CN110147732A - Refer to vein identification method, device, computer equipment and storage medium - Google Patents
Refer to vein identification method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN110147732A CN110147732A CN201910304095.9A CN201910304095A CN110147732A CN 110147732 A CN110147732 A CN 110147732A CN 201910304095 A CN201910304095 A CN 201910304095A CN 110147732 A CN110147732 A CN 110147732A
- Authority
- CN
- China
- Prior art keywords
- finger
- vein
- sample
- finger vein
- vein image
- 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
- 210000003462 vein Anatomy 0.000 title claims abstract description 381
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000003860 storage Methods 0.000 title claims abstract description 14
- 238000012549 training Methods 0.000 claims abstract description 101
- 239000013598 vector Substances 0.000 claims abstract description 68
- 238000000605 extraction Methods 0.000 claims abstract description 11
- 230000001537 neural effect Effects 0.000 claims abstract description 9
- 239000000523 sample Substances 0.000 claims description 103
- 238000013528 artificial neural network Methods 0.000 claims description 65
- 239000013074 reference sample Substances 0.000 claims description 30
- 230000006870 function Effects 0.000 claims description 22
- 230000008569 process Effects 0.000 claims description 22
- 238000004590 computer program Methods 0.000 claims description 13
- 210000004218 nerve net Anatomy 0.000 claims description 2
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 239000000284 extract Substances 0.000 description 6
- 210000000653 nervous system Anatomy 0.000 description 6
- 238000013527 convolutional neural network Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 210000001367 artery Anatomy 0.000 description 4
- 238000001514 detection method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000004044 response Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000001151 other effect Effects 0.000 description 1
- 230000000149 penetrating effect Effects 0.000 description 1
Classifications
-
- 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
- G06N3/08—Learning methods
-
- 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
- G06V40/13—Sensors therefor
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
- Image Analysis (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention discloses a kind of finger vein identification method, device, computer equipment and storage mediums, are related to artificial intelligence field.This refers to that vein identification method includes: to obtain finger vein image to be identified and finger vein image to be compared；Finger vein image to be identified and finger vein image to be compared are input to and referred in hand vein recognition model, wherein refers to that hand vein recognition model is obtained using twin neural metwork training；By referring to hand vein recognition model extraction first eigenvector relevant to finger vein image to be identified and second feature vector relevant with finger vein image to be compared；Finger hand vein recognition is carried out according to first eigenvector and second feature vector, judges whether finger vein image to be identified and finger vein image to be compared come from same root finger.Refer to that vein identification method can accurately carry out referring to vein image identification using this.
Description
[technical field]
The present invention relates to artificial intelligence field more particularly to a kind of finger vein identification method, device, computer equipment and deposit
Storage media.
[background technique]
Refer to that vein identification technology is that resulting veinprint image carries out personal knowledge after penetrating finger using near infrared ray
Other biological identification technology.In various biological identification technologies, because referring to that vein identification technology is the biology that can't see using outside
The technology that internal feature is identified, so antifalsification with higher.However, referring to the recognition accuracy of hand vein recognition still at present
It is universal lower.
[summary of the invention]
In view of this, the embodiment of the invention provides a kind of finger vein identification method, device, computer equipment and storages to be situated between
Matter refers to that the recognition accuracy of hand vein recognition is lower to solve the problems, such as at present.
In a first aspect, the embodiment of the invention provides a kind of finger vein identification methods, comprising:
Obtain finger vein image to be identified and finger vein image to be compared；
The finger vein image to be identified and the finger vein image to be compared are input to and referred in hand vein recognition model,
In, the finger hand vein recognition model is obtained using twin neural metwork training；
By finger hand vein recognition model extraction first eigenvector relevant to the finger vein image to be identified and
Second feature vector relevant to the finger vein image to be compared；
Finger hand vein recognition is carried out according to the first eigenvector and the second feature vector, judges the finger to be identified
Whether vein image and the finger vein image to be compared come from same root finger.
The aspect and any possible implementation manners as described above, it is further provided a kind of implementation, described by institute
It states finger vein image to be identified and the finger vein image to be compared is input to before referring in hand vein recognition model, further includes:
Acquisition refers to vein training sample；
The finger vein training sample is input in the twin neural network and is trained, obtains producing in training process
Raw penalty values；
The network parameter that the twin neural network is updated according to the penalty values obtains the finger hand vein recognition model.
The aspect and any possible implementation manners as described above, it is further provided a kind of implementation, it is described will be described
Refer to that vein training sample is input in the twin neural network to be trained, obtains the penalty values generated in training process, wrap
It includes:
The finger vein training sample is input in the twin neural network as unit of triple and is trained,
In, triple includes a reference sample, a similar sample and foreign peoples's sample, the reference sample be
Randomly selected finger vein training sample in the finger vein training sample, the similar sample are affiliated user and the ginseng
The identical finger vein training sample of user belonging to sample is examined, foreign peoples's sample is affiliated user and the reference sample institute
The different finger vein training sample of the user of category；
Calculate what the reference sample, the similar sample and foreign peoples's sample exported in the twin neural network
Feature vector；
Based on described eigenvector, the penalty values generated in training process are calculated using triple loss function,
In, the triple loss function is I table
Show that triple group number, N indicate total group of number of triple,Indicate that L2 norm is squared,Indicate i-th group three
The feature vector that the reference sample of tuple exports in twin neural network,Indicate the similar sample of i-th group of triple
The feature vector exported in twin neural network,Indicate foreign peoples's sample of i-th group of triple in twin neural network
The feature vector of middle output, α indicate interval threshold, and when the value in+expression [...] is greater than 0, taking the value greater than 0 is penalty values,
When less than 0, penalty values are taken as 0.
The aspect and any possible implementation manners as described above, it is further provided a kind of implementation, it is described according to institute
The network parameter that penalty values update the twin neural network is stated, the finger hand vein recognition model is obtained, comprising:
According to the penalty values, the network parameter of the twin neural network is updated using back-propagation algorithm；
When the changing value of the network parameter, which is respectively less than, to be stopped iteration threshold or reach trained the number of iterations, stop updating
The network parameter obtains the finger hand vein recognition model.
The aspect and any possible implementation manners as described above, it is further provided a kind of implementation, it is described according to institute
It states first eigenvector and the second feature vector carries out finger hand vein recognition, judge the finger vein image to be identified and described
Whether finger vein image to be compared comes from same root finger, comprising:
Calculate the metric range between the first eigenvector and the second feature vector；
If the metric range is within the scope of preset metric range, it is determined that the finger vein image to be identified and described
Finger vein image to be compared is derived from same root finger.
Second aspect, the embodiment of the invention provides a kind of finger vein identification devices, comprising:
Refer to that vein image obtains module, for obtaining finger vein image to be identified and finger vein image to be compared；
Input module, for the finger vein image to be identified and the finger vein image to be compared to be input to finger vein
In identification model, wherein the finger hand vein recognition model is obtained using twin neural metwork training；
Feature vector obtains module, for passing through the finger hand vein recognition model extraction and the finger vein image to be identified
Relevant first eigenvector and second feature vector relevant to the finger vein image to be compared；
Judgment module is sentenced for carrying out finger hand vein recognition according to the first eigenvector and the second feature vector
Break the finger vein image to be identified and whether the finger vein image to be compared comes from same root finger.
The third aspect, a kind of computer equipment, including memory, processor and storage are in the memory and can be
The computer program run on the processor, the processor realize above-mentioned finger hand vein recognition when executing the computer program
The step of method.
Fourth aspect, the embodiment of the invention provides a kind of computer readable storage mediums, comprising: computer program, institute
State the step of above-mentioned finger vein identification method is realized when computer program is executed by processor.
In embodiments of the present invention, finger vein image to be identified and finger vein image to be compared are obtained first, with using to
Comparison refers to vein image as reference, judges whether finger vein image to be identified is from same root with finger vein image to be compared
Finger；Then finger vein image to be identified and finger vein image to be compared are input to what use twin nervous system training obtained
Refer to hand vein recognition model, by refer to relevant to the finger vein image to be identified first eigenvector of hand vein recognition model extraction and with
The relevant second feature vector of finger vein image to be compared, the finger hand vein recognition model energy obtained using the training of twin nervous system
Enough primary just extract obtains first eigenvector and second feature vector, and can protrude finger vein image to be identified and to be compared
Refer to the difference between vein image, helps to improve the accuracy rate for referring to hand vein recognition model；Finally according to first eigenvector and
Second feature vector carries out finger hand vein recognition, and it is same to judge whether finger vein image to be identified and finger vein image to be compared come from
Root finger obtains accurately referring to hand vein recognition result by the difference between feature to judging.
[Detailed description of the invention]
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for this field
For those of ordinary skill, without any creative labor, it can also be obtained according to these attached drawings other attached
Figure.
Fig. 1 is a flow chart of one embodiment of the invention middle finger vein identification method；
Fig. 2 is a schematic diagram of one embodiment of the invention middle finger vein identification device；
Fig. 3 is a schematic diagram of computer equipment in one embodiment of the invention.
[specific embodiment]
For a better understanding of the technical solution of the present invention, being retouched in detail to the embodiment of the present invention with reference to the accompanying drawing
It states.
It will be appreciated that described embodiments are 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
Its embodiment, shall fall within the protection scope of the present invention.
The term used in embodiments of the present invention is only to be not intended to be limiting merely for for the purpose of describing particular embodiments
The present invention.In the embodiment of the present invention and the "an" of singular used in the attached claims, " described " and "the"
It is also intended to including most forms, unless the context clearly indicates other meaning.
It should be appreciated that term "and/or" used herein is only a kind of identical field for describing affiliated partner, table
Show there may be three kinds of relationships, for example, A and/or B, can indicate: individualism A exists simultaneously A and B, individualism B this three
Kind situation.In addition, character "/" herein, typicallys represent the relationship that forward-backward correlation object is a kind of "or".
It will be appreciated that though default model may be described using term first, second, third, etc. in embodiments of the present invention
It encloses, but these preset ranges should not necessarily be limited by these terms.These terms are only used to for preset range being distinguished from each other out.For example,
In the case where not departing from range of embodiment of the invention, the first preset range can also be referred to as the second preset range, similarly,
Second preset range can also be referred to as the first preset range.
Depending on context, word as used in this " if " can be construed to " ... when " or " when ...
When " or " in response to determination " or " in response to detection ".Similarly, depend on context, phrase " if it is determined that " or " if detection
(condition or event of statement) " can be construed to " when determining " or " in response to determination " or " when the detection (condition of statement
Or event) when " or " in response to detection (condition or event of statement) ".
Fig. 1 shows a flow chart of the present embodiment middle finger vein identification method.This refers to that vein identification method can be applicable to finger
In vein recognition system, this can be used when judging whether any two fingers vein image comes from same root finger and refers to hand vein recognition
System is realized.This refers to that vein recognition system specifically can be applicable in computer equipment, wherein the computer equipment is can be with user
The equipment for carrying out human-computer interaction, the including but not limited to equipment such as computer, smart phone and plate.As shown in Figure 1, this refers to that vein is known
Other method includes the following steps:
S10: finger vein image to be identified and finger vein image to be compared are obtained.
This programme can identify any two fingers vein image, judge that the two finger vein images identified are
It is no to come from same root finger, refer to that the eigenvector method of vein image is different from preparatory typing is needed, this programme can be directly right
The finger vein image that scene obtains is identified.Wherein, finger vein image to be identified and finger vein image to be compared are opposite
Concept, finger vein image during referring to hand vein recognition as reference is finger vein image to be compared, and another Zhang Jinhang refers to
The finger vein image of hand vein recognition is finger vein image to be identified, and finger vein image to be identified and finger vein image to be compared can be with
Converted according to demand, such as using it is to be identified finger vein image as refer to finger vein image when, the finger vein figure to be identified
As being converted to finger vein image to be compared, and original finger vein image to be compared is then converted to finger vein image to be identified.It can
To understand ground, the corresponding finger of fingers vein image to be compared identify can be it is known be also possible to it is unknown, wherein finger mark
Know the mark for referring to unique difference finger, is known, and finger to be identified when the corresponding finger of fingers vein image to be compared identifies
When vein image and finger vein image to be compared come from same root finger, then the corresponding hand of finger vein image to be identified can be determined
Index is known, and is unknown, and finger vein image to be identified and finger to be compared when the corresponding finger of fingers vein image to be compared identifies
When vein image comes from same root finger, although finger vein image to be identified or the finger to be compared for referring to vein image can not be determined
Mark, but still can determine that finger vein image to be identified and finger vein image to be compared from same root finger, correspond to identical
Finger mark.
In one embodiment, finger vein image to be identified and finger vein image to be compared are obtained, with quiet using finger to be compared
Arteries and veins image judges whether finger vein image to be identified is from same root finger with finger vein image to be compared as reference.
S20: finger vein image to be identified and finger vein image to be compared are input to and referred in hand vein recognition model, wherein
Refer to that hand vein recognition model is obtained using twin neural metwork training.
Wherein, twin neural network is a kind of neural network including two or more identical sub-neural networks,
Sub-neural network network parameter having the same.
Specifically, refer to that hand vein recognition model is can to judge whether two different finger vein images come from same root finger
Model, using twin neural network in advance training obtain, wherein the twin neural network specifically include two it is identical son mind
Through network, which is specially convolutional neural networks, and in other words, twin neural network includes two identical convolution
Neural network, convolutional neural networks have preferable effect for extracting characteristics of image, can be improved the knowledge for referring to hand vein recognition model
Other accuracy rate.
In one embodiment, finger vein image to be identified and finger vein image to be compared are input to and refer to hand vein recognition model
In, it can be regarded as finger vein image to be identified and finger vein image to be compared being separately input to two identical sub- nerve nets
In network, the finger vein image to be identified of input and finger vein image to be compared can be handled simultaneously, and does not interfere with each other.
Further, it before step S20, i.e., is input to by finger vein image to be identified and finger vein image to be compared
Before referring in hand vein recognition model, further includes:
S21: acquisition refers to vein training sample.
In one embodiment, training refers to that hand vein recognition model need to obtain and refers to vein training sample.This refers to vein training sample
It including the finger vein image from different fingers, while also including the finger vein image from identical finger, for from same
The finger vein image of root finger identifies the identical finger of well in advance.Refer to vein training sample by obtaining, it can be according to pre-
First ready-made finger mark training refers to hand vein recognition model, so that it is identical to train obtained finger hand vein recognition model that can identify
The finger vein image of finger mark.
S22: will refer to that vein training sample is input in twin neural network and is trained, and obtain generating in training process
Penalty values.
Wherein, the finger vein training sample of input be divided into training set, verifying collection and test set, and test set, verifying collection and
Training set is not overlapped between each other, can effectively improve the generalization ability of model in this way, and guarantee refers to the practical of hand vein recognition model
Property, enable and refer to that hand vein recognition model identifies any two different finger vein images, judges whether from same
Root finger.It is to be appreciated that carrying out referring to the finger vein image to be identified and finger vein to be compared of hand vein recognition under actual scene
Image typically refers to not to be occurred in vein training sample.
In one embodiment, will refer to that vein training sample is input in twin neural network to be trained, it will be according to preparatory
The loss function of setting is indicated to measure the loss generated in training process with specific penalty values.
S23: updating the network parameter of twin neural network according to penalty values, obtains referring to hand vein recognition model.
In one embodiment, constantly model will be updated according to penalty values during model training, and will obtain referring to quiet
Arteries and veins identification model.Specifically, the sub-neural network in twin neural network, which will synchronize, is updated, the net between sub-neural network
Network parameter is still identical.
Refer to that hand vein recognition model is based on twin neural metwork training and obtains, remains the original network knot of twin neural network
Structure can identify any two fingers vein image, it is not limited to refer to the finger vein image in vein training sample, prop up
Any scene held in real life carries out finger hand vein recognition, this is that the neural network of other single channels cannot accomplish
's.
In step S21-S23, the specific embodiment that a kind of training obtains referring to hand vein recognition model is provided, by using
Based on the network structure of twin neural network, the label about classification is done to finger vein image in advance and follows opener mark
Standard, so that test set, verifying collect and the processing such as training set is not overlapped between each other are trained finger hand vein recognition, so that training
Obtained finger hand vein recognition model has stronger generalization ability, can identify the different knowledges for referring to vein image under any scene
Not, whether the finger vein image that accurate judgement is identified comes from same root finger.
Further, in step S22, it will refer to that vein training sample is input in twin neural network and be trained, obtain
The penalty values generated into training process, specifically include:
S221: it will refer to that vein training sample is input in twin neural network as unit of triple and be trained, wherein
One triple includes a reference sample, a similar sample and foreign peoples's sample, and reference sample is to refer to vein training
Randomly selected finger vein training sample in sample, similar sample are that affiliated user is identical with user belonging to reference sample
Refer to that vein training sample, foreign peoples's sample are the different finger vein training sample of user belonging to affiliated user and reference sample
This.
Wherein, triple is used to indicate to refer to that vein training sample progress model training can be further for the mode of unit
Characterization refers to the similarity degree between vein training sample, can expand the distance between class, reduces the distance in class, so that model is instructed
Similar sample and foreign peoples's sample can be further discriminated between when practicing.It is to be appreciated that similar sample can be allowed during model training
The distance between it is closer, the distance of foreign peoples's sample is farther, use the triple for the training method of unit can be improved refer to it is quiet
The separating capacity of arteries and veins identification model helps to improve the order of accuarcy for referring to that hand vein recognition model is identified.
S222: reference sample is calculated, the feature vector that similar sample and foreign peoples's sample export in twin neural network.
Wherein, twin neural network includes two identical convolutional neural networks, and two convolutional neural networks are in model
It is not interfere with each other in renewal process.What the feature vector of twin neural network output represented is the output of training process as a result, this is defeated
Result and reality it is expected and (if output result is same class, are actually desired for not being same class) there may be certain loss out,
Therefore, it is necessary to analyze the feature vector of the output, twin neural network is adjusted according to the feature vector of the output, so that
Refer to that the accuracy rate of hand vein recognition model is higher.
S223: being based on feature vector, the penalty values generated in training process be calculated using triple loss function,
In, triple loss function is I indicates three
Tuple group number, N indicate total group of number of triple,Indicate that L2 norm is squared,Indicate i-th group of triple
The feature vector that is exported in twin neural network of reference sample,Indicate the similar sample of i-th group of triple twin
The feature vector exported in raw neural network,Indicate that foreign peoples's sample of i-th group of triple is defeated in twin neural network
Feature vector out, α indicate interval threshold, and when the value in+expression [...] is greater than 0, taking the value greater than 0 is penalty values, less than 0
When, penalty values are taken as 0.
In one embodiment, triple loss function is Wherein, α indicate the distance between reference sample and similar sample and reference sample and foreign peoples's sample it
Between distance a minimum interval threshold value.The triple loss function describes between class distance (the distance between foreign peoples's sample)
The loss generated with inter- object distance (the distance between similar sample) in training, can be according to the loss to twin neural network
It is adjusted, so that the between class distance of training sample is bigger, inter- object distance is smaller, further increases the knowledge for referring to hand vein recognition model
Other effect.
In step S221-S223, the specific embodiment for calculating the penalty values generated in training process is provided, by adopting
Triple loss function is set up with the mode as unit of triple, which can expand when calculating penalty values
Distance between class, reduce class in distance, for about refer to hand vein recognition have preferable effect, enable model training when into
One step distinguishes similar sample and foreign peoples's sample, the loss that can accurately generate in descriptive model training process.Using this three
The finger hand vein recognition model that the penalty values progress model modification that tuple loss function obtains obtains has stronger about finger vein
The separating capacity of image, can the finger vein image that is identified of accurate judgement whether come from same root finger.
Further, in step S23, the network parameter of twin neural network is updated according to penalty values, obtains referring to vein
Identification model specifically includes:
S231: according to penalty values, the network parameter of twin neural network is updated using back-propagation algorithm.
Wherein, back-propagation algorithm (Back propagation algorithm, abbreviation BP algorithm) is suitable for multilayer
A kind of learning algorithm of neuroid, it is established on the basis of gradient descent method, can quickly and accurately update depth
Network parameter in habit.
In one embodiment, it according to penalty values, is successively returned using back-propagation algorithm to each in twin neural network
The network parameter that each layer of convolutional neural networks is modified, so that mapping ability representated by the network parameter after change, energy
It is enough that more accurately the feature vector of input is mapped, it exports more representative of the feature vector for referring to vein image feature, so that
The finger hand vein recognition model that training obtains can make accurate knowledge to finger vein image to be identified and finger vein image to be compared
Not.
S232: when the changing value of network parameter, which is respectively less than, to be stopped iteration threshold or reach trained the number of iterations, stop more
New network parameter obtains referring to hand vein recognition model.
In step S231-S232, a kind of network parameter for updating twin neural network according to penalty values is provided, is referred to
The specific embodiment of hand vein recognition model can be improved by using back-propagation algorithm and refer to hand vein recognition model training
Efficiency；And it is respectively less than according to the changing value of network parameter and stops iteration threshold or stop in time more when reaching trained the number of iterations
New network parameter can effectively improve the recognition accuracy for referring to hand vein recognition model.
S30: by refer to relevant to the finger vein image to be identified first eigenvector of hand vein recognition model extraction and with to
Comparison refers to the relevant second feature vector of vein image.
In one embodiment, refer to that hand vein recognition model is obtained based on twin neural metwork training, can extract simultaneously
The relevant first eigenvector of finger vein image and the relevant second feature vector of finger vein image to be compared to be identified, and extract
It does not interfere with each other in the process, without the process that repeated characteristic extracts, finger hand vein recognition can be quickly carried out under actual scene.
S40: carrying out finger hand vein recognition according to first eigenvector and second feature vector, judges finger vein image to be identified
Same root finger whether is come from finger vein image to be compared.
It is to be appreciated that referring to that hand vein recognition model is one from feature is extracted to the integrated mould for exporting recognition result
Type, will be according to the first eigenvector of extraction and second feature vector determination finger vein figure to be identified in referring to hand vein recognition model
Whether picture and finger vein image to be compared come from same root finger.
Further, in step s 40, finger hand vein recognition is carried out according to first eigenvector and second feature vector, sentenced
Break finger vein image to be identified and whether finger vein image to be compared come from same root finger, specifically includes:
S41: the metric range between first eigenvector and second feature vector is calculated.
Wherein, metric range refers to the distance of tightness degree between two feature vectors of measurement.Specifically, metric range can
To be Euclidean distance, COS distance or mahalanobis distance isometry distance.It is European specifically using Euclidean distance in this implementation
Distance can embody the tightness degree on geometric distance, and the similarity degree for measuring image has higher accuracy rate.
S42: if metric range is within the scope of preset metric range, it is determined that finger vein image to be identified and finger to be compared
Vein image is derived from same root finger.
When metric range is using Euclidean distance, and Euclidean distance is within the scope of preset metric range, then can
With think finger vein image to be identified and finger vein image to be compared be to the extent permitted by the error it is identical, can be accurately
Determine that finger vein image to be identified and finger vein image to be compared are derived from same root finger.
Step S41-S42, which is provided, judges whether finger vein image to be identified and finger vein image to be compared come from same root
The specific embodiment of finger assesses finger vein image to be identified and finger vein image to be compared from the angle of metric range
Similarity degree, accurate recognition result can be obtained.
In embodiments of the present invention, finger vein image to be identified and finger vein image to be compared are obtained first, with using to
Comparison refers to vein image as reference, judges whether finger vein image to be identified is from same root with finger vein image to be compared
Finger；Then finger vein image to be identified and finger vein image to be compared are input to what use twin nervous system training obtained
Refer to hand vein recognition model, by refer to relevant to the finger vein image to be identified first eigenvector of hand vein recognition model extraction and with
The relevant second feature vector of finger vein image to be compared, the finger hand vein recognition model energy obtained using the training of twin nervous system
Enough primary just extract obtains first eigenvector and second feature vector, and can protrude finger vein image to be identified and to be compared
Refer to the difference between vein image, helps to improve the accuracy rate for referring to hand vein recognition model；Finally according to first eigenvector and
Second feature vector carries out finger hand vein recognition, and it is same to judge whether finger vein image to be identified and finger vein image to be compared come from
Root finger obtains accurately referring to hand vein recognition result by the difference between feature to judging.
It should be understood that the size of the serial number of each step is not meant that the order of the execution order in above-described embodiment, each process
Execution sequence should be determined by its function and internal logic, the implementation process without coping with the embodiment of the present invention constitutes any limit
It is fixed.
Based on finger vein identification method provided in embodiment, the embodiment of the present invention further provides the realization above method
The Installation practice of each step and method in embodiment.
Fig. 2 shows the functional block diagrams for referring to vein identification device correspondingly with embodiment middle finger vein identification method.Such as
Shown in Fig. 2, this refers to that vein identification device includes referring to that vein image obtains module 10, input module 20, feature vector and obtains module
30 and judgment module 40.Wherein, refer to that vein image obtains module 10, input module 20, feature vector and obtains module 30 and judgement
The realization function of module 40 step corresponding with embodiment middle finger vein identification method corresponds, to avoid repeating, this implementation
Example is not described in detail one by one.
Refer to that vein image obtains module 10, for obtaining finger vein image to be identified and finger vein image to be compared.
Input module 20 refers to hand vein recognition mould for finger vein image to be identified and finger vein image to be compared to be input to
In type, wherein refer to that hand vein recognition model is obtained using twin neural metwork training.
Feature vector obtains module 30, for by referring to that hand vein recognition model extraction is relevant to finger vein image to be identified
First eigenvector and second feature vector relevant to finger vein image to be compared.
Judgment module 40 judges for carrying out finger hand vein recognition according to first eigenvector and second feature vector wait know
Do not refer to whether vein image and finger vein image to be compared come from same root finger.
Optionally, refer to that vein identification device further includes referring to vein training sample acquiring unit, penalty values acquiring unit and referring to
Hand vein recognition model acquiring unit.
Refer to vein training sample acquiring unit, refers to vein training sample for obtaining.
Penalty values acquiring unit is trained for that will refer to that vein training sample is input in twin neural network, obtains
The penalty values generated in training process.
Refer to that hand vein recognition model acquiring unit is obtained for updating the network parameter of twin neural network according to penalty values
Refer to hand vein recognition model.
Optionally, penalty values acquiring unit includes that input subelement, feature vector computation subunit and penalty values obtain son
Unit.
Subelement being inputted, being carried out for will refer to that vein training sample is input in twin neural network as unit of triple
Training, wherein triple includes a reference sample, a similar sample and foreign peoples's sample, reference sample be
Refer to that randomly selected finger vein training sample in vein training sample, similar sample are belonging to affiliated user and reference sample
The identical finger vein training sample of user, foreign peoples's sample are that the different finger of user belonging to affiliated user and reference sample is quiet
Arteries and veins training sample.
Feature vector computation subunit, for calculating reference sample, similar sample and foreign peoples's sample in twin neural network
The feature vector of middle output.
Penalty values obtain subelement, and for being based on feature vector, training process is calculated using triple loss function
The penalty values of middle generation, wherein triple loss function is I indicates that triple group number, N indicate total group of number of triple,Indicate L2
Norm is squared,Indicate the feature vector that the reference sample of i-th group of triple exports in twin neural network,Indicate the feature vector that the similar sample of i-th group of triple exports in twin neural network,Indicate i-th
The feature vector that foreign peoples's sample of group triple exports in twin neural network, α indicate interval threshold ,+indicate in [...]
Value when being greater than 0, taking value greater than 0 is penalty values, and when less than 0, penalty values are taken as 0.
Optionally, refer to that hand vein recognition model acquiring unit includes updating subelement and referring to that hand vein recognition model acquisition is single
Member.
Subelement is updated, for updating the network parameter of twin neural network using back-propagation algorithm according to penalty values.
Refer to that hand vein recognition model obtains subelement, is respectively less than for the changing value when network parameter and stops iteration threshold or reach
To when training the number of iterations, stop updating network parameter, obtains referring to hand vein recognition model.
Optionally, judgment module 40 includes metric range computing unit and determination unit.
Metric range computing unit, for calculating the metric range between first eigenvector and second feature vector.
Determination unit, if for metric range within the scope of preset metric range, it is determined that finger vein image to be identified
Same root finger is derived from finger vein image to be compared.
In embodiments of the present invention, finger vein image to be identified and finger vein image to be compared are obtained first, with using to
Comparison refers to vein image as reference, judges whether finger vein image to be identified is from same root with finger vein image to be compared
Finger；Then finger vein image to be identified and finger vein image to be compared are input to what use twin nervous system training obtained
Refer to hand vein recognition model, by refer to relevant to the finger vein image to be identified first eigenvector of hand vein recognition model extraction and with
The relevant second feature vector of finger vein image to be compared, the finger hand vein recognition model energy obtained using the training of twin nervous system
Enough primary just extract obtains first eigenvector and second feature vector, and can protrude finger vein image to be identified and to be compared
Refer to the difference between vein image, helps to improve the accuracy rate for referring to hand vein recognition model；Finally according to first eigenvector and
Second feature vector carries out finger hand vein recognition, and it is same to judge whether finger vein image to be identified and finger vein image to be compared come from
Root finger obtains accurately referring to hand vein recognition result by the difference between feature to judging.
The present embodiment provides a computer readable storage medium, computer journey is stored on the computer readable storage medium
Sequence realizes embodiment middle finger vein identification method, to avoid repeating, herein not one by one when the computer program is executed by processor
It repeats.Alternatively, realizing each module/unit in embodiment middle finger vein identification device when the computer program is executed by processor
Function does not repeat one by one herein to avoid repeating.
Fig. 3 is the schematic diagram for the computer equipment that one embodiment of the invention provides.As shown in figure 3, the calculating of the embodiment
Machine equipment 50 includes: processor 51, memory 52 and is stored in the calculating that can be run in memory 52 and on processor 51
Machine program 53 realizes the finger vein identification method in embodiment when the computer program 53 is executed by processor 51, to avoid weight
It is multiple, it does not repeat one by one herein.Alternatively, realizing that embodiment middle finger hand vein recognition fills when the computer program 53 is executed by processor 51
The function of each model/unit does not repeat one by one herein in setting to avoid repeating.
Computer equipment 50 can be desktop PC, notebook, palm PC and cloud server etc. and calculate equipment.
Computer equipment 50 may include, but be not limited only to, processor 51, memory 52.It will be understood by those skilled in the art that Fig. 3 is only
It is only the example of computer equipment 50, does not constitute the restriction to computer equipment 50, may include more more or less than illustrating
Component, perhaps combine certain components or different components, for example, computer equipment can also include input-output equipment,
Network access equipment, bus etc..
Alleged processor 51 can be central processing unit (Central Processing Unit, CPU), can also be
Other general processors, digital signal processor (Digital Signal Processor, DSP), specific integrated circuit
(Application Specific Integrated Circuit, ASIC), field programmable gate array (Field-
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,
Discrete hardware components etc..General processor can be microprocessor or the processor is also possible to any conventional processor
Deng.
Memory 52 can be the internal storage unit of computer equipment 50, such as the hard disk or interior of computer equipment 50
It deposits.Memory 52 is also possible to the plug-in type being equipped on the External memory equipment of computer equipment 50, such as computer equipment 50
Hard disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card, flash card
(Flash Card) etc..Further, memory 52 can also both including computer equipment 50 internal storage unit and also including
External memory equipment.Memory 52 is for storing other programs and data needed for computer program and computer equipment.It deposits
Reservoir 52 can be also used for temporarily storing the data that has exported or will export.
It is apparent to those skilled in the art that for convenience of description and succinctly, only with above-mentioned each function
Can unit, module division progress for example, in practical application, can according to need and by above-mentioned function distribution by different
Functional unit, module are completed, i.e., the internal structure of device are divided into different functional unit or module, to complete above description
All or part of function.
The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although with reference to the foregoing embodiments
Invention is explained in detail, those skilled in the art should understand that: it still can be to aforementioned each implementation
Technical solution documented by example is modified or equivalent replacement of some of the technical features；And these modification or
Replacement, the spirit and scope for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution should all include
Within protection scope of the present invention.
Claims (10)
1. a kind of finger vein identification method, which is characterized in that the described method includes:
Obtain finger vein image to be identified and finger vein image to be compared；
The finger vein image to be identified and the finger vein image to be compared are input to and referred in hand vein recognition model, wherein
The finger hand vein recognition model is obtained using twin neural metwork training；
By finger hand vein recognition model extraction first eigenvector relevant to the finger vein image to be identified and with institute
State the relevant second feature vector of finger vein image to be compared；
Finger hand vein recognition is carried out according to the first eigenvector and the second feature vector, judges the finger vein to be identified
Whether image and the finger vein image to be compared come from same root finger.
2. the method according to claim 1, wherein it is described by the finger vein image to be identified and it is described to
Before comparison refers to that vein image is input in finger hand vein recognition model, further includes:
Acquisition refers to vein training sample；
The finger vein training sample is input in the twin neural network and is trained, obtains generating in training process
Penalty values；
The network parameter that the twin neural network is updated according to the penalty values obtains the finger hand vein recognition model.
3. according to the method described in claim 2, it is characterized in that, it is described the finger vein training sample is input to it is described twin
It is trained in raw neural network, obtains the penalty values generated in training process, comprising:
The finger vein training sample is input in the twin neural network as unit of triple and is trained, wherein
One triple includes a reference sample, a similar sample and foreign peoples's sample, and the reference sample is in institute
It states and refers to randomly selected finger vein training sample in vein training sample, the similar sample is affiliated user and the reference
The identical finger vein training sample of user belonging to sample, foreign peoples's sample are belonging to affiliated user and the reference sample
The different finger vein training sample of user；
Calculate the feature that the reference sample, the similar sample and foreign peoples's sample export in the twin neural network
Vector；
Based on described eigenvector, the penalty values generated in training process are calculated using triple loss function, wherein institute
Stating triple loss function is I indicates three
Tuple group number, N indicate total group of number of triple,Indicate that L2 norm is squared,Indicate i-th group of triple
The feature vector that is exported in twin neural network of reference sample,Indicate the similar sample of i-th group of triple twin
The feature vector exported in raw neural network,Indicate that foreign peoples's sample of i-th group of triple is defeated in twin neural network
Feature vector out, α indicate interval threshold, and when the value in+expression [...] is greater than 0, taking the value greater than 0 is penalty values, less than 0
When, penalty values are taken as 0.
4. according to the method described in claim 2, it is characterized in that, described update the twin nerve net according to the penalty values
The network parameter of network obtains the finger hand vein recognition model, comprising:
According to the penalty values, the network parameter of the twin neural network is updated using back-propagation algorithm；
When the changing value of the network parameter, which is respectively less than, to be stopped iteration threshold or reach trained the number of iterations, stop described in update
Network parameter obtains the finger hand vein recognition model.
5. the method according to claim 1, which is characterized in that described according to the first eigenvector
Finger hand vein recognition is carried out with the second feature vector, judges the finger vein image to be identified and the finger vein figure to be compared
It seem no from same root finger, comprising:
Calculate the metric range between the first eigenvector and the second feature vector；
If the metric range is within the scope of preset metric range, it is determined that the finger vein image to be identified and it is described to than
Same root finger is derived to finger vein image.
6. a kind of finger vein identification device, which is characterized in that described device includes:
Refer to that vein image obtains module, for obtaining finger vein image to be identified and finger vein image to be compared；
Input module, for the finger vein image to be identified and the finger vein image to be compared to be input to finger hand vein recognition
In model, wherein the finger hand vein recognition model is obtained using twin neural metwork training；
Feature vector obtains module, for related to the finger vein image to be identified by the finger hand vein recognition model extraction
First eigenvector and second feature vector relevant to the finger vein image to be compared；
Judgment module judges institute for carrying out finger hand vein recognition according to the first eigenvector and the second feature vector
It states finger vein image to be identified and whether the finger vein image to be compared comes from same root finger.
7. device according to claim 6, which is characterized in that described device further include:
Refer to vein training sample acquiring unit, refers to vein training sample for obtaining；
Penalty values acquiring unit is trained for the finger vein training sample to be input in the twin neural network,
Obtain the penalty values generated in training process；
Refer to hand vein recognition model acquiring unit, for updating the network parameter of the twin neural network according to the penalty values,
Obtain the finger hand vein recognition model.
8. device according to claim 7, which is characterized in that the penalty values acquiring unit includes:
Subelement is inputted, for the finger vein training sample to be input in the twin neural network as unit of triple
It is trained, wherein a triple includes a reference sample, a similar sample and foreign peoples's sample, described
Reference sample is the randomly selected finger vein training sample in the finger vein training sample, and the similar sample is affiliated
User's finger vein training sample identical with user belonging to the reference sample, foreign peoples's sample are affiliated user and institute
State the different finger vein training sample of user belonging to reference sample；
Feature vector computation subunit, for calculating the reference sample, the similar sample and foreign peoples's sample described
The feature vector exported in twin neural network；
Penalty values obtain subelement, and for being based on described eigenvector, training process is calculated using triple loss function
The penalty values of middle generation, wherein the triple loss function is I indicates that triple group number, N indicate total group of number of triple,Indicate L2
Norm is squared,Indicate the feature vector that the reference sample of i-th group of triple exports in twin neural network,Indicate the feature vector that the similar sample of i-th group of triple exports in twin neural network,Indicate i-th group
The feature vector that foreign peoples's sample of triple exports in twin neural network, α indicate interval threshold ,+indicate in [...]
When value is greater than 0, taking the value greater than 0 is penalty values, and when less than 0, penalty values are taken as 0.
9. a kind of computer equipment, including memory, processor and storage are in the memory and can be in the processor
The computer program of upper operation, which is characterized in that the processor realized when executing the computer program as claim 1 to
The step of any one of 5 finger vein identification method.
10. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, and feature exists
In realization refers to the step of vein identification method as described in any one of claim 1 to 5 when the computer program is executed by processor
Suddenly.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910304095.9A CN110147732A (en) | 2019-04-16 | 2019-04-16 | Refer to vein identification method, device, computer equipment and storage medium |
| PCT/CN2019/116472 WO2020211339A1 (en) | 2019-04-16 | 2019-11-08 | Finger vein recognition method and apparatus, and computer device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910304095.9A CN110147732A (en) | 2019-04-16 | 2019-04-16 | Refer to vein identification method, device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110147732A true CN110147732A (en) | 2019-08-20 |
Family
ID=67589800
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910304095.9A Pending CN110147732A (en) | 2019-04-16 | 2019-04-16 | Refer to vein identification method, device, computer equipment and storage medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN110147732A (en) |
| WO (1) | WO2020211339A1 (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110516745A (en) * | 2019-08-28 | 2019-11-29 | 北京达佳互联信息技术有限公司 | Training method, device and the electronic equipment of image recognition model |
| CN111008550A (en) * | 2019-09-06 | 2020-04-14 | 上海芯灵科技有限公司 | Identification method for finger vein authentication identity based on Multiple loss function |
| CN111079785A (en) * | 2019-11-11 | 2020-04-28 | 深圳云天励飞技术有限公司 | Image identification method and device and terminal equipment |
| CN111166070A (en) * | 2019-12-17 | 2020-05-19 | 五邑大学 | Medical storage cabinet based on finger vein authentication and management method thereof |
| CN111242951A (en) * | 2020-01-08 | 2020-06-05 | 上海眼控科技股份有限公司 | Vehicle detection method, device, computer equipment and storage medium |
| CN111767940A (en) * | 2020-05-14 | 2020-10-13 | 北京迈格威科技有限公司 | Target object identification method, device, equipment and storage medium |
| WO2020211339A1 (en) * | 2019-04-16 | 2020-10-22 | 平安科技（深圳）有限公司 | Finger vein recognition method and apparatus, and computer device and storage medium |
| CN112200159A (en) * | 2020-12-01 | 2021-01-08 | 四川圣点世纪科技有限公司 | Non-contact palm vein identification method based on improved residual error network |
| CN112200156A (en) * | 2020-11-30 | 2021-01-08 | 四川圣点世纪科技有限公司 | Vein recognition model training method and device based on clustering assistance |
| WO2023093838A1 (en) * | 2021-11-25 | 2023-06-01 | 北京字跳网络技术有限公司 | Super-resolution image processing method and apparatus, and device and medium |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113362070A (en) * | 2021-06-03 | 2021-09-07 | 中国工商银行股份有限公司 | Method, apparatus, electronic device, and medium for identifying operating user |
| CN113379779B (en) * | 2021-06-07 | 2023-04-07 | 华南理工大学 | Edge calculation method, apparatus, medium, and device of stack width learning system |
| CN113792632A (en) * | 2021-09-02 | 2021-12-14 | 广州广电运通金融电子股份有限公司 | Finger vein identification method, system and storage medium based on multi-party cooperation |
| CN113920532A (en) * | 2021-09-18 | 2022-01-11 | 浙江工业大学 | Finger vein identification method applying 2DPCA dimension reduction mechanism |
| CN114170687B (en) * | 2021-12-08 | 2024-05-07 | 山东大学 | Human skeleton action early recognition method and system based on guide information |
| CN114399763B (en) * | 2021-12-17 | 2024-04-16 | 西北大学 | Single-sample and small-sample micro-body paleobiological fossil image identification method and system |
| CN114998950B (en) * | 2022-08-01 | 2022-11-22 | 北京圣点云信息技术有限公司 | Vein encryption and identification method based on deep learning |
| CN118279578A (en) * | 2022-12-30 | 2024-07-02 | 同方威视科技江苏有限公司 | CT image processing method and device, and international express inspection method and device |
| CN116543330A (en) * | 2023-04-13 | 2023-08-04 | 北京京东乾石科技有限公司 | Crop information storage method, device, electronic equipment and computer readable medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106991368A (en) * | 2017-02-20 | 2017-07-28 | 北京大学 | A kind of finger vein checking personal identification method based on depth convolutional neural networks |
| CN107392114A (en) * | 2017-06-29 | 2017-11-24 | 广州智慧城市发展研究院 | A kind of finger vein identification method and system based on neural network model |
| CN107480785A (en) * | 2017-07-04 | 2017-12-15 | 北京小米移动软件有限公司 | The training method and device of convolutional neural networks |
| CN107967442A (en) * | 2017-09-30 | 2018-04-27 | 广州智慧城市发展研究院 | A kind of finger vein identification method and system based on unsupervised learning and deep layer network |
| CN107977609A (en) * | 2017-11-20 | 2018-05-01 | 华南理工大学 | A kind of finger vein identity verification method based on CNN |
| CN108009528A (en) * | 2017-12-26 | 2018-05-08 | 广州广电运通金融电子股份有限公司 | Face authentication method, device, computer equipment and storage medium based on Triplet Loss |
| CN108009520A (en) * | 2017-12-21 | 2018-05-08 | 东南大学 | A kind of finger vein identification method and system based on convolution variation self-encoding encoder neutral net |
| WO2018137358A1 (en) * | 2017-01-24 | 2018-08-02 | 北京大学 | Deep metric learning-based accurate target retrieval method |
| CN108960289A (en) * | 2018-06-08 | 2018-12-07 | 清华大学 | Medical imaging sorter and method |
| CN109376602A (en) * | 2018-09-21 | 2019-02-22 | 厦门中控智慧信息技术有限公司 | A kind of finger vein identification method, device and terminal device |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI599964B (en) * | 2010-09-03 | 2017-09-21 | 國立台灣科技大學 | Finger vein recognition system and method |
| CN110147732A (en) * | 2019-04-16 | 2019-08-20 | 平安科技（深圳）有限公司 | Refer to vein identification method, device, computer equipment and storage medium |
-
2019
- 2019-04-16 CN CN201910304095.9A patent/CN110147732A/en active Pending
- 2019-11-08 WO PCT/CN2019/116472 patent/WO2020211339A1/en active Application Filing
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018137358A1 (en) * | 2017-01-24 | 2018-08-02 | 北京大学 | Deep metric learning-based accurate target retrieval method |
| CN106991368A (en) * | 2017-02-20 | 2017-07-28 | 北京大学 | A kind of finger vein checking personal identification method based on depth convolutional neural networks |
| CN107392114A (en) * | 2017-06-29 | 2017-11-24 | 广州智慧城市发展研究院 | A kind of finger vein identification method and system based on neural network model |
| CN107480785A (en) * | 2017-07-04 | 2017-12-15 | 北京小米移动软件有限公司 | The training method and device of convolutional neural networks |
| CN107967442A (en) * | 2017-09-30 | 2018-04-27 | 广州智慧城市发展研究院 | A kind of finger vein identification method and system based on unsupervised learning and deep layer network |
| CN107977609A (en) * | 2017-11-20 | 2018-05-01 | 华南理工大学 | A kind of finger vein identity verification method based on CNN |
| CN108009520A (en) * | 2017-12-21 | 2018-05-08 | 东南大学 | A kind of finger vein identification method and system based on convolution variation self-encoding encoder neutral net |
| CN108009528A (en) * | 2017-12-26 | 2018-05-08 | 广州广电运通金融电子股份有限公司 | Face authentication method, device, computer equipment and storage medium based on Triplet Loss |
| CN108960289A (en) * | 2018-06-08 | 2018-12-07 | 清华大学 | Medical imaging sorter and method |
| CN109376602A (en) * | 2018-09-21 | 2019-02-22 | 厦门中控智慧信息技术有限公司 | A kind of finger vein identification method, device and terminal device |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020211339A1 (en) * | 2019-04-16 | 2020-10-22 | 平安科技（深圳）有限公司 | Finger vein recognition method and apparatus, and computer device and storage medium |
| CN110516745A (en) * | 2019-08-28 | 2019-11-29 | 北京达佳互联信息技术有限公司 | Training method, device and the electronic equipment of image recognition model |
| CN110516745B (en) * | 2019-08-28 | 2022-05-24 | 北京达佳互联信息技术有限公司 | Training method and device of image recognition model and electronic equipment |
| CN111008550A (en) * | 2019-09-06 | 2020-04-14 | 上海芯灵科技有限公司 | Identification method for finger vein authentication identity based on Multiple loss function |
| CN111079785A (en) * | 2019-11-11 | 2020-04-28 | 深圳云天励飞技术有限公司 | Image identification method and device and terminal equipment |
| CN111166070A (en) * | 2019-12-17 | 2020-05-19 | 五邑大学 | Medical storage cabinet based on finger vein authentication and management method thereof |
| CN111242951A (en) * | 2020-01-08 | 2020-06-05 | 上海眼控科技股份有限公司 | Vehicle detection method, device, computer equipment and storage medium |
| CN111767940A (en) * | 2020-05-14 | 2020-10-13 | 北京迈格威科技有限公司 | Target object identification method, device, equipment and storage medium |
| CN112200156A (en) * | 2020-11-30 | 2021-01-08 | 四川圣点世纪科技有限公司 | Vein recognition model training method and device based on clustering assistance |
| CN112200156B (en) * | 2020-11-30 | 2021-04-30 | 四川圣点世纪科技有限公司 | Vein recognition model training method and device based on clustering assistance |
| CN112200159A (en) * | 2020-12-01 | 2021-01-08 | 四川圣点世纪科技有限公司 | Non-contact palm vein identification method based on improved residual error network |
| WO2023093838A1 (en) * | 2021-11-25 | 2023-06-01 | 北京字跳网络技术有限公司 | Super-resolution image processing method and apparatus, and device and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2020211339A1 (en) | 2020-10-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110147732A (en) | Refer to vein identification method, device, computer equipment and storage medium | |
| Gao et al. | Multiscale analysis of complex time series: integration of chaos and random fractal theory, and beyond | |
| Han et al. | Image annotation by input–output structural grouping sparsity | |
| US20190102655A1 (en) | Training data acquisition method and device, server and storage medium | |
| CN107168992A (en) | Article sorting technique and device, equipment and computer-readable recording medium based on artificial intelligence | |
| CN110059465A (en) | Auth method, confrontation generate training method, device and the equipment of network | |
| CN108959474B (en) | Entity relation extraction method | |
| Zhu et al. | Convolutional ordinal regression forest for image ordinal estimation | |
| CN109685104B (en) | Determination method and device for recognition model | |
| CN112990302B (en) | Model training method and device based on text generated image and image generation method | |
| CN111401219B (en) | Palm key point detection method and device | |
| Dering et al. | Generative adversarial networks for increasing the veracity of big data | |
| CN114896067B (en) | Automatic generation method and device of task request information, computer equipment and medium | |
| CN114239083B (en) | Efficient state register identification method based on graph neural network | |
| CN114093022A (en) | Activity detection device, activity detection system, and activity detection method | |
| CN109948680A (en) | The classification method and system of medical record data | |
| CN108492301A (en) | A kind of Scene Segmentation, terminal and storage medium | |
| CN112949469A (en) | Image recognition method, system and equipment for face tampered image characteristic distribution | |
| CN109271546A (en) | The foundation of image retrieval Feature Selection Model, Database and search method | |
| CN116311539B (en) | Sleep motion capturing method, device, equipment and storage medium based on millimeter waves | |
| Zhang et al. | A Learnable Discrete-Prior Fusion Autoencoder with Contrastive Learning for Tabular Data Synthesis | |
| CN112131587B (en) | Intelligent contract pseudo-random number security inspection method, system, medium and device | |
| CN111914772B (en) | Age identification method, age identification model training method and device | |
| CN114298299A (en) | Model training method, device, equipment and storage medium based on course learning | |
| Dong et al. | Scene-oriented hierarchical classification of blurry and noisy images |
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 |