CN107958217A - A kind of fingerprint classification identifying system and method based on deep learning - Google Patents
A kind of fingerprint classification identifying system and method based on deep learning Download PDFInfo
- Publication number
- CN107958217A CN107958217A CN201711211237.4A CN201711211237A CN107958217A CN 107958217 A CN107958217 A CN 107958217A CN 201711211237 A CN201711211237 A CN 201711211237A CN 107958217 A CN107958217 A CN 107958217A
- Authority
- CN
- China
- Prior art keywords
- fingerprint
- image
- characteristic
- classification
- module
- 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
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1365—Matching; Classification
- G06V40/1376—Matching features related to ridge properties or fingerprint texture
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1347—Preprocessing; Feature extraction
- G06V40/1359—Extracting features related to ridge properties; Determining the fingerprint type, e.g. whorl or loop
Abstract
The invention discloses a kind of fingerprint classification identifying system based on deep learning, the system comprises：Image capture module, image pre-processing module, CNN deep learning modules, characteristic extracting module, template memory module and characteristic matching module.The present invention carries out classification self-identifying based on CNN convolutional neural networks to fingerprint image, same type fingerprint characteristic image is carried out again on the basis of the classification of quick line type to compare, error rate is lower compared with conventional method, identification is more accurate, and in the case of the comparison of large-scale data amount, fingerprint classification recognition methods based on deep learning has more preferable expressive force, faster more accurate.
Description
Technical field
The present invention relates to fingerprint classification identification technology field, more particularly to a kind of fingerprint classification identification based on deep learning
System and method.
Background technology
On living things feature recognition field, fingerprint identification technology is the most ripe and most widely used living things feature recognition
One of technology, since it is with unchangeable property, uniqueness and convenience, and technology maturation, collection simplicity and cost ratio are relatively low
It is the features such as honest and clean, maximum in the market accounting of living things feature recognition at present, and it is widely used in identity authentication, information security, access
Control, quick payment etc..In daily life, the visible fingerprint identification technology such as gate inhibition, attendance, unlock of smart mobile phone
Application.
Automated Fingerprint Identification System (Automatic fingerprint identification system, AFIS) profit
Personal identification is identified with the features such as uniqueness of fingerprint, generality, convenience and unchangeable property, all kinds of AFIS into
During row fingerprint recognition, it (is typically 1 that fingerprint to be certified, which usually requires to be compared with the template fingerprint in database,：The ratio of N
It is right).Compare that number is more to be brought huge time-consuming to reduce, it is quick again that AFIS often carries out fingerprint classification according to Finger print
Screening, i.e., fingerprint to be certified need to only be compared with the template fingerprint of same line type.
With the increase of database, the difference of similar line type becomes larger and the boundary between different line types thickens,
Especially for incomplete fingerprint and the fingerprint there are much noise, their line type judge increasingly difficult.Conventional fingerprint
The accuracy rate of line type sorting algorithm is directly influenced be subject to individual features extraction algorithm.In magnanimity fingerprint base, similar line type refers to
Line metamorphosis significantly increases, and inhomogeneity line type boundary thickens, and it is difficult suitable only by the feature of Manual definition classify
Should whole finger print datas.One kind is disclosed in the Chinese invention patent specification of Publication No. CN105975909A and is based on six classes
The fingerprint three-level sorting technique of crestal line number and fractal dimension between Finger print feature, fingerprint singularity, but calculate image FRACTAL DIMENSION
Several algorithm generally existings is larger to high dimension image calculation error, it is computationally intensive the shortcomings of.
To solve the problems, such as the coupling of the feature extraction of line type Classification and Identification and Manual definition, the present invention proposes that one kind directly exists
In fingerprint artwork carry out line type identification algorithm, using convolutional neural networks (Convolutional Neural Networks,
CNN) ability of Automatic Feature Extraction obtains line type feature from a large amount of finger print data learnings, and passes through the design to training data
Enable the network to adapt to the diversity of fingerprint, the robustness of boosting algorithm.In addition, multiple dimensioned network model averaging method can make point
Class accuracy is further lifted.
The content of the invention
The purpose of the present invention is to solve shortcoming existing in the prior art, and propose a kind of based on deep learning
Fingerprint classification identifying system and method.
To achieve these goals, present invention employs following technical solution：
A kind of fingerprint classification identifying system based on deep learning, the system comprises：Image capture module, image are located in advance
Manage module, CNN deep learning modules, characteristic extracting module, template memory module and characteristic matching module.
Preferably, described image acquisition module is different from general pressing touch sensing fingerprint imaging, Image Acquisition mould
Block is made of light source, high-definition camera and imaging sensor, takes contactless high-definition shooting to obtain finger print image,
The problem of can easily being deformed upon to avoid the fingerprint obtained by pressing finger at the same time.
Preferably, described image pretreatment module is connected with image capture module, and image pre-processing module is to Image Acquisition
The fingerprint image that module collects is pre-processed, and pretreatment includes determining the fingerprint image that finger print acquisition module collects
Position, image segmentation, image enhancement, image binaryzation and micronization processes, finally obtain the fingerprint image being disposed.
Preferably, the CNN deep learnings module is the registered fingerprint image for using line type calibration information as reference
Image simultaneously generates training data, according to training data design and training convolutional neural networks, then inputs fingerprint image, refers to according to reference
The generation method generation input finger print data of print image, and the finger print data is input in the CNN of training completion, according to CNN
Output judged and finally obtain bow, left dustpan, right dustpan and struggle against 4 class fingerprint image results.
Preferably, the characteristic extracting module is connected with CNN deep learning modules, and characteristic extracting module is classified through CNN
Take the fingerprint minutiae point in the fingerprint image of completion, and the sampled point in then being extracted again in fingerprint image on crestal line, finally carries
The convex closure of print image sampling point, generates the finger of the convex closure containing fingerprint minutiae, all crestal lines up-sampling point and sampled point
Line characteristic image.
Preferably, the template memory module is connected with characteristic extracting module, and user's input simultaneously passes through classification, feature extraction
The registered fingerprint characteristic image obtained afterwards is stored into database as fingerprint characteristic image template, for the finger inputted to system
Print image is compared.
Preferably, the characteristic matching module is connected with characteristic extracting module and template memory module respectively, when there is finger
After feature that line is inputted and classified by CNN, feature extraction unit takes the fingerprint, characteristic matching unit extracts the mould with fingerprint pattern
The fingerprint template characteristic image of plate storage unit is contrasted with the fingerprint characteristic image inputted, judges to input fingerprint characteristic and mould
Whether plate fingerprint characteristic matches.
Present invention also offers a kind of fingerprint classification based on deep learning to know method for distinguishing, and the method includes as follows
Implementation steps：
Step S1, fingerprint image, including registered fingerprint and input fingerprint are gathered using image capture module；
Step S2, fingerprint image is pre-processed, obtains the refinement fingerprint image easy to CNN Classification and Identifications；
Step S3, using the registered fingerprint image for having line type calibration information as reference image and training data is generated, root
CNN convolutional neural networks, the CNN of training structure are designed according to training data；
Step S4, fingerprint image generation finger print data is inputted, and finger print data is input in the CNN of training completion, root
Judged according to CNN outputs and finally obtain the fingerprint image result of classification；
Step S5, feature point extraction is carried out in the fingerprint image that classification is completed, generation contains fingerprint minutiae, all ridges
The fingerprint characteristic image of the convex closure of line up-sampling point and sampled point；
Step S6, it is special to be used as fingerprint for user's input and the registered fingerprint characteristic image by being obtained after classification, feature extraction
Image template storage is levied into database, the fingerprint characteristic image for being inputted to system is compared；
Step S7, when there is fingerprint to be identified to input, characteristic matching unit is according to similar in type-collection template storage unit
The fingerprint characteristic image of type is compared with the fingerprint characteristic image inputted, judges that input fingerprint characteristic is characterized in template fingerprint
No matching.
Compared with prior art, the beneficial effects of the invention are as follows：The present invention is based on CNN convolutional neural networks to fingerprint image
Classification self-identifying is carried out, carrying out same type fingerprint characteristic image again on the basis of the classification of quick line type compares, and error rate relatively passes
System method is lower, and identification is more accurate, and in the case of the comparison of large-scale data amount, the fingerprint classification based on deep learning
Recognition methods has more preferable expressive force, faster more accurate.
Brief description of the drawings
Fig. 1 is fingerprint pattern classification schematic diagram of the present invention；
Fig. 2 is fingerprint classification identifying system structure chart of the present invention；
Fig. 3 is fingerprint image pondization processing schematic diagram of the present invention；
Fig. 4 is CNN deep learnings Finger print classification process figure of the present invention；
Fig. 5 is CNN deep learnings network structure of the present invention.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, below in conjunction with specific embodiment, to this
Invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, not
For limiting the present invention.
Embodiment 1
As shown in Figs. 1-5, the fingerprint pattern classification schematic diagram wherein shown in Fig. 1, will refer to according to Henry fingerprint classifications system
Print image is divided into 5 types according to its topological structure, that is, bend (Arch), account bow (Tented-arch), left dustpan (Left loop),
Right dustpan (Right loop) and bucket (Whorl), according to statistics, the NATURAL DISTRIBUTION ratio of this 5 kinds of fingerprint patterns is respectively 3.7%,
2.9%, 33.8%, 31.7% and 27.9%.Since bow and account bend, 2 class fingerprint proportions are very low, in actual AFIS,
The two is usually combined into one kind, i.e., Finger print is divided into bow, 4 classes of left dustpan, right dustpan and bucket.
By fingerprint classification, fingerprint characteristic image to be identified only need to the fingerprint comparison of same type, i.e., in finger print data
Matched in a subset in storehouse, avoid and carry out largely comparing one by one in fingerprint database.
As shown in Fig. 2, the present invention fingerprint classification identifying system structure core be employ CNN deep learnings network into
The classification self-identifying of row fingerprint image, is illustrated by the fingerprint classification recognition methods of the present invention and with reference to system below.
Method for distinguishing is known in a kind of fingerprint classification based on deep learning, and implementation step is as follows：
Step S1, utilize image capture module collection fingerprint image, including registered fingerprint and input fingerprint；
Specifically, Image Acquisition difference of the present invention to fingerprint is used in general pressing touch sensing fingerprint imaging
Contactless high speed high-definition shooting is imaged to obtain the fingerprint image of finger, can avoid fingerprint image because pressing deformation；
Wherein, registered fingerprint is for the fingerprint of known fingerprint type, and collection registered fingerprint image is as Finger print reference
Image, input fingerprint verify whether matched fingerprint to be to be identified.
Step S2, fingerprint image pre-processed, obtain the refinement fingerprint image easy to CNN Classification and Identifications；
Specifically, the fingerprint image collected is pre-processed, including carries out zone location, image point to fingerprint image
Cut, image enhancement, image binaryzation and micronization processes, acquisition are suitable for the refinement fingerprint image of CNN Classification and Identifications；
Wherein, in fingerprint image preprocessing, on the premise of fingerprint image center is ensured it is unified by image cutting-out into
512*512 pixel sizes are as original training data, since fingerprint will not change with the change of resolution ratio, to make convolutional Neural
Network is adapted to different resolution, improves algorithm robustness, the pond of different multiples is carried out to initial data, as shown in figure 3, this hair
The bright pond that 2 times and 3 times are respectively adopted, obtains the image of 256*256 pixels and 171*171 pixels；
In addition, in order to overcome training data insufficient, while prevent over-fitting from producing, data volume is further expanded
Fill, the data of Chi Huahou are cut, select upper left, lower-left, upper right, bottom right and intermediate region to be cut as training data respectively
Size after cutting is 224*224 pixels and 148*148 pixels respectively.
Step S3, using the registered fingerprint image for having line type calibration information as reference image and generate training data, root
CNN convolutional neural networks, the CNN of training structure are designed according to training data；
Specifically, the calibration of line type information is carried out to the registered fingerprint image of known fingerprint type, registered fingerprint image is
Data training, the i.e. template data as the identification of line type are carried out as reference image；
As shown in figure 4, determining convolution window according to registered fingerprint view data, and CNN convolutional neural networks are built, built
Registered fingerprint view data is trained into rear, CNN points using registered fingerprint image as line pattern plate are obtained after the completion of training
Class device.
Step S4, input fingerprint image generation finger print data, and by finger print data be input to training completion CNN in, root
Judged according to CNN outputs and finally obtain the fingerprint image result of classification；
Further, in order to Finger print carry out Classification and Identification, the present invention devise one using 224*224 pixels with
The multiple dimensioned CNN deep learnings model that two kinds of dimension images of 148*148 pixels are trained, the model include convolutional layer, Chi Hua
Layer, full articulamentum and SVM classifier layer, as shown in Figure 5；
Wherein, convolutional layer (Convolutions) is represented with C, and C (1,2,3) is three convolutional layers, and numeral is big for convolution kernel
Small, S is convolution step-length；Pond layer (Subsampling) represents that P (1,2,3) is three pond layers with P, and numeral is big for pond core
Small, S is pond step-length；Full connection (1,2) are two full articulamentums, and SVM is grader layer；
In the network structure of the present invention, exemplified by inputting 148*148 pixel fingerprint images, fingerprint image passes through convolution
The characteristic image of 64 35*35 is obtained more than layer C1 after the layer P1 of pond, then by second of convolution with obtaining 128 after pondization
The characteristic image of 15*15, finally obtains the characteristic image of 512 5*5 by the operation of third time convolution pondization；
Characteristic point connect entirely and obtains 32 tie points in second articulamentum by characteristic image in full articulamentum, leads to
Cross full articulamentum image classified by SVM classifier after export 4 class Finger print data；
Specifically, in whole convolutional neural networks, the fingerprint image file inputted first passes through C1 volumes of convolutional layer
Product processing, convolution algorithm can reduce noise, the feature of enhancing input fingerprint image, Convolution Formula definition to a certain extent
For：
Cx=f (∑ I*kx+bx)
Wherein, I is input, k for can training convolutional window, b is can train additivity to bias, and Cx is the convolution characteristic pattern obtained, f
For activation primitive, k and b is a random value when initial, is adjusted in the continuous training of network；
Obtain operating, it is necessary to carry out a pondization after convolutional layer, it is therefore an objective to while information in fingerprint is retained subtract
Few calculation amount, compared with last layer, the operation of pond layer reduces image resolution ratio, but obtains more plane spaces, has
Beneficial to the detection of feature, the formula of pondization operation is defined as：
Sx+1=σ (ωx+1*Cx+bx+1)
Wherein, ω is the biasing of multiplying property, and b is can train additivity to bias, and Cx is the corresponding upper convolutional layer of the pond layer, Sx+1
For the pond characteristic pattern of acquisition, σ (x) is Sig-moid functions, makes its slight change to median sensitive and improves resolution；
For each input x, there are a predicted value y ' and actual value y, loss function τ (y ', y) is describing both
Between actual loss, target makes this loss function in whole training set be minimum；
Q (z, ω)=τ (fω(x),y)
Wherein, Q (z, ω) is average loss function, fω(x) predicted value for being sample x when weights are ω, En(f) it is experience
Risk function, τ (f (xi),yi) be some sample loss function, empirical risk function is weighing the training effect of network；
A SVM classifier is included between articulamentum and output layer entirely at second, the grader is via convolutional Neural net
The feature that network is extracted is trained, its kernel function is linear, and type C-SVM, had both avoided the cumbersome of manual extraction feature
With unilateral, and the maximum marginal classification feature of SVM is combined, there is stronger classification capacity relative to original CNN networks,
SVM classifier is classified and is exported to the image for completing training according to pre-set 4 class Finger print.
Step S5, in the fingerprint image that classification is completed carry out feature point extraction, generation contains fingerprint minutiae, all ridges
The fingerprint characteristic image of the convex closure of line up-sampling point and sampled point；
Specifically, characteristic extracting module takes the fingerprint minutiae point in the fingerprint image that classification is completed, then again in fingerprint
Sampled point in being extracted in image on crestal line, the convex closure for the image sampling point that finally takes the fingerprint.
Step S6, user's input and by classification, that the registered fingerprint characteristic image obtained after feature extraction be used as fingerprint is special
Image template storage is levied into database, the fingerprint characteristic image for being inputted to system is compared；
The fingerprint characteristic image of user's registration is using as the database of the template data of matching identification storage to generic
In subset, verification is compared as the fingerprint characteristic image inputted with system.
Step S7, when there is fingerprint to be identified to input, characteristic matching unit is according to similar in type-collection template storage unit
The fingerprint characteristic image of type is compared with the fingerprint characteristic image inputted, judges that input fingerprint characteristic is characterized in template fingerprint
No matching.
Specifically, when system has fingerprint input to be identified, obtain and pre-processed after the fingerprint image, classified, feature
Extraction, obtains system input fingerprint characteristic image to be identified, characteristic matching unit is according to similar in type-collection modular unit
The template fingerprint characteristic image of type is compared, and judges to input whether fingerprint characteristic matches sound with template fingerprint feature.
The foregoing is only a preferred embodiment of the present invention, but protection scope of the present invention be not limited thereto,
Any one skilled in the art the invention discloses technical scope in, technique according to the invention scheme and its
Inventive concept is subject to equivalent substitution or change, should be covered by the protection scope of the present invention.
Claims (8)
- A kind of 1. fingerprint classification identifying system based on deep learning, it is characterised in that the system comprises：Image Acquisition mould Block, image pre-processing module, CNN deep learning modules, characteristic extracting module, template memory module and characteristic matching module.
- A kind of 2. fingerprint classification identifying system based on deep learning according to claim 1, it is characterised in that the figure As pressing touch sensing fingerprint imaging of the acquisition module from is different, image capture module by light source, high-definition camera and Imaging sensor is formed, and takes contactless high-definition shooting to obtain finger print image, while can be to avoid because pressing hand Refer to the problem of fingerprint obtained easily deforms upon.
- A kind of 3. fingerprint classification identifying system based on deep learning according to claim 1, it is characterised in that the figure As pretreatment module is connected with image capture module, the fingerprint image that image pre-processing module collects image capture module into Row pretreatment, pretreatment include the fingerprint image that finger print acquisition module collects is positioned, image segmentation, image enhancement, Image binaryzation and micronization processes, finally obtain the fingerprint image being disposed.
- 4. a kind of fingerprint classification identifying system based on deep learning according to claim 1, it is characterised in that described CNN deep learning modules are to use line type to demarcate the registered fingerprint image of information as reference image and generate training data, According to training data design and training convolutional neural networks, then fingerprint image is inputted, given birth to according to the generation method with reference to fingerprint image It is input into input finger print data, and by the finger print data in the CNN of training completion, according to CNN outputs judge and final Bent, left dustpan, right dustpan and bucket 4 class fingerprint image results.
- A kind of 5. fingerprint classification identifying system based on deep learning according to claim 1, it is characterised in that the spy Sign extraction module is connected with CNN deep learning modules, and characteristic extracting module is extracted in the fingerprint image completed through CNN classification to be referred to Line minutiae point, the sampled point in then being extracted again in fingerprint image on crestal line, the convex closure for the image sampling point that finally takes the fingerprint, Generate the fingerprint characteristic image of the convex closure containing fingerprint minutiae, all crestal lines up-sampling point and sampled point.
- A kind of 6. fingerprint classification identifying system based on deep learning according to claim 1, it is characterised in that the mould Plate memory module is connected with characteristic extracting module, user's input and the registered fingerprint feature by being obtained after classification, feature extraction Image is stored into database as fingerprint characteristic image template, for the fingerprint image that system inputs to be compared.
- 7. a kind of fingerprint classification identifying system based on deep learning according to claim 1, it is characterised in that described Characteristic matching module is connected with characteristic extracting module and template memory module respectively, when having fingerprint to input and classified by CNN, special Sign extraction unit takes the fingerprint after feature, and the fingerprint template that characteristic matching unit extracts the template storage unit with fingerprint pattern is special Sign image is contrasted with the fingerprint characteristic image inputted, judges to input whether fingerprint characteristic matches with template fingerprint feature.
- 8. a kind of fingerprint classification recognition methods according to claim 1 based on deep learning, it is characterised in that described Method includes following implementation steps：Step S1, fingerprint image, including registered fingerprint and input fingerprint are gathered using image capture module；Step S2, fingerprint image is pre-processed, obtains the refinement fingerprint image easy to CNN Classification and Identifications；Step S3, using the registered fingerprint image for having line type calibration information as reference image and training data is generated, according to instruction Practice design data CNN convolutional neural networks, the CNN of training structure；Step S4, fingerprint image generation finger print data is inputted, and finger print data is input in the CNN of training completion, according to CNN Output is judged and finally obtains the fingerprint image result of classification；Step S5, feature point extraction is carried out in the fingerprint image that classification is completed, generation is containing on fingerprint minutiae, all crestal lines The fingerprint characteristic image of the convex closure of sampled point and sampled point；Step S6, user inputs and the registered fingerprint characteristic image by being obtained after classification, feature extraction is used as fingerprint characteristic figure Into database, the fingerprint characteristic image for being inputted to system is compared for picture template storage；Step S7, when there is fingerprint to be identified to input, characteristic matching unit is according to same type in type-collection template storage unit Fingerprint characteristic image is compared with the fingerprint characteristic image inputted, judge to input fingerprint characteristic and template fingerprint feature whether Match somebody with somebody.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711211237.4A CN107958217A (en) | 2017-11-28 | 2017-11-28 | A kind of fingerprint classification identifying system and method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711211237.4A CN107958217A (en) | 2017-11-28 | 2017-11-28 | A kind of fingerprint classification identifying system and method based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107958217A true CN107958217A (en) | 2018-04-24 |
Family
ID=61962336
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711211237.4A Pending CN107958217A (en) | 2017-11-28 | 2017-11-28 | A kind of fingerprint classification identifying system and method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107958217A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108764093A (en) * | 2018-05-21 | 2018-11-06 | 中国人民解放军战略支援部队信息工程大学 | Non-contact fingerprint picture collector based on camera and method |
| CN108846327A (en) * | 2018-05-29 | 2018-11-20 | 中国人民解放军总医院 | A kind of intelligent distinguishing system and method for mole and melanoma |
| CN109145834A (en) * | 2018-08-27 | 2019-01-04 | 河南丰泰光电科技有限公司 | A kind of fingerprint recognition neural network based and verification method |
| CN109508692A (en) * | 2018-11-30 | 2019-03-22 | 深圳大学 | A kind of gender identification method and system based on 3D fingerprint image |
| CN110164003A (en) * | 2019-05-09 | 2019-08-23 | 深圳市英泰斯达智能技术有限公司 | A kind of fingerprint control switch, system and method |
| CN111160424A (en) * | 2019-12-16 | 2020-05-15 | 南方电网科学研究院有限责任公司 | NFC equipment fingerprint authentication method and system based on CNN image identification |
| CN112087443A (en) * | 2020-09-04 | 2020-12-15 | 浙江大学 | Intelligent detection method for sensing data abnormity under large-scale industrial sensing network information physical attack |
| CN112287732A (en) * | 2019-07-25 | 2021-01-29 | 上海车景网络科技有限公司 | Fingerprint quick comparison method and system |
| CN112966610A (en) * | 2021-03-09 | 2021-06-15 | 金陵科技学院 | Elevator intelligence fingerprint identification system based on increment study |
| CN113128364A (en) * | 2021-03-31 | 2021-07-16 | 杭州电子科技大学 | Fingerprint biological key generation method based on deep neural network coding |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1581206A (en) * | 2003-08-14 | 2005-02-16 | 上海一维科技有限公司 | Finger print identification method |
| CN101751555A (en) * | 2009-09-25 | 2010-06-23 | 中国科学院自动化研究所 | Deformation fingerprint identification method and system |
| US20150110340A1 (en) * | 2013-10-23 | 2015-04-23 | Gracenote, Inc. | Identifying video content via color-based fingerprint matching |
| CN105303179A (en) * | 2015-10-28 | 2016-02-03 | 小米科技有限责任公司 | Fingerprint identification method and fingerprint identification device |
| CN106557761A (en) * | 2016-11-29 | 2017-04-05 | 深圳天珑无线科技有限公司 | A kind of method and device of fingerprint recognition |
-
2017
- 2017-11-28 CN CN201711211237.4A patent/CN107958217A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1581206A (en) * | 2003-08-14 | 2005-02-16 | 上海一维科技有限公司 | Finger print identification method |
| CN101751555A (en) * | 2009-09-25 | 2010-06-23 | 中国科学院自动化研究所 | Deformation fingerprint identification method and system |
| US20150110340A1 (en) * | 2013-10-23 | 2015-04-23 | Gracenote, Inc. | Identifying video content via color-based fingerprint matching |
| CN105303179A (en) * | 2015-10-28 | 2016-02-03 | 小米科技有限责任公司 | Fingerprint identification method and fingerprint identification device |
| CN106557761A (en) * | 2016-11-29 | 2017-04-05 | 深圳天珑无线科技有限公司 | A kind of method and device of fingerprint recognition |
Non-Patent Citations (4)
| Title |
|---|
| 彭博 等: "基于深度学习的车标识别方法研究", 《计算机科学》 * |
| 江璐 等: "基于深度卷积神经网络的指纹纹型分类算法", 《中国科学院大学学报》 * |
| 清华大学自动化系 编: "《智能信息处理和智能控制》", 31 December 1998, 杭州：浙江科学技术出版社 * |
| 陈成霞 等: "基于高清摄像技术的非接触式指纹识别系统", 《通讯世界》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108764093A (en) * | 2018-05-21 | 2018-11-06 | 中国人民解放军战略支援部队信息工程大学 | Non-contact fingerprint picture collector based on camera and method |
| CN108846327A (en) * | 2018-05-29 | 2018-11-20 | 中国人民解放军总医院 | A kind of intelligent distinguishing system and method for mole and melanoma |
| CN109145834A (en) * | 2018-08-27 | 2019-01-04 | 河南丰泰光电科技有限公司 | A kind of fingerprint recognition neural network based and verification method |
| WO2020107922A1 (en) * | 2018-11-30 | 2020-06-04 | 深圳大学 | 3d fingerprint image-based gender recognition method and system |
| CN109508692A (en) * | 2018-11-30 | 2019-03-22 | 深圳大学 | A kind of gender identification method and system based on 3D fingerprint image |
| CN109508692B (en) * | 2018-11-30 | 2020-06-16 | 深圳大学 | Gender identification method and system based on 3D fingerprint image |
| CN110164003A (en) * | 2019-05-09 | 2019-08-23 | 深圳市英泰斯达智能技术有限公司 | A kind of fingerprint control switch, system and method |
| CN112287732A (en) * | 2019-07-25 | 2021-01-29 | 上海车景网络科技有限公司 | Fingerprint quick comparison method and system |
| CN111160424A (en) * | 2019-12-16 | 2020-05-15 | 南方电网科学研究院有限责任公司 | NFC equipment fingerprint authentication method and system based on CNN image identification |
| CN112087443A (en) * | 2020-09-04 | 2020-12-15 | 浙江大学 | Intelligent detection method for sensing data abnormity under large-scale industrial sensing network information physical attack |
| CN112966610A (en) * | 2021-03-09 | 2021-06-15 | 金陵科技学院 | Elevator intelligence fingerprint identification system based on increment study |
| CN113128364A (en) * | 2021-03-31 | 2021-07-16 | 杭州电子科技大学 | Fingerprint biological key generation method based on deep neural network coding |
| CN113128364B (en) * | 2021-03-31 | 2024-02-02 | 杭州电子科技大学 | Fingerprint biological key generation method based on deep neural network coding |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107958217A (en) | A kind of fingerprint classification identifying system and method based on deep learning | |
| US11657525B2 (en) | Extracting information from images | |
| CN110348319B (en) | Face anti-counterfeiting method based on face depth information and edge image fusion | |
| US11450152B2 (en) | Detection of manipulated images | |
| AU2016214084B2 (en) | Systems and methods for performing fingerprint based user authentication using imagery captured using mobile devices | |
| CN105518709B (en) | The method, system and computer program product of face for identification | |
| US20230252662A1 (en) | Extracting information from images | |
| WO2016150240A1 (en) | Identity authentication method and apparatus | |
| CN109376631A (en) | A kind of winding detection method and device neural network based | |
| CN107808120A (en) | Glasses localization method, device and storage medium | |
| CN104951940A (en) | Mobile payment verification method based on palmprint recognition | |
| CN111582197A (en) | Living body based on near infrared and 3D camera shooting technology and face recognition system | |
| CN109376717A (en) | Personal identification method, device, electronic equipment and the storage medium of face comparison | |
| CN108846269A (en) | One kind is towards manifold identity identifying method and identification authentication system | |
| Benziane et al. | Dorsal hand vein identification based on binary particle swarm optimization | |
| CN106156739A (en) | A kind of certificate photo ear detection analyzed based on face mask and extracting method | |
| CN106803307B (en) | Banknote face value orientation identification method based on template matching | |
| KR101037759B1 (en) | Robust Face Recognition Method using AAM and Gabor Feature Vectors | |
| Sharifi | Score-level-based face anti-spoofing system using handcrafted and deep learned characteristics | |
| Verma et al. | Static Signature Recognition System for User Authentication Based Two Level Cog, Hough Tranform and Neural Network | |
| Khan et al. | Dorsal hand vein biometric using Independent Component Analysis (ICA) | |
| Okokpujie et al. | A COMPARATIVE ANALYSIS PERFORMANCE OF DATA AUGMENTATION ON AGE-INVARIANT FACE RECOGNI-TION USING PRETRAINED RESIDUAL NEURAL NETWORK | |
| Sivapriyan et al. | Analysis of facial recognition techniques | |
| Tian et al. | A practical iris recognition algorithm | |
| Kara et al. | Detection of Fake Indian Currency Using Deep Convolutional Neural Network |
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: 20180424 |
|
| RJ01 | Rejection of invention patent application after publication |