CN110705352A - Fingerprint image detection method based on deep learning - Google Patents
Fingerprint image detection method based on deep learningDownload PDFInfo
- Publication number
- CN110705352A CN110705352ACN201910808923.2ACN201910808923ACN110705352ACN 110705352 ACN110705352 ACN 110705352ACN 201910808923 ACN201910808923 ACN 201910808923ACN 110705352 ACN110705352 ACN 110705352A
- Authority
- CN
- China
- Prior art keywords
- fingerprint
- living body
- fingerprint image
- score
- probability
- 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
Images
Classifications
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; 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 OR CALCULATING; 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
- G06V40/1382—Detecting the live character of the finger, i.e. distinguishing from a fake or cadaver finger
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06V40/1382—Detecting the live character of the finger, i.e. distinguishing from a fake or cadaver finger
- G06V40/1388—Detecting the live character of the finger, i.e. distinguishing from a fake or cadaver finger using image processing
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)
- Computer Vision & Pattern Recognition (AREA)
- Collating Specific Patterns (AREA)
- Image Input (AREA)
Abstract
Description
Technical Field
The invention relates to the field of digital image processing, in particular to a fingerprint image detection method based on deep learning.
Background
The acquisition of the fingerprint image relies on a fingerprint sensor. Fingerprint sensors are roughly classified into optical fingerprint sensors, semiconductor capacitance fingerprint sensors, semiconductor pressure sensors, thermal sensors, and the like according to the principle and technology of fingerprint imaging. The fingerprint sensor realizes the description of the fingerprint information of a human body, and the actual imaging quality difference of different sensors is larger due to the difference of the principles. Differences in imaging quality from sensor to sensor may lead to differences in performance under the same algorithm.
Deep learning, particularly convolutional neural networks, which are widely used in the field of computer vision, is an advanced machine learning means and method today. The method is mainly characterized in that a target mode is automatically learned and fitted from a large amount of data according to a network structure and a corresponding operation rule. The establishment of the deep learning model requires a large amount of data, and for fingerprint living body detection, a large amount of living body fingerprint images and non-living body fingerprint images need to be collected to construct a training set, a verification set, a test set and other image sets. The main operations of the convolutional neural network comprise convolution, separable convolution, batch standardization, pooling, full connection and the like. The convolution fitting effect is good, but the operation amount is maximum. The separable convolution parameters and the operation quantity are much less than those of convolution, and the parameters required by full-connection operation are the largest.
The fingerprint is an important biological characteristic, is widely applied to the fields of attendance checking, security protection, smart home and the like, has a large number of phenomena of checking, card punching and the like of false fingerprints such as fingerprint films and the like, and is easy to deceive equipment. Therefore, the existing fingerprint detection has been a detection as to whether it is a living body.
For fingerprint in-vivo detection, the existing in-vivo detection implementation methods are basically divided into two types, namely a pure software method and a method requiring additional hardware cooperation. The method needing extra hardware cooperation needs to additionally arrange hardware outside the existing fingerprint sensor to detect the biological characteristics of the user, mainly signals such as skin color, blood and the like, so as to obtain the skin color and heart rate characteristics of the user, and the additionally obtained biological characteristics are combined with a fingerprint image to carry out fingerprint living body detection. The pure software method is mainly realized based on fingerprint deformation detection, finger sweat pore detection and the like. The equipment detection is cumbersome, the equipment needs extra hardware to detect whether the living body exists, the equipment cost rises, the size is increased, and the detection efficiency is low.
Disclosure of Invention
The invention aims to provide a fingerprint image detection method based on deep learning, which aims at overcoming the defects of the prior art, utilizes a convolutional neural network to automatically extract fingerprint image characteristics, effectively realizes the distinguishing of biological living fingerprints and artificial false fingerprints, and prevents the attack of the false fingerprints.
In order to achieve the purpose, the invention adopts the following technical scheme:
the fingerprint image detection method based on deep learning comprises the following steps:
s1, collecting a fingerprint image by the fingerprint sensor, and preprocessing the fingerprint image;
s2, intercepting local parts of the fingerprint images preprocessed in the S1 as input of a living body detection model, setting the fingerprint images as i, and intercepting the number of the images as n;
s3, calculating the living body probability of each input fingerprint image, wherein the living body probability is Pi, and the living body probability is obtained through a convolutional neural network model and a classification function;
and S4, setting the score of the living body probability judgment as score ═ score/n, judging that the fingerprint belongs to the living body if the score is greater than a preset value T, and judging that the fingerprint is not the living body if the score is less than the preset value T.
Further, the fingerprint image acquired in S1 has poor quality, and needs to be preprocessed for live body detection. The preprocessing mode comprises operations of filtering and denoising, image normalization, image enhancement and the like, and the filtering and denoising can reduce image noise caused by various external noises during imaging. Image enhancement is often performed by operations such as thinning of image texture.
Furthermore, the convolution part of the convolution neural network model comprises a module A and a module B, wherein the input of the module A is an output characteristic diagram calculated in the previous layer, and then a convolution layer, a batch normalization layer and an activation function layer are sequentially calculated; the input of the module B is also the output characteristic diagram of the previous layer, and then the separable convolution layer, the batch normalization layer and the activation function layer are calculated in sequence.
The parameters such as the size and number of convolution templates in the module a, and the number of channels of each parameter for batch normalization are not constants, and vary with the change in the network configuration. Parameters such as the size and the number of channels of the separable convolution template in the module B, and the number of channels of each parameter of batch normalization are not constants and vary with the variation of the network configuration.
Further, the living body probability is calculated by connecting a plurality of modules A and B in sequence, wherein the value of the activation function of the last module B is a 2-dimensional vector, and the 2-dimensional vector is used as the input of the classification function.
Further, the classification function is a normalized exponential function.
Further, the activation function layer is a Relu6 function, and the Relu6 function is a convolutional neural network activation function, which is defined as follows:
wherein x is: the convolution layer calculation result is specifically a result of performing convolution operation on a portion of an image or a feature map (feature map) overlapping with a convolution template, and x is a scalar.
By adopting the technical scheme of the invention, the invention has the beneficial effects that: compared with the prior art, the method utilizes the convolutional neural network to automatically extract the fingerprint image characteristics, effectively realizes the distinguishing of the living biological fingerprint and the artificial fake fingerprint, and prevents the attack of the fake fingerprint. According to the method, additional detection hardware is not needed, and whether the provider of the fingerprint picture is a living organism or not is judged only through the fingerprint picture acquired by the fingerprint sensor. The invention automatically extracts the fingerprint living characteristics by using a deep learning mode without manually constructing a characteristic descriptor. The convolutional neural network model is constructed in a modular assembly mode, is flexible and changeable, gives consideration to the calculation parameter quantity and the characteristic description capacity of different modules, and can be conveniently constructed into a proper living body detection model on different calculation force platforms.
Drawings
FIG. 1 is a flowchart of a fingerprint image detection method based on deep learning according to the present invention;
FIG. 2 is a schematic diagram of capturing a plurality of images in a deep learning-based fingerprint image detection method provided by the invention;
FIG. 3 is a schematic diagram of a module A in a deep learning-based fingerprint image detection method according to the present invention;
FIG. 4 is a schematic diagram of a module B in a deep learning-based fingerprint image detection method according to the present invention;
FIG. 5 is a flowchart of a live body detection probability in a fingerprint image detection method based on deep learning according to the present invention;
FIG. 6 is a flowchart of an embodiment of a deep learning-based fingerprint image detection method according to the present invention
Fig. 7 is a schematic diagram of an embodiment of a flow chart of a live detection probability in a fingerprint image detection method based on deep learning according to the present invention.
Detailed Description
Specific embodiments of the present invention will be further described with reference to the accompanying drawings.
The fingerprint is an important biological characteristic and is widely applied to the fields of attendance checking, security protection, smart home and the like. In reality, an attack mode aiming at the biological characteristics of the fingerprint still appears. The false fingerprint attack is characterized in that a biological fingerprint film and a fingerprint sleeve are manufactured by using specific glue or other means, so that a non-fingerprint real owner can deceive fingerprint comparison equipment and pass identity verification. The fingerprint sensor in the actual product is very small, and only partial fingerprints of the finger can be collected at one time, so the identification degree of the fingerprint sensor is lower than that of a complete fingerprint. Meanwhile, the fingerprint features contain many very common features. Thus, it is not coincidental that the fake fingerprint can be matched with the real fingerprint.
The convolutional neural network is an important theoretical branch of deep learning in the field of image processing, and one of the main characteristics of the convolutional neural network is that the features in an image can be automatically extracted and described. The traditional fingerprint feature extraction algorithm is still a way of artificially describing and constructing descriptors in a structured way, and the way is difficult to completely describe the features of the fingerprint and the differences among different fingerprints. The invention utilizes the convolution neural network to automatically extract the fingerprint image characteristics, effectively realizes the distinguishing of the biological living fingerprint and the artificial false fingerprint, and defends the attack of the false fingerprint.
As shown in fig. 1, the fingerprint image detection method based on deep learning includes the following steps:
s1, collecting a fingerprint image by the fingerprint sensor, and preprocessing the fingerprint image;
s2, intercepting local parts of the fingerprint images preprocessed in the S1 as input of a living body detection model, setting the fingerprint images as i, and intercepting the number of the images as n;
s3, calculating the living body probability of each input fingerprint image, wherein the living body probability is Pi, and the living body probability is obtained through a convolutional neural network model and a classification function;
and S4, setting the score of the living body probability judgment as score ═ score/n, judging that the fingerprint belongs to the living body if the score is greater than a preset value T, and judging that the fingerprint is not the living body if the score is less than the preset value T.
As shown in fig. 3 and 4, the convolution part of the convolutional neural network model includes a module a and a module B, the input of the module a is the output characteristic diagram calculated in the previous layer, and then the convolutional layer, the batch normalization layer and the activation function layer are calculated in sequence; the input of the module B is also the output characteristic diagram of the previous layer, and then the separable convolution layer, the batch normalization layer and the activation function layer are calculated in sequence.
The parameters such as the size and number of convolution templates in the module a, and the number of channels of each parameter for batch normalization are not constants, and vary with the change in the network configuration. Parameters such as the size and the number of channels of the separable convolution template in the module B, and the number of channels of each parameter of batch normalization are not constants and vary with the variation of the network configuration.
As shown in fig. 5, the living body probability is calculated by sequentially connecting a plurality of modules a and B, the value of the activation function of the last module B is a 2-dimensional vector, and the probability that the pixel region belongs to a living body fingerprint is calculated by using the 2-dimensional vector as the input of the classification function.
The classification function is a normalized exponential function. The activation function layer is a Relu6 function, and the Relu6 function is a convolutional neural network activation function, which is defined as follows:
wherein x is the calculation result of the convolution layer, specifically the result of convolution operation on the part of the image or feature map (feature map) overlapped with the convolution template, and x is a scalar.
The first embodiment,
The invention is further illustrated by the following specific example, as shown in figure 6,
s1, the image captured by the fingerprint sensor has a relatively large scale, such as 160X160 pixel size. Preprocessing the fingerprint image; the quality of the acquired fingerprint image is poor, and the fingerprint image can be used for in-vivo detection only by preprocessing. The preprocessing mode comprises operations of filtering and denoising, image normalization, image enhancement and the like, and the filtering and denoising can reduce image noise caused by various external noises during imaging. Image enhancement is often performed by operations such as thinning of image texture. The pretreatment method is a conventional and common method.
S2, as shown in fig. 2, the parts of the fingerprint images preprocessed in S1 are extracted as the input of the biometric model, the fingerprint image is set to i, and the number of the extracted images is set to 5. On the premise of ensuring the performance of the algorithm, in order to reduce the amount of computation and time consumption, a partial image needs to be cut from a larger original image as the input of a model, and the partial image is a plurality of original fingerprint image areas with the same size. For example: on a picture of the original size 160X160 pixels, 5 32X32 pixel sub-pictures are taken at the center, upper left, lower left, upper right, and lower right at a time. At this time, each 32 × 32 area subgraph is used as an input image of the model, and the probability that the subregion is a living organism is calculated sequentially.
S3, calculating the living body probability of each input fingerprint image, wherein the living body probability is Pi, and the living body probability is obtained through a convolutional neural network model and a classification function; the convolution part of the convolution neural network model comprises a module A and a module B, wherein the input of the module A is an output characteristic diagram calculated in the previous layer, and then a convolution layer, a batch normalization layer and an activation function layer are sequentially calculated; the input of the module B is also the output characteristic diagram of the previous layer, and then the separable convolution layer, the batch normalization layer and the activation function layer are calculated in sequence.
As shown in fig. 7, the biometric model inputs a fingerprint image of 32X32 pixels, and the model is formed by sequentially connecting 1 a module and 12B modules. The final 12 th B-module activation function has the value of a 2-dimensional vector. Taking this 2-dimensional vector as input to a classification function (most often the Softmax function is used), the probability that the 32X32 pixel region belongs to a live fingerprint is calculated. The probability is calculated using the Softmax method. The Softmax method is represented by the following equation:
wherein K represents that the results output by the classification model are totally divided into several classes, and z represents the score of the class input into the Softmax function. The assignment of the class scores to terms based on the natural constant e is to increase the degree of model non-linearity. The output of the layers of convolution, namely the input of the Softmax function, is a vector, the elements in the vector represent the score of the class, the score is a scalar quantity, and is the result of abstraction of the model layer by layer and does not directly represent the information of the color of the living body and the like. There are only two possibilities of authenticity in the fingerprint liveness detection problem, so K here is 2.
And then the live fingerprint probabilities for the other 4 32X32 pixel regions are calculated. And circulating the steps until the living body probability of each subgraph is calculated. I.e. the steps in the figure, i ═ i +1, score ═ score + Pi, when i is greater than 5, i.e. all calculations are completed.
And S4, taking the average value of the 5 region probabilities as the probability that the whole fingerprint belongs to the living body, setting the living body probability judgment score as score ═ score/5, judging that the fingerprint belongs to the living body if the score is greater than a preset value T, and judging that the fingerprint is not the living body if the score is less than the preset value T.
The system presets a probability threshold value for passing the living body detection, compares the probability with the probability obtained by the calculation of the convolutional neural network model, and when the model calculation probability is greater than the threshold value, the fingerprint is considered to be collected from the living body, so that the user identity comparison link can be carried out, otherwise, the fingerprint comparison is rejected.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (5)
1. The fingerprint image detection method based on deep learning is characterized by comprising the following steps of:
s1, collecting a fingerprint image by the fingerprint sensor, and preprocessing the fingerprint image;
s2, intercepting local parts of the fingerprint images preprocessed in the S1 as input of a living body detection model, setting the fingerprint images as i, and intercepting the number of the images as n;
s3, calculating the living body probability of each input fingerprint image, wherein the living body probability is Pi, and the living body probability is obtained through a convolutional neural network model and a classification function;
and S4, setting the score of the living body probability judgment as score ═ score/n, judging that the fingerprint belongs to the living body if the score is greater than a preset value T, and judging that the fingerprint is not the living body if the score is less than the preset value T.
2. The deep learning-based fingerprint image detection method according to claim 1, wherein the convolution part of the convolution neural network model comprises a module A and a module B, the input of the module A is the output feature map calculated by the previous layer, and then the convolution layer, the batch normalization layer and the activation function layer are sequentially calculated; the input of the module B is also the output characteristic diagram of the previous layer, and then the separable convolution layer, the batch normalization layer and the activation function layer are calculated in sequence.
3. The method for detecting fingerprint images based on deep learning of claim 2, wherein the live body probability is calculated by sequentially connecting a plurality of modules a and B, the value of the activation function of the last module B is a 2-dimensional vector, and the probability that the pixel region belongs to the live body fingerprint is calculated by using the 2-dimensional vector as the input of the classification function.
4. The deep learning-based fingerprint image detection method of claim 1, wherein the classification function is a normalized exponential function.
5. The deep learning-based fingerprint image detection method according to claim 2, wherein the activation function layer is a Relu6 function, and the Relu6 function is a convolutional neural network activation function, which is defined as follows:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910808923.2ACN110705352A (en) | 2019-08-29 | 2019-08-29 | Fingerprint image detection method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910808923.2ACN110705352A (en) | 2019-08-29 | 2019-08-29 | Fingerprint image detection method based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110705352Atrue CN110705352A (en) | 2020-01-17 |
Family
ID=69194171
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910808923.2APendingCN110705352A (en) | 2019-08-29 | 2019-08-29 | Fingerprint image detection method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110705352A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112215194A (en)* | 2020-10-23 | 2021-01-12 | 敦泰电子（深圳）有限公司 | Optical fingerprint identification module, identification method thereof and display device |
| CN117152800A (en)* | 2023-09-21 | 2023-12-01 | 北京大学 | Fingerprint presentation attack detection method and device |
| CN117253295A (en)* | 2023-09-28 | 2023-12-19 | 东南大学 | Fingerprint living body detection method and system based on depth robust forest |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106096535A (en)* | 2016-06-07 | 2016-11-09 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | A kind of face verification method based on bilinearity associating CNN |
| CN106683104A (en)* | 2017-01-06 | 2017-05-17 | 西北工业大学 | Prostate magnetic resonance image segmentation method based on integrated depth convolution neural network |
| CN108564040A (en)* | 2018-04-17 | 2018-09-21 | 杭州景联文科技有限公司 | A kind of fingerprint activity test method based on depth convolution feature |
| CN108875338A (en)* | 2018-05-04 | 2018-11-23 | 北京旷视科技有限公司 | unlocking method, device and system and storage medium |
| CN109508678A (en)* | 2018-11-16 | 2019-03-22 | 广州市百果园信息技术有限公司 | Training method, the detection method and device of face key point of Face datection model |
| CN109829446A (en)* | 2019-03-06 | 2019-05-31 | 百度在线网络技术（北京）有限公司 | Eye fundus image recognition methods, device, electronic equipment and storage medium |
- 2019
- 2019-08-29CNCN201910808923.2Apatent/CN110705352A/enactivePending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106096535A (en)* | 2016-06-07 | 2016-11-09 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | A kind of face verification method based on bilinearity associating CNN |
| CN106683104A (en)* | 2017-01-06 | 2017-05-17 | 西北工业大学 | Prostate magnetic resonance image segmentation method based on integrated depth convolution neural network |
| CN108564040A (en)* | 2018-04-17 | 2018-09-21 | 杭州景联文科技有限公司 | A kind of fingerprint activity test method based on depth convolution feature |
| CN108875338A (en)* | 2018-05-04 | 2018-11-23 | 北京旷视科技有限公司 | unlocking method, device and system and storage medium |
| CN109508678A (en)* | 2018-11-16 | 2019-03-22 | 广州市百果园信息技术有限公司 | Training method, the detection method and device of face key point of Face datection model |
| CN109829446A (en)* | 2019-03-06 | 2019-05-31 | 百度在线网络技术（北京）有限公司 | Eye fundus image recognition methods, device, electronic equipment and storage medium |
Non-Patent Citations (4)
| Title |
|---|
| ANDREW G.HOWARD,ET AL.: ""MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications"", 《ARXIV》* |
| NINE-DAYS: ""Relu6作用"", 《CSDN》* |
| 大写的ZDQ: ""MobileNet_v1详详讲"", 《CSDN》* |
| 高志强等: "《深度学习：从入门到实战》", 30 June 2018, 中国铁道出版社* |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112215194A (en)* | 2020-10-23 | 2021-01-12 | 敦泰电子（深圳）有限公司 | Optical fingerprint identification module, identification method thereof and display device |
| WO2022083048A1 (en)* | 2020-10-23 | 2022-04-28 | 敦泰电子（深圳）有限公司 | Optical fingerprint recognition module, recognition method thereof, and electronic apparatus |
| CN117152800A (en)* | 2023-09-21 | 2023-12-01 | 北京大学 | Fingerprint presentation attack detection method and device |
| CN117152800B (en)* | 2023-09-21 | 2025-08-19 | 北京大学 | Fingerprint presentation attack detection method and device |
| CN117253295A (en)* | 2023-09-28 | 2023-12-19 | 东南大学 | Fingerprint living body detection method and system based on depth robust forest |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11830230B2 (en) | Living body detection method based on facial recognition, and electronic device and storage medium | |
| CN112801057B (en) | Image processing method, image processing device, computer equipment and storage medium | |
| CN112464690B (en) | Living body recognition method, device, electronic device and readable storage medium | |
| CN112487922B (en) | Multi-mode human face living body detection method and system | |
| CN111462379A (en) | Access control management method, system and medium containing palm vein and face recognition | |
| CN106529414A (en) | Method for realizing result authentication through image comparison | |
| WO2022127112A1 (en) | Cross-modal face recognition method, apparatus and device, and storage medium | |
| CN107169479A (en) | Intelligent mobile equipment sensitive data means of defence based on fingerprint authentication | |
| CN110443146A (en) | Auth method, device, equipment and readable medium based on bio-identification | |
| CN110705352A (en) | Fingerprint image detection method based on deep learning | |
| CN111814682A (en) | Face living body detection method and device | |
| CN111178130A (en) | Face recognition method, system and readable storage medium based on deep learning | |
| CN112381042A (en) | Method for extracting palm vein features from palm vein image and palm vein identification method | |
| CN112308035A (en) | Image detection method, image detection device, computer equipment and storage medium | |
| CN113420582B (en) | Anti-counterfeiting detection method and system for palm vein recognition | |
| CN105631285A (en) | Biological feature identity recognition method and apparatus | |
| Lazimul et al. | Fingerprint liveness detection using convolutional neural network and fingerprint image enhancement | |
| CN118658213B (en) | A palm print and palm vein recognition method based on ShuffleNetV2 neural network | |
| CN119206845A (en) | Iris recognition method and device, electronic device, and storage medium | |
| CN113343927A (en) | Intelligent face recognition method and system suitable for facial paralysis patient | |
| CN113506274A (en) | Detection system for human cognitive condition based on visual saliency difference map | |
| CN109359616B (en) | Pseudo-concatenation small-size fingerprint identification algorithm based on SIFT | |
| CN112906613B (en) | Method and device for collecting identity information | |
| CN113408416B (en) | Fingerprint frequency estimation method and device and fingerprint information extraction method and device | |
| CN110738174B (en) | Finger vein recognition method, device, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | Application publication date:20200117 | |
| RJ01 | Rejection of invention patent application after publication |