CN111814570B - Face recognition method, system and storage medium based on dynamic threshold - Google Patents
Face recognition method, system and storage medium based on dynamic threshold Download PDFInfo
- Publication number
- CN111814570B CN111814570B CN202010533170.1A CN202010533170A CN111814570B CN 111814570 B CN111814570 B CN 111814570B CN 202010533170 A CN202010533170 A CN 202010533170A CN 111814570 B CN111814570 B CN 111814570B
- Authority
- CN
- China
- Prior art keywords
- feature vector
- face
- feature
- updating
- face recognition
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000003860 storage Methods 0.000 title claims abstract description 13
- 239000013598 vector Substances 0.000 claims abstract description 147
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 15
- 230000007246 mechanism Effects 0.000 claims description 12
- 238000013459 approach Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 6
- 238000004088 simulation Methods 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 238000013508 migration Methods 0.000 abstract description 3
- 230000005012 migration Effects 0.000 abstract description 3
- 238000012545 processing Methods 0.000 description 10
- 230000008569 process Effects 0.000 description 7
- 238000013136 deep learning model Methods 0.000 description 6
- 230000009471 action Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000001815 facial effect Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013441 quality evaluation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
-
- 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
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention discloses a face recognition method, a face recognition system and a storage medium based on a dynamic threshold, wherein the method comprises the following steps: acquiring a face image to be identified, and calculating a feature vector of the face image to be identified; the feature vector of the face is close to the feature center direction by using an updating algorithm; and using the updated face recognition base feature vector for face recognition. In the aspect of face recognition, the invention updates the feature vector of the face recognition base by using the face image acquired by the equipment end on the premise of not changing the face recognition model, so that the feature vector of the face is iterated gradually towards the direction of the feature center, thereby realizing style migration of the field environment, improving the face recognition rate and reducing the false recognition rate.
Description
Technical Field
The invention relates to the technical field of face recognition, in particular to a face recognition method, a face recognition system and a storage medium based on a dynamic threshold.
Background
At present, the face recognition technology is widely applied to the production and life of people. In a general workflow, personnel information is recorded in the background, including a face database, in order to ensure the recognition effect and safety, photos of the database need to be subjected to strict quality evaluation and living body detection, and the photos are not homologous to images acquired by a terminal camera, which may generate data distribution with large difference, thereby causing the reduction of recognition rate. And the terminal equipment is well controlled due to the fact that the environment is not good, for example, the equipment can be deployed in a dark and backlit environment, and quality judgment of a face can not be controlled too severely. These widely differing images may result in a poor face experience during recognition.
The prior art is therefore still in need of further development.
Disclosure of Invention
Aiming at the technical problems, the embodiment of the invention provides a face recognition method, a face recognition system and a storage medium based on a dynamic threshold, which can solve the related technical problems in the prior art.
The embodiment of the invention provides a face recognition method based on a dynamic threshold, which comprises the following steps:
Acquiring a face image to be identified, and calculating a feature vector of the face image to be identified;
Updating the feature vector of the face identification base by using an updating algorithm and the feature vector of the face image to be identified, so that the feature vector of the face approaches to the feature center direction;
and using the updated face recognition base feature vector for face recognition.
Optionally, the updating the face recognition base feature vector by using an updating algorithm and the feature vector of the face image to be recognized includes:
Comparing the feature vector of the face image to be recognized with the feature vector of a face database;
Triggering an updating mechanism if the comparison score of the feature vector of the face image to be identified and the feature vector of the face database is larger than a first preset threshold value and the quality of the face image meets a preset requirement;
and taking the feature vector of the face image to be identified as update data.
Optionally, the updating the face recognition base feature vector by using an updating algorithm and the feature vector of the face image to be recognized further includes:
updating the face recognition bottom library feature vector by the following formula:
; wherein fn represents the updated feature vector of the background library, f (n-1) represents the feature vector of the background library before updating, fq represents the feature vector acquired by the equipment in the complex environment, and fg represents the feature vector of the face of the background library which is originally input; /(I) Is a coefficient,/>。
Optionally, the face recognition method based on the dynamic threshold value further comprises the following steps:
Setting different thresholds based on applicable frequency usage frequencies of different device IDs;
The updated parameters of the threshold values are obtained from the simulation dataset.
Optionally, the face recognition method based on the dynamic threshold value further comprises the following steps:
different update strategies are set for face comparison thresholds of different device IDs according to the following formula:
；
wherein S_k is the comparison score passing each comparison, the value range is 0-1, T_n is the updated threshold, and T_ (n-1) is the threshold before updating. a is an empirical value, α, β are constants, and α+β=1.
The embodiment of the invention also provides a face recognition system based on the dynamic threshold value, which comprises:
The feature acquisition module is used for acquiring a face image to be identified and calculating a feature vector of the face image to be identified;
The feature updating module is used for updating the feature vector of the face recognition base by using an updating algorithm and the feature vector of the face image to be recognized so that the feature vector of the face approaches to the feature center direction;
and the identification module is used for identifying the face by using the updated face identification base feature vector.
Optionally, the feature acquisition module includes:
Comparing the feature vector of the face image to be recognized with the feature vector of a face database;
Triggering an updating mechanism if the comparison score of the feature vector of the face image to be identified and the feature vector of the face database is larger than a first preset threshold value and the quality of the face image meets a preset requirement;
And taking the feature vector of the face image to be recognized as the updating data of the feature vector of the face recognition base.
Optionally, the feature updating module includes:
updating the face recognition bottom library feature vector by the following formula:
; wherein, middle/> Representing the updated bottom library feature vector,Representing the bottom library feature vector before update,/>Acquiring feature vectors acquired in a complex environment for a device,/>The face feature vector is the original input face feature vector of the base; /(I)Is a coefficient,/>。
Optionally, the feature updating module includes:
a threshold updating module:
different update strategies are set for face comparison thresholds of different device IDs according to the following formula:
；
Wherein the method comprises the steps of For each comparison pass, the comparison score is in the value range of/>，/>For updated threshold,/>Is the threshold before updating. /(I)Is an empirical value,/>、/>Is constant and/>。
The embodiment of the invention also provides a computer readable storage medium, wherein a computer program is stored in the computer readable storage medium, and the computer program realizes the face recognition method steps based on the dynamic threshold when being executed by a processor.
In the aspect of face recognition, the invention updates the feature vector of the face recognition base by using the face image acquired by the equipment end on the premise of not changing the face recognition model, so that the feature vector of the face is iterated gradually towards the direction of the feature center, thereby realizing style migration of the field environment, improving the face recognition rate and reducing the false recognition rate.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following description will simply explain the drawings that are required to be used in the description of the embodiments.
Fig. 1 is a flowchart of a face recognition method based on a dynamic threshold in an embodiment of the present invention.
Fig. 2 is a schematic diagram of partial vectors of face feature vectors according to an embodiment of the present invention.
Fig. 3 is a flowchart of a face recognition method based on a dynamic threshold according to another embodiment of the present invention.
Fig. 4 is a flowchart of a face recognition method based on a dynamic threshold according to another embodiment of the present invention.
Fig. 5 is a block diagram of a face recognition system based on dynamic threshold according to another embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to fall within the scope of the invention.
Embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
As shown in fig. 1, the face recognition method based on the dynamic threshold provided by the invention comprises the following steps:
Step S100: and acquiring a face image to be identified, and calculating the feature vector of the face image to be identified.
The training of the face recognition model is to perform feature dimension processing on the image, and input the face image to be recognized into the deep learning model; and receiving the face characteristic data output by the deep learning model. In the deep learning model, after receiving a face image to be recognized, an input layer carries out data processing on the face image to be recognized so as to obtain high-dimensional characteristic data; performing data dimension reduction processing on the high-dimensional characteristic data to obtain low-dimensional characteristic data; the feature combination layer pre-processes the low-dimensional feature data and combines the low-dimensional feature data to obtain combined feature data; and carrying out feature fusion on the combined feature data to obtain the face feature data. However, the present invention is not limited to this, and there may be a plurality of processing methods for image processing and feature data.
Similarly, when the face image to be recognized is obtained, the recognition module is used for processing the image, the image is firstly processed into high-dimensional feature data in the deep learning model, and then feature processing, such as dimension reduction processing, is performed, so that the deep learning model can conveniently perform a series of operations and processes, and available face features are extracted. And face feature vectors which can also be understood as being suitable for the final recognition task of the deep learning model.
Step S200: and updating the feature vector of the face identification base by using an updating algorithm and the feature vector of the face image to be identified, so that the feature vector of the face approaches to the feature center direction.
Face recognition algorithms rely on image quality, and recognition performance is different in different scenes, and experience is generally worse than normal in environments with abnormal light such as backlight and darkness. The existing solution is a method for acquiring scene pictures for retraining and replacing models, but the method is not a problem which can be solved in a short term. The technology provided by the invention mainly improves the performance of the algorithm on the premise of not replacing the model.
Specifically, a first-order low-pass filtering algorithm using updating parameters is adopted to update the face feature vectors, and class center points of different feature vectors are continuously updated through the feature vectors recorded later.
As shown in fig. 2, there are a plurality of feature vectors, where f1, f2, f3 are face feature vectors that are relatively close, wa, wb are representative classes, and the feature classes formed by the deep learning classification algorithm represent cluster centers of different IDs. If the feature vectors f1, f2, and f3 belong to the class Wa, assuming that f3 is used alone as the feature vector of the face database, the phenomenon that f3 is close to (misrecognized by) another class Wb and far from the class Wa (unrecognized) is likely to occur.
The feature vectors of the face base are updated by using a first-order low-pass filtering algorithm of the updating parameters, and f1, f2 and f3 are continuously updated through the feature vectors which are fed in, so that the feature vectors gradually approach to the clustering center, and the optimization is realized in the actual use process of reducing the distance between classes and increasing the distance between classes, so that the situations of misrecognition and incapability of being identified can be well solved.
Of course, fig. 2 is only a schematic diagram, and there are many actual facial feature vectors, which are only illustrated here for ease of understanding.
Step S300: and using the updated face recognition base feature vector for face recognition.
The face recognition module updated by the method is not changed, and only the face recognition base feature vector is updated, so that the face recognition model is more suitable for the environment where equipment is located, the recognition rate is improved compared with the original recognition rate, and the false recognition rate is reduced.
According to the technical scheme provided by the invention, the characteristics of the human face are iterated step by step towards the characteristic center direction. Finally, style migration from the bottom library to the field environment is realized, and the passing rate is greatly improved. In order to further reduce the false recognition rate, a strategy of dynamic threshold values is adopted, namely, faces of different bottom libraries have different comparison threshold values. The strategy remarkably improves the recognition rate and reduces the false recognition rate on the premise of not changing the face recognition model.
As shown in fig. 3, the present invention further provides another face recognition method based on a dynamic threshold, which includes the following steps:
Step S201, comparing the feature vector of the face image to be recognized with the feature vector of the face database.
Step S202, triggering an updating mechanism if the comparison score of the feature vector of the face image to be identified and the feature vector of the face database is larger than a first preset threshold value and the quality of the face image meets a preset requirement.
And step 203, taking the feature vector of the face image to be recognized as the update data of the feature vector of the face recognition base.
It is assumed that in the initial state, N original different face images exist in the face base, and their corresponding feature vectors fn=0=fg. Generally, these raw face libraries are subjected to rigorous quality judgment, including judgment of living body, face illumination intensity, angle offset, shielding condition, angle and the like.
For example, the face recognition device (a common face recognition card punch is not limited to this, and may be a mobile terminal such as a mobile phone) may collect a face image of a user to perform feature extraction in the use process, and compare the extracted feature fq with fg in the base, and if the score of the comparison is greater than a certain threshold and the quality of the face image is detected to a certain degree, trigger an update mechanism.
The updating of the face recognition base feature vector is realized by the following formula:
；
Wherein, the middle part Representing updated base feature vectors,/>Representing the bottom library feature vector before update,/>Acquiring feature vectors acquired in a complex environment for a device,/>The face feature vector is the original input face feature vector of the base; /(I)Is a coefficient,/>。
When the feature vectors of the collected face images of the user are compared with the feature vectors in the face database after being extracted, the score is larger than a preset threshold, the quality of the face images meets the requirements (living body, face illumination intensity, angle offset, shielding condition and the like), and an updating mechanism is triggered:
New updated base feature vectors:
When the feature vector of the collected face image of the user triggers the updating mechanism again:
New updated base feature vectors:
in summary, each update of the feature vector of the face recognition base is more prone to the use environment, and the face recognition model itself optimizes the self recognition capability by using the face image acquired by the use environment of the device. Wherein the threshold value is not fixed because the use environments of the devices are different, and the use time is changed, the brightness caused by weather is changed, and the like; as will be described in detail later.
And S204, using the updated face recognition base feature vector for face recognition.
As shown in fig. 4, the above steps can be further optimized, and specifically include the following steps:
Step 301, comparing the feature vector of the face image to be identified with the feature vector of a face database;
step S302, triggering an updating mechanism if the comparison score of the feature vector of the face image to be identified and the feature vector of the face database is larger than a first preset threshold value and the quality of the face image meets a preset requirement;
step S303, taking the feature vector of the face image to be recognized as the update data of the feature vector of the face recognition base;
step S304, setting different thresholds based on the use frequency of different device IDs.
Steps S301 to 303 are the same as the above embodiment, and only step S304 is described.
Further, different thresholds are set based on the usage frequency of different device IDs; the updated parameters of the dynamic threshold are obtained from a simulation dataset. The use of a first order low pass filter addresses the threshold update problem described above. For example, different update policies are set for face alignment thresholds for different device IDs according to the following formula:
；
Wherein the method comprises the steps of For each comparison pass, the comparison score is in the value range of/>，/>For updated threshold,/>Is the threshold before updating. /(I)Is an empirical value,/>、/>Is constant and/>。
Through the verification of the inventor, the technical scheme provided by the inventor can not meet the application scenes of all face recognition devices, but can meet the recognition requirements of most face recognition devices. To better solve the related technical problems, the inventors have made the following further technical findings:
Updating the features of the base is also a bad technical drawback, and if false recognition occurs, the updating of the feature vectors of the base will be greatly affected, so the setting of the threshold is generally severe, and if the same threshold is used for each different ID, the situation of difficult recognition occurs. And it can also be found that some people use the device frequently, the faster the frequency of the characteristic update is, and some people use the device rarely, the frequency of the update is reduced, and even no update occurs; these conditions are common.
For the ID of the frequently updated bottom library feature, the score of each alignment may be high, and a high threshold may be set to reduce false recognition. For the IDs of the features which are not updated frequently, the score of the comparison is often lower, and a lower threshold value is set; therefore, the method can be better suitable for updating the face recognition base feature vector of different people and different face images, and improves the updating mechanism of the face recognition base feature vector, so that the updating mechanism is more reasonable.
The second strategy proposed by the present invention is to take a different threshold for each ID. The threshold value is set according to a statistical rule. To prevent the occurrence of singular values, a first order low pass filter is used.
Sk is the comparison score of each comparison pass, the value range is 0-Sk-1, the total passes of m times are assumed, tn is the updated threshold, and T (n-1) is the threshold before update. a is an empirical value, which is a parameter obtained after a test on a simulated dataset, meaning that a necessary compensation value is generated in order to determine the lowest threshold value, while guaranteeing a certain false recognition rate. α, β are constants and α+β=1.
In sum, the aim of improving the recognition rate and reducing the false recognition rate can be achieved.
As shown in fig. 5, the present invention provides a face recognition system based on dynamic threshold, including:
The feature acquisition module 401 is configured to acquire a face image to be identified, and calculate a feature vector of the face image to be identified;
The feature updating module 402 is configured to use an updating algorithm and feature vectors of the face image to be identified, and face recognition base feature vectors to make the feature vectors of the face approach to a feature center direction;
the recognition module 403 uses the updated feature vector of the face recognition base for face recognition.
The feature acquisition module 401 includes:
Comparing the feature vector of the face image to be recognized with the feature vector of a face database;
Triggering an updating mechanism if the comparison score of the feature vector of the face image to be identified and the feature vector of the face database is larger than a first preset threshold value and the quality of the face image meets a preset requirement;
And taking the feature vector of the face image to be recognized as the updating data of the feature vector of the face recognition base.
The feature update module 402 includes:
face recognition is performed on the feature vectors of the bottom library through the following formula:
; wherein, middle/> Representing the updated bottom library feature vector,Representing the bottom library feature vector before update,/>Acquiring feature vectors acquired in a complex environment for a device,/>The face feature vector is the original input face feature vector of the base; /(I)Is a coefficient,/>。
The feature update module 402 includes:
different update strategies are set for face comparison thresholds of different device IDs according to the following formula:
；
Wherein the method comprises the steps of For each comparison pass, the comparison score is in the value range of/>，/>For updated threshold,/>Is the threshold before updating. /(I)Is an empirical value,/>、/>Is constant and/>。
A computer readable storage medium having stored therein a computer program which when executed by a processor implements the dynamic threshold based face recognition method steps.
The embodiment of the invention also provides a computer readable storage medium, wherein a computer program is stored in the computer readable storage medium, and the computer program realizes any data processing method based on the dynamic white box when being executed by a processor.
The storage medium mentioned in the above electronic device may include a random access Memory (Random Access Memory, RAM) or may include a Non-Volatile Memory (NVM), such as at least one magnetic disk Memory. Optionally, the memory may also be at least one memory device located remotely from the aforementioned processor.
The processor may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU), a network processor (Network Processor, NP), etc.; but may also be a digital signal processor (DIGITAL SIGNAL Processing, DSP), application SPECIFIC INTEGRATED Circuit (ASIC), field-Programmable gate array (Field-Programmable GATE ARRAY, FPGA) or other Programmable logic device, discrete gate or transistor logic device, discrete hardware components.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (3)
1. A face recognition method based on a dynamic threshold, comprising:
Acquiring a face image to be identified, and calculating a feature vector of the face image to be identified;
Updating the feature vector of the face identification base by using an updating algorithm and the feature vector of the face image to be identified, so that the feature vector of the face approaches to the feature center direction;
Using the updated face recognition base feature vector for face recognition;
The updating of the feature vector of the face recognition base by using the updating algorithm and the feature vector of the face image to be recognized comprises the following steps:
Comparing the feature vector of the face image to be recognized with the feature vector of a face database;
Triggering an updating mechanism if the comparison score of the feature vector of the face image to be identified and the feature vector of the face database is larger than a first preset threshold value and the quality of the face image meets a preset requirement;
Taking the feature vector of the face image to be recognized as the update data of the feature vector of the face recognition base;
the updating of the feature vector of the face recognition base by using the updating algorithm and the feature vector of the face image to be recognized further comprises:
updating the face recognition bottom library feature vector by the following formula: ; wherein/> Representing updated base feature vectors,/>Representing the bottom library feature vector before update,/>Acquiring feature vectors acquired in a complex environment for a device,/>The face feature vector is the original input face feature vector of the base; alpha, beta,/>As a coefficient, alpha+beta +/>=1；
Setting different thresholds based on the usage frequency of different device IDs;
the updating parameters of the threshold value are obtained by a simulation data set;
different update strategies are set for face comparison thresholds of different device IDs according to the following formula:
；
Wherein the method comprises the steps of For each comparison pass, the comparison score is in the value range of/>，/>In order to update the threshold value of the threshold value,Is a threshold before update; a is an empirical value, which is a parameter obtained after testing on a simulated dataset,/>、/>Is constant and/>+/>=1。
2. A dynamic threshold-based face recognition system, comprising:
The feature acquisition module is used for acquiring a face image to be identified and calculating a feature vector of the face image to be identified;
The feature updating module is used for updating the feature vector of the face recognition base by using an updating algorithm and the feature vector of the face image to be recognized so that the feature vector of the face approaches to the feature center direction;
the identification module is used for identifying the face by using the updated face identification base feature vector;
the feature updating module comprises:
Comparing the feature vector of the face image to be recognized with the feature vector of a face database;
Triggering an updating mechanism if the comparison score of the feature vector of the face image to be identified and the feature vector of the face database is larger than a first preset threshold value and the quality of the face image meets a preset requirement;
taking the feature vector of the face image to be recognized as the update data of the feature vector of the face recognition base;
the feature updating module further includes:
updating the face recognition bottom library feature vector by the following formula:
; wherein/> Representing updated base feature vectors,/>Representing the bottom library feature vector before update,/>Acquiring feature vectors acquired in a complex environment for a device,/>The face feature vector is the original input face feature vector of the base; alpha, beta,/>As a coefficient, alpha+beta +/>=1;
The feature updating module further comprises:
a threshold updating module:
different update strategies are set for face comparison thresholds of different device IDs according to the following formula:
；
Wherein the method comprises the steps of For each comparison pass, the comparison score is in the value range of/>，/>In order to update the threshold value of the threshold value,For the pre-update threshold, a is an empirical value, which is a parameter obtained after testing on a simulated dataset,/>、/>Is constant and/>+/>=1。
3. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored therein a computer program which, when executed by a processor, implements the method steps of claim 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010533170.1A CN111814570B (en) | 2020-06-12 | 2020-06-12 | Face recognition method, system and storage medium based on dynamic threshold |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010533170.1A CN111814570B (en) | 2020-06-12 | 2020-06-12 | Face recognition method, system and storage medium based on dynamic threshold |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111814570A CN111814570A (en) | 2020-10-23 |
| CN111814570B true CN111814570B (en) | 2024-04-30 |
Family
ID=72844939
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010533170.1A Active CN111814570B (en) | 2020-06-12 | 2020-06-12 | Face recognition method, system and storage medium based on dynamic threshold |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111814570B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112307234A (en) * | 2020-11-03 | 2021-02-02 | 厦门兆慧网络科技有限公司 | Face bottom library synthesis method, system, device and storage medium |
| CN112633113A (en) * | 2020-12-17 | 2021-04-09 | 厦门大学 | Cross-camera human face living body detection method and system |
| TWI806030B (en) * | 2021-03-31 | 2023-06-21 | 瑞昱半導體股份有限公司 | Processing circuit and processing method applied to face recognition system |
| CN113221655B (en) * | 2021-04-12 | 2022-09-30 | 重庆邮电大学 | Face spoofing detection method based on feature space constraint |
| CN113255631B (en) * | 2021-07-15 | 2021-10-15 | 浙江大华技术股份有限公司 | Similarity threshold updating method, face recognition method and related device |
Citations (28)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102004908A (en) * | 2010-11-30 | 2011-04-06 | 汉王科技股份有限公司 | Self-adapting face identification method and device |
| KR20110109695A (en) * | 2010-03-31 | 2011-10-06 | 경북대학교 산학협력단 | Face recognition system and method thereof |
| CN105938552A (en) * | 2016-06-29 | 2016-09-14 | 北京旷视科技有限公司 | Face recognition method capable of realizing base image automatic update and face recognition device |
| CN107590212A (en) * | 2017-08-29 | 2018-01-16 | 深圳英飞拓科技股份有限公司 | The Input System and method of a kind of face picture |
| CN108170750A (en) * | 2017-12-21 | 2018-06-15 | 深圳英飞拓科技股份有限公司 | A kind of face database update method, system and terminal device |
| CN108446387A (en) * | 2018-03-22 | 2018-08-24 | 百度在线网络技术（北京）有限公司 | Method and apparatus for updating face registration library |
| CN108875534A (en) * | 2018-02-05 | 2018-11-23 | 北京旷视科技有限公司 | Method, apparatus, system and the computer storage medium of recognition of face |
| CN109034100A (en) * | 2018-08-13 | 2018-12-18 | 成都盯盯科技有限公司 | Face pattern detection method, device, equipment and storage medium |
| CN109063691A (en) * | 2018-09-03 | 2018-12-21 | 武汉普利商用机器有限公司 | A kind of recognition of face bottom library optimization method and system |
| CN109086739A (en) * | 2018-08-23 | 2018-12-25 | 成都睿码科技有限责任公司 | A kind of face identification method and system of no human face data training |
| WO2019037346A1 (en) * | 2017-08-25 | 2019-02-28 | 广州视源电子科技股份有限公司 | Method and device for optimizing human face picture quality evaluation model |
| CN109492560A (en) * | 2018-10-26 | 2019-03-19 | 深圳力维智联技术有限公司 | Facial image Feature fusion, device and storage medium based on time scale |
| CN109711357A (en) * | 2018-12-28 | 2019-05-03 | 北京旷视科技有限公司 | A kind of face identification method and device |
| CN109872406A (en) * | 2019-01-23 | 2019-06-11 | 北京影谱科技股份有限公司 | A kind of gradual judgement update method, device and face punch card system |
| CN109871767A (en) * | 2019-01-17 | 2019-06-11 | 平安科技（深圳）有限公司 | Face identification method, device, electronic equipment and computer readable storage medium |
| WO2019134246A1 (en) * | 2018-01-03 | 2019-07-11 | 平安科技（深圳）有限公司 | Facial recognition-based security monitoring method, device, and storage medium |
| CN110197107A (en) * | 2018-08-17 | 2019-09-03 | 平安科技（深圳）有限公司 | Micro- expression recognition method, device, computer equipment and storage medium |
| CN110232134A (en) * | 2019-06-13 | 2019-09-13 | 上海商汤智能科技有限公司 | Data-updating method, server and computer storage medium |
| CN110348315A (en) * | 2019-06-14 | 2019-10-18 | 深圳英飞拓科技股份有限公司 | Dynamic updates method and device, the face snap system in face characteristic bottom library |
| CN110363150A (en) * | 2019-07-16 | 2019-10-22 | 深圳市商汤科技有限公司 | Data-updating method and device, electronic equipment and storage medium |
| CN110532991A (en) * | 2019-09-04 | 2019-12-03 | 深圳市捷顺科技实业股份有限公司 | A kind of face identification method, device and equipment |
| CN110647823A (en) * | 2019-09-02 | 2020-01-03 | 中国建设银行股份有限公司 | Method and device for optimizing human face base |
| CN110765885A (en) * | 2019-09-29 | 2020-02-07 | 武汉大学 | City expansion detection method and device based on heterogeneous luminous remote sensing image |
| WO2020038136A1 (en) * | 2018-08-24 | 2020-02-27 | 深圳前海达闼云端智能科技有限公司 | Facial recognition method and apparatus, electronic device and computer-readable medium |
| CN111104825A (en) * | 2018-10-26 | 2020-05-05 | 北京陌陌信息技术有限公司 | Face registry updating method, device, equipment and medium |
| CN111160066A (en) * | 2018-11-07 | 2020-05-15 | 北京陌陌信息技术有限公司 | Face recognition method, device, equipment and medium |
| WO2020108268A1 (en) * | 2018-11-28 | 2020-06-04 | 杭州海康威视数字技术股份有限公司 | Face recognition system, method and apparatus |
| CN111241928A (en) * | 2019-12-30 | 2020-06-05 | 新大陆数字技术股份有限公司 | Face recognition base optimization method, system, equipment and readable storage medium |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006122009A2 (en) * | 2005-05-09 | 2006-11-16 | Lockheed Martin Corporation | Continuous extended range image processing |
| JP2010165305A (en) * | 2009-01-19 | 2010-07-29 | Sony Corp | Information processing apparatus, information processing method, and program |
| US9135680B2 (en) * | 2011-09-08 | 2015-09-15 | Bae Systems Information And Electronic Systems Integration Inc. | Method for reducing row and column noise in imaging systems |
| KR102185854B1 (en) * | 2017-09-09 | 2020-12-02 | 애플 인크. | Implementation of biometric authentication |
| US11216541B2 (en) * | 2018-09-07 | 2022-01-04 | Qualcomm Incorporated | User adaptation for biometric authentication |
-
2020
- 2020-06-12 CN CN202010533170.1A patent/CN111814570B/en active Active
Patent Citations (28)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20110109695A (en) * | 2010-03-31 | 2011-10-06 | 경북대학교 산학협력단 | Face recognition system and method thereof |
| CN102004908A (en) * | 2010-11-30 | 2011-04-06 | 汉王科技股份有限公司 | Self-adapting face identification method and device |
| CN105938552A (en) * | 2016-06-29 | 2016-09-14 | 北京旷视科技有限公司 | Face recognition method capable of realizing base image automatic update and face recognition device |
| WO2019037346A1 (en) * | 2017-08-25 | 2019-02-28 | 广州视源电子科技股份有限公司 | Method and device for optimizing human face picture quality evaluation model |
| CN107590212A (en) * | 2017-08-29 | 2018-01-16 | 深圳英飞拓科技股份有限公司 | The Input System and method of a kind of face picture |
| CN108170750A (en) * | 2017-12-21 | 2018-06-15 | 深圳英飞拓科技股份有限公司 | A kind of face database update method, system and terminal device |
| WO2019134246A1 (en) * | 2018-01-03 | 2019-07-11 | 平安科技（深圳）有限公司 | Facial recognition-based security monitoring method, device, and storage medium |
| CN108875534A (en) * | 2018-02-05 | 2018-11-23 | 北京旷视科技有限公司 | Method, apparatus, system and the computer storage medium of recognition of face |
| CN108446387A (en) * | 2018-03-22 | 2018-08-24 | 百度在线网络技术（北京）有限公司 | Method and apparatus for updating face registration library |
| CN109034100A (en) * | 2018-08-13 | 2018-12-18 | 成都盯盯科技有限公司 | Face pattern detection method, device, equipment and storage medium |
| CN110197107A (en) * | 2018-08-17 | 2019-09-03 | 平安科技（深圳）有限公司 | Micro- expression recognition method, device, computer equipment and storage medium |
| CN109086739A (en) * | 2018-08-23 | 2018-12-25 | 成都睿码科技有限责任公司 | A kind of face identification method and system of no human face data training |
| WO2020038136A1 (en) * | 2018-08-24 | 2020-02-27 | 深圳前海达闼云端智能科技有限公司 | Facial recognition method and apparatus, electronic device and computer-readable medium |
| CN109063691A (en) * | 2018-09-03 | 2018-12-21 | 武汉普利商用机器有限公司 | A kind of recognition of face bottom library optimization method and system |
| CN109492560A (en) * | 2018-10-26 | 2019-03-19 | 深圳力维智联技术有限公司 | Facial image Feature fusion, device and storage medium based on time scale |
| CN111104825A (en) * | 2018-10-26 | 2020-05-05 | 北京陌陌信息技术有限公司 | Face registry updating method, device, equipment and medium |
| CN111160066A (en) * | 2018-11-07 | 2020-05-15 | 北京陌陌信息技术有限公司 | Face recognition method, device, equipment and medium |
| WO2020108268A1 (en) * | 2018-11-28 | 2020-06-04 | 杭州海康威视数字技术股份有限公司 | Face recognition system, method and apparatus |
| CN109711357A (en) * | 2018-12-28 | 2019-05-03 | 北京旷视科技有限公司 | A kind of face identification method and device |
| CN109871767A (en) * | 2019-01-17 | 2019-06-11 | 平安科技（深圳）有限公司 | Face identification method, device, electronic equipment and computer readable storage medium |
| CN109872406A (en) * | 2019-01-23 | 2019-06-11 | 北京影谱科技股份有限公司 | A kind of gradual judgement update method, device and face punch card system |
| CN110232134A (en) * | 2019-06-13 | 2019-09-13 | 上海商汤智能科技有限公司 | Data-updating method, server and computer storage medium |
| CN110348315A (en) * | 2019-06-14 | 2019-10-18 | 深圳英飞拓科技股份有限公司 | Dynamic updates method and device, the face snap system in face characteristic bottom library |
| CN110363150A (en) * | 2019-07-16 | 2019-10-22 | 深圳市商汤科技有限公司 | Data-updating method and device, electronic equipment and storage medium |
| CN110647823A (en) * | 2019-09-02 | 2020-01-03 | 中国建设银行股份有限公司 | Method and device for optimizing human face base |
| CN110532991A (en) * | 2019-09-04 | 2019-12-03 | 深圳市捷顺科技实业股份有限公司 | A kind of face identification method, device and equipment |
| CN110765885A (en) * | 2019-09-29 | 2020-02-07 | 武汉大学 | City expansion detection method and device based on heterogeneous luminous remote sensing image |
| CN111241928A (en) * | 2019-12-30 | 2020-06-05 | 新大陆数字技术股份有限公司 | Face recognition base optimization method, system, equipment and readable storage medium |
Non-Patent Citations (5)
| Title |
|---|
| 基于多阈值机制的动态人脸识别警务实战应用;贾永洪;言专艺;何永铿;萧博铭;;现代信息科技;20180125(第01期);全文 * |
| 基于尺度不变特征变换优化算法的带遮挡人脸识别;周玲丽;赖剑煌;;计算机应用;20110630(第S1期);全文 * |
| 基于深度学习算法的改进型人脸识别系统实现――以智慧校园安防系统为例;张怡;戴闽鲁;雷国平;;信息系统工程;20180520(第05期);全文 * |
| 基于轻量网络的近红外光和可见光融合的异质人脸识别;张典;汪海涛;姜瑛;陈星;;小型微型计算机系统;20200409(第04期);全文 * |
| 自适应中心对称局部三值模式的人脸识别;闫河;王朴;刘婕;陈伟栋;;计算机应用与软件;20160915(第09期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111814570A (en) | 2020-10-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111814570B (en) | Face recognition method, system and storage medium based on dynamic threshold | |
| WO2021057848A1 (en) | Network training method, image processing method, network, terminal device and medium | |
| US11544628B2 (en) | Information processing apparatus and information processing method for generating classifier using target task learning data and source task learning data, and storage medium | |
| CN110728294A (en) | Cross-domain image classification model construction method and device based on transfer learning | |
| CN108875797B (en) | Method for determining image similarity, photo album management method and related equipment | |
| CN112948612B (en) | Human body cover generation method and device, electronic equipment and storage medium | |
| CN111325067B (en) | Illegal video identification method and device and electronic equipment | |
| CN111291887A (en) | Neural network training method, image recognition method, device and electronic equipment | |
| CN112766218A (en) | Cross-domain pedestrian re-identification method and device based on asymmetric joint teaching network | |
| KR102488789B1 (en) | Prediction and classification method, apparatus and program using one class anomaly detection model based on artificial intelligence | |
| CN116543261A (en) | Model training method for image recognition, image recognition method device and medium | |
| CN112581355A (en) | Image processing method, image processing device, electronic equipment and computer readable medium | |
| CN108257117B (en) | Image exposure evaluation method and device | |
| CN112070682A (en) | Method and device for compensating image brightness | |
| CN113902944A (en) | Model training and scene recognition method, device, equipment and medium | |
| CN112997148A (en) | Sleep prediction method, device, storage medium and electronic equipment | |
| CN112926663A (en) | Method and device for training classification model, computer equipment and storage medium | |
| WO2021047453A1 (en) | Image quality determination method, apparatus and device | |
| CN114219051B (en) | Image classification method, classification model training method and device and electronic equipment | |
| CN111126147A (en) | Image processing method, device and electronic system | |
| CN108875572B (en) | Pedestrian re-identification method based on background suppression | |
| CN114549502A (en) | Method and device for evaluating face quality, electronic equipment and storage medium | |
| CN115690514A (en) | Image recognition method and related equipment | |
| CN112884040B (en) | Training sample data optimization method, system, storage medium and electronic equipment | |
| CN114549884A (en) | Abnormal image detection method, device, equipment and 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |