
In the present embodiment, as shown in Fig. 2, the handwritten signature acquisition module 1 includes acquisition module 6 and memory module
7。
In the present embodiment, the handwritten signature acquisition module 1 is that acquisition module 6 will be adopted by electromagnetical type touch screen 9
The handwritten signature data collected are transferred to the memory module 7 of storage handwritten signature data by central control unit 8.
In the present embodiment, as shown in figure 3, the handwritten signature figure that the CPU module 2 first imports memory module 7
Sheet data pre-processes, and normalizes image data, while matching a label to every a kind of handwritten signature, then using preservation
The training and test that good convolutional neural networks algorithm model is classified to image data and identified.Trained data use
Be that signer repeatedly signs the data of accumulation, and new signature is first identified as test data later, if signer
I is signed, then is stored in the history signature record of the signer and as training data later.It is used in an experiment
To signature have 17 classes, training data has 3672, and test data has 1020.
In the present embodiment, as shown in figure 4, the neural network algorithm of the CPU module 2 includes data standard in an experiment
Standby and network model two parts.
In the present embodiment, there are two types of data sets for the data preparation, are practical handwritten signature data, are from HWDB
Common data sets, practical handwritten signature data by pretreatment after can be divided into training data and test data in proportion, by it
In training data be sent in network model and carry out Training, the common data sets of HWDB are single handwritten Chinese character, by it
It carries out 2-3 Chinese character splicing and carries out unsupervised training to simulate handwritten signature data and then be sent in network model.
In the present embodiment, the network model is there are three types of model structure, is based on convolutional neural networks, the is
The innovatory algorithm of classic algorithm LeNet-5, second is the algorithm that joined residual error function, the third is and Recognition with Recurrent Neural Network
The algorithm of binary channels network is formed, the targeted data characteristics of three kinds of network models is different, lead to the difference of final recognition effect,
Final test data can preferentially be chosen by filtering algorithm.
In the present embodiment, the display module 3 will appear identification knot using UFB display screen on a display screen
Fruit correctness, and can there are two button, if correct click of identification continues button, if identification mistake click re-enter by
Button.
In the present embodiment, the interface chip module 4 is that the data in memory module 7 are transferred to CPU module 2.
In the present embodiment, the interface chip module 5 is that the recognition result after identifying CPU module 2 is shown
Display module 3.
The above specific embodiment introduction and describe basic principles and main features and advantages of the present invention of the invention.
It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, described in above embodiments and description
Merely illustrate the principles of the invention, without departing from the spirit and scope of the present invention, those skilled in the art are when can root
Various corresponding changes and modifications are made according to the present invention, these changes and improvements all fall within the protetion scope of the claimed invention.
Claims (4)
1. handwritten signature identifying system neural network based, it is characterised in that: the handwritten signature identifying system includes: hand-written
Signature acquisition module, CPU module, display module and interface chip module, wherein
The handwritten signature acquisition module, for acquiring a large amount of handwritten signature data;
The CPU module, for network model build and the training and identification of data;
The display module shows true and false result later for identification;
The interface chip module, for the data-linkage between handwritten signature acquisition module and CPU module.
2. handwritten signature identifying system neural network based according to claim 1, it is characterised in that: the handwritten signature
Acquisition module includes acquisition module and memory module, wherein
The acquisition module is using electromagnetical type touch screen, for collected handwritten signature data are single by center control
Member is transferred to the memory module of storage handwritten signature data;
The memory module, for storing a large amount of practical handwritten signature image datas and being transmitted to CPU module.
3. handwritten signature identifying system neural network based according to claim 1, it is characterised in that: the CPU module
It is that building and different data collection being trained and tested for Extended Model is carried out based on convolutional neural networks algorithm, wherein
The data set includes practical handwritten signature data set and public individual Chinese character data set and according to the ratio cut partition of 3:1 is
Test used test collection and training set;
The practical hand-written data collection is obtained by the handwritten signature acquisition module, and the common data sets derive from HWDB
The data set for the simulation handwritten signature that data set is spliced;
The training set may originate from practical hand-written data collection and HWDB common data sets, do the network training with and without supervision respectively,
Test set can only practical hand-written data collection described in source, and be the brand-new data with training dataset without duplicate the same category.
4. handwritten signature identifying system neural network based according to claim 1, it is characterised in that: the display module
Using UFB display screen, wherein
The UFB display screen, for the true and false visual result after identification to be shown, and picture quality with higher.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910015205.XA CN109766825A (en) | 2019-01-08 | 2019-01-08 | Handwritten signature identifying system neural network based |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910015205.XA CN109766825A (en) | 2019-01-08 | 2019-01-08 | Handwritten signature identifying system neural network based |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109766825A true CN109766825A (en) | 2019-05-17 |
Family
ID=66453487
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910015205.XA Pending CN109766825A (en) | 2019-01-08 | 2019-01-08 | Handwritten signature identifying system neural network based |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109766825A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112347981A (en) * | 2020-11-26 | 2021-02-09 | 国网山东省电力公司建设公司 | Signature identification method and system |
CN112906829A (en) * | 2021-04-13 | 2021-06-04 | 成都四方伟业软件股份有限公司 | Digital recognition model construction method and device based on Mnist data set |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101593324A (en) * | 2009-06-17 | 2009-12-02 | 浙江师范大学 | The network multi-level measures and procedures for the examination and approval and system based on dependable computing application technique |
CN105740786A (en) * | 2016-01-25 | 2016-07-06 | 中国科学院深圳先进技术研究院 | Identity identification method and device of writer |
CN106408039A (en) * | 2016-09-14 | 2017-02-15 | 华南理工大学 | Off-line handwritten Chinese character recognition method carrying out data expansion based on deformation method |
WO2018210796A1 (en) * | 2017-05-15 | 2018-11-22 | Deepmind Technologies Limited | Neural network systems for action recognition in videos |
CN108875696A (en) * | 2018-07-05 | 2018-11-23 | 五邑大学 | The Off-line Handwritten Chinese Recognition method of convolutional neural networks is separated based on depth |
-
2019
- 2019-01-08 CN CN201910015205.XA patent/CN109766825A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101593324A (en) * | 2009-06-17 | 2009-12-02 | 浙江师范大学 | The network multi-level measures and procedures for the examination and approval and system based on dependable computing application technique |
CN105740786A (en) * | 2016-01-25 | 2016-07-06 | 中国科学院深圳先进技术研究院 | Identity identification method and device of writer |
CN106408039A (en) * | 2016-09-14 | 2017-02-15 | 华南理工大学 | Off-line handwritten Chinese character recognition method carrying out data expansion based on deformation method |
WO2018210796A1 (en) * | 2017-05-15 | 2018-11-22 | Deepmind Technologies Limited | Neural network systems for action recognition in videos |
CN108875696A (en) * | 2018-07-05 | 2018-11-23 | 五邑大学 | The Off-line Handwritten Chinese Recognition method of convolutional neural networks is separated based on depth |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112347981A (en) * | 2020-11-26 | 2021-02-09 | 国网山东省电力公司建设公司 | Signature identification method and system |
CN112906829A (en) * | 2021-04-13 | 2021-06-04 | 成都四方伟业软件股份有限公司 | Digital recognition model construction method and device based on Mnist data set |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Escalante et al. | Modeling, recognizing, and explaining apparent personality from videos | |
CN108052577B (en) | Universal text content mining method, device, server and storage medium | |
Ferreira et al. | Towards automatic content analysis of social presence in transcripts of online discussions | |
WO2022105118A1 (en) | Image-based health status identification method and apparatus, device and storage medium | |
US11928985B2 (en) | Content pre-personalization using biometric data | |
WO2020237872A1 (en) | Method and apparatus for testing accuracy of semantic analysis model, storage medium, and device | |
CN103258157A (en) | On-line handwriting authentication method and system based on finger information | |
CN113722474A (en) | Text classification method, device, equipment and storage medium | |
WO2021139316A1 (en) | Method and apparatus for establishing expression recognition model, and computer device and storage medium | |
CN105095415A (en) | Method and apparatus for confirming network emotion | |
CN109670559A (en) | Recognition methods, device, equipment and the storage medium of handwritten Chinese character | |
CN111428448A (en) | Text generation method and device, computer equipment and readable storage medium | |
CN106227836B (en) | Unsupervised joint visual concept learning system and unsupervised joint visual concept learning method based on images and characters | |
US9830533B2 (en) | Analyzing and exploring images posted on social media | |
CN109766825A (en) | Handwritten signature identifying system neural network based | |
CN116824278A (en) | Image content analysis method, device, equipment and medium | |
CN113742488A (en) | Embedded knowledge graph completion method and device based on multitask learning | |
Maneriker et al. | SYSML: StYlometry with Structure and Multitask Learning: Implications for Darknet forum migrant analysis | |
Zhu et al. | NAGNet: A novel framework for real‐time students' sentiment analysis in the wisdom classroom | |
Zhang et al. | # HowYouTagTweets: Learning user hashtagging preferences via personalized topic attention | |
CN110689249A (en) | Comprehensive analysis system and analysis method for literacy capability of mathematical disciplines | |
Theophilo et al. | Explainable artificial intelligence for authorship attribution on social media | |
CN116976321A (en) | Text processing method, apparatus, computer device, storage medium, and program product | |
CN113610080B (en) | Cross-modal perception-based sensitive image identification method, device, equipment and medium | |
KR101607128B1 (en) | Providing related problem method about incorrect answer |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190517 |
|
WD01 | Invention patent application deemed withdrawn after publication |