CN111553429B - Fingerprint identification model migration method, device and system - Google Patents
Fingerprint identification model migration method, device and system Download PDFInfo
- Publication number
- CN111553429B CN111553429B CN202010360556.7A CN202010360556A CN111553429B CN 111553429 B CN111553429 B CN 111553429B CN 202010360556 A CN202010360556 A CN 202010360556A CN 111553429 B CN111553429 B CN 111553429B
- Authority
- CN
- China
- Prior art keywords
- fingerprint
- feature data
- fingerprint feature
- destination
- feature
- 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 41
- 238000013508 migration Methods 0.000 title claims abstract description 24
- 230000005012 migration Effects 0.000 title claims abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 97
- 230000008569 process Effects 0.000 claims abstract description 13
- 238000000605 extraction Methods 0.000 claims description 2
- 230000009286 beneficial effect Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000035755 proliferation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/214—Database migration support
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Collating Specific Patterns (AREA)
Abstract
The application provides a fingerprint identification model migration method and device, comprising the following steps: acquiring a fingerprint identification model corresponding to a source region; acquiring a fingerprint characteristic data set of a user corresponding to a source region and a fingerprint characteristic data set of a user corresponding to a destination region; comparing the fingerprint feature data set of the user corresponding to the source region with the fingerprint feature data set of the user corresponding to the destination region, and determining that the destination region is different from the source region; determining a personalized parameter set corresponding to the personalized fingerprint feature set in the fingerprint identification model; extracting a personal fingerprint feature data set corresponding to the personal fingerprint feature set from fingerprint feature data of a user corresponding to the destination area, and constructing a training sample set based on the personal fingerprint feature data set; performing supervised training on the fingerprint identification model based on the training sample set, and only adjusting the personalized parameter set in the supervised training process; and obtaining a fingerprint identification model suitable for the destination area after reaching the training ending condition.
Description
Technical Field
The present application relates to the field of communications technologies, and in particular, to a fingerprint identification model migration method, apparatus, and system.
Background
Most of the prior banking systems start to adopt the mode of biological recognition such as fingerprint and the like to realize client authentication, fingerprint information needs to be collected in the process of establishing a fingerprint system, the fingerprint information is converted into digital information through training a fingerprint recognition model, and the client fingerprint is recorded through the digital information.
The proliferation of random banks, some of which have begun to lay out global markets. User fingerprints in different regions in the global scope have differences, so that the recognition accuracy of the existing fingerprint recognition model in other countries is low.
Disclosure of Invention
In view of the above, the application provides a fingerprint identification model migration method, device and system, which can migrate the existing fingerprint identification model to different countries, simply and conveniently obtain fingerprint identification models suitable for different countries, and avoid retraining new fingerprint identification models.
In order to achieve the above object, the present application provides the following technical features:
a fingerprint recognition model migration method, comprising:
acquiring a fingerprint identification model corresponding to a source region;
acquiring a fingerprint characteristic data set of a user corresponding to a source region and a fingerprint characteristic data set of a user corresponding to a destination region;
comparing the fingerprint feature data set of the user corresponding to the source region with the fingerprint feature data set of the user corresponding to the destination region, and determining that the destination region is different from the source region personalized fingerprint feature set;
determining a personalized parameter set corresponding to the personalized fingerprint feature set in the fingerprint identification model;
extracting a personal fingerprint feature data set corresponding to a personal fingerprint feature set from fingerprint feature data of a user corresponding to the destination area, and constructing a training sample set based on the personal fingerprint feature data set;
performing supervised training on the fingerprint identification model based on the training sample set, and only adjusting the personalized parameter set in the supervised training process;
and obtaining a fingerprint identification model suitable for the destination area after reaching the training ending condition.
Optionally, the acquiring the fingerprint feature data set of the user corresponding to the source region, and the fingerprint feature data set of the user corresponding to the destination region includes:
acquiring fingerprint feature data sets of a preset number of users in the source region;
acquiring fingerprint feature data sets of a preset number of users in the destination area;
wherein the fingerprint feature data set comprises a plurality of fingerprint features and corresponding fingerprint feature data;
the plurality of fingerprint features includes: the pattern feature, the pattern region feature, the core point feature, the triangular point feature and the pattern number feature.
Optionally, the comparing the fingerprint feature data set of the user corresponding to the source region with the fingerprint feature data set of the user corresponding to the destination region, and determining that the destination region is different from the source region personalized fingerprint feature set includes:
calculating the similarity of the fingerprint feature data of the source region and the fingerprint feature data of the destination region aiming at each fingerprint feature in the fingerprint feature data set;
if the similarity of the fingerprint features is greater than the preset similarity, determining that the fingerprint features are common fingerprint features of the source region and the destination region;
if the similarity of the fingerprint features is not greater than the preset similarity, determining that the fingerprint features are personalized fingerprint features of which the destination area is different from the source area;
and combining the individual fingerprint features into an individual fingerprint feature set of the destination area which is different from the source area.
Optionally, the determining the personality parameter set corresponding to the personality fingerprint feature set in the fingerprint recognition model includes:
inquiring each individual fingerprint feature in the individual fingerprint feature set in the corresponding relation between the pre-constructed fingerprint feature and the model parameter;
and forming the model parameters corresponding to the individual fingerprint characteristics into the individual parameter set.
Optionally, after the obtaining the fingerprint identification model applicable to the destination area after reaching the training end condition, the method further includes:
constructing a training sample set based on fingerprint feature data sets of a plurality of users in the destination area;
performing supervised training on the fingerprint identification model based on the training sample set;
and obtaining the optimized fingerprint identification model of the destination area after reaching the training ending condition.
A fingerprint recognition model migration apparatus comprising:
the first acquisition unit is used for acquiring a fingerprint identification model corresponding to a source region;
the second acquisition unit is used for acquiring a fingerprint characteristic data set of the user corresponding to the source region and a fingerprint characteristic data set of the user corresponding to the destination region;
the comparison unit is used for comparing the fingerprint feature data set of the user corresponding to the source region with the fingerprint feature data set of the user corresponding to the destination region, and determining that the destination region is different from the source region personalized fingerprint feature set;
the determining unit is used for determining a personalized parameter set corresponding to the personalized fingerprint feature set in the fingerprint identification model;
the extraction unit is used for extracting a personalized fingerprint feature data set corresponding to the personalized fingerprint feature set from the fingerprint feature data of the user corresponding to the destination area, and constructing a training sample set based on the personalized fingerprint feature data set;
the training unit is used for performing supervised training on the fingerprint identification model based on the training sample set, and only adjusting the personalized parameter set in the supervised training process;
and the obtaining unit is used for obtaining the fingerprint identification model applicable to the destination area after reaching the training ending condition.
Optionally, the second obtaining unit includes:
the source acquisition unit is used for acquiring fingerprint characteristic data sets of a preset number of users in the source region;
the target acquisition unit is used for acquiring fingerprint characteristic data sets of a preset number of users in the target area;
wherein the fingerprint feature data set comprises a plurality of fingerprint features and corresponding fingerprint feature data;
the plurality of fingerprint features includes: the pattern feature, the pattern region feature, the core point feature, the triangular point feature and the pattern number feature.
Optionally, the comparing unit includes:
calculating the similarity of the fingerprint feature data of the source region and the fingerprint feature data of the destination region aiming at each fingerprint feature in the fingerprint feature data set;
if the similarity of the fingerprint features is greater than the preset similarity, determining that the fingerprint features are common fingerprint features of the source region and the destination region;
if the similarity of the fingerprint features is not greater than the preset similarity, determining that the fingerprint features are personalized fingerprint features of which the destination area is different from the source area;
and combining the individual fingerprint features into an individual fingerprint feature set of the destination area which is different from the source area.
Optionally, the determining unit includes:
the query unit is used for querying each individual fingerprint feature in the individual fingerprint feature set in the corresponding relation between the pre-constructed fingerprint feature and the model parameter;
and the composition unit is used for composing the model parameters corresponding to the individual fingerprint characteristics into the individual parameter set.
Optionally, after obtaining the unit, further comprising:
the fine tuning unit is used for constructing a training sample set based on fingerprint characteristic data sets of a plurality of users in the destination area; performing supervised training on the fingerprint identification model based on the training sample set; and obtaining the optimized fingerprint identification model of the destination area after reaching the training ending condition.
Through the technical means, the following beneficial effects can be realized:
the application provides a fingerprint identification model migration method, which can identify a personalized parameter set of a fingerprint identification model, wherein a destination area is different from a source area, then acquire a training sample set corresponding to the personalized parameter set, and perform supervised training on the fingerprint identification model of the source area based on the training sample set.
The method has the advantages that a brand new fingerprint identification model does not need to be built and retrained, and migration is performed on the basis of a source region, so that a fingerprint identification model suitable for a destination region is obtained, and the existing fingerprint identification model of a meta region is migrated to the destination region simply and conveniently.
Drawings
In order to more clearly illustrate the embodiments of the application or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a first embodiment of a fingerprint identification model migration method disclosed in the present application;
FIG. 2 is a flowchart of a first embodiment of a fingerprint identification model migration method according to an embodiment of the present application;
FIG. 3 is a flowchart of a second embodiment of a fingerprint identification model migration method according to the present application;
fig. 4 is a schematic structural diagram of a fingerprint identification model migration device according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application 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 application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Referring to fig. 1, the present application provides a first embodiment of a fingerprint identification model migration method, including:
step S101: and acquiring a fingerprint identification model corresponding to the source region.
The fingerprint identification model corresponding to the source region is existing, and is very suitable for fingerprint identification of the source region, so that the accuracy is high.
Step S102: and acquiring a fingerprint characteristic data set of the user corresponding to the source region and a fingerprint characteristic data set of the user corresponding to the destination region.
It will be appreciated that the fingerprint feature data set of the user corresponding to the source region already exists, i.e. the fingerprint feature data of the user corresponding to the source region is already obtained before training the fingerprint recognition model corresponding to the source region.
Optionally acquiring fingerprint feature data sets of a preset number of users in the source region; wherein the fingerprint feature data set comprises a plurality of fingerprint features and corresponding fingerprint feature data; the plurality of fingerprint features includes: the pattern feature, the pattern region feature, the core point feature, the triangular point feature and the pattern number feature.
This step may determine only the fingerprint feature data set of the user corresponding to the destination area.
Optionally, fingerprint images of a preset number of users in the destination area may be acquired, and image processing and pattern recognition are performed on the fingerprint images, so as to obtain a fingerprint feature dataset of the preset number of users in the source area.
Wherein the fingerprint feature data set comprises a plurality of fingerprint features and corresponding fingerprint feature data; the plurality of fingerprint features includes: the pattern feature, the pattern region feature, the core point feature, the triangular point feature and the pattern number feature.
It will be appreciated that the fingerprint data set of the source region and the fingerprint data set of the destination region are identical, but the fingerprint data are not identical.
Step S103: comparing the fingerprint feature data set of the user corresponding to the source region with the fingerprint feature data set of the user corresponding to the destination region, and determining that the destination region is different from the source region personalized fingerprint feature set.
It will be appreciated that the users in both the source and destination regions are human users, so there must be common fingerprint features of the human user's fingerprint, except for some personalized fingerprint features.
For this purpose, the fingerprint feature data set of the user corresponding to the source region and the fingerprint feature data set of the user corresponding to the destination region may be compared, and it may be determined that the destination region is different from the source region personalized fingerprint feature set.
Referring to fig. 2, the present step includes the following technical features:
step S201: and calculating the similarity of the fingerprint feature data of the source region and the fingerprint feature data of the destination region aiming at each fingerprint feature in the fingerprint feature data set.
There are various calculation modes for calculating the similarity between the fingerprint feature data of the source region and the fingerprint feature data of the destination region, and one calculation mode is provided in this embodiment.
For each fingerprint feature in the fingerprint feature dataset, calculating variances of a preset number of fingerprint feature data in the source region and a preset number of fingerprint feature data in the destination region.
Step S202: and if the similarity of the fingerprint features is greater than the preset similarity, determining that the fingerprint features are common fingerprint features of the source region and the destination region.
Continuing the calculation mode, if the variance of one fingerprint feature is smaller than the preset variance, that is, the similarity of one fingerprint feature is larger than the preset similarity, determining that the fingerprint feature is the common fingerprint feature of the source region and the destination region.
Step S203: and if the similarity of the fingerprint features is not greater than the preset similarity, determining that the fingerprint features are the personalized fingerprint features of which the destination area is different from the source area.
Continuing the calculation mode, if the variance of one fingerprint feature is not smaller than the preset variance, namely the similarity of one fingerprint feature is not larger than the preset similarity, determining that the fingerprint feature is the personalized fingerprint feature of the source region and the destination region.
Step S204: and combining the individual fingerprint features into an individual fingerprint feature set of the destination area which is different from the source area.
Step S103 proceeds to step S104: and determining a personalized parameter set corresponding to the personalized fingerprint feature set in the fingerprint identification model.
In the training process of the fingerprint identification model of the source region, the corresponding relation between each fingerprint feature and the model parameter can be clearly understood, for example, one fingerprint feature 1 can influence the model parameter 1 and the model parameter 2, the other fingerprint feature can influence the model parameter 3 and the model parameter 4, and the like.
For this reason, each individual fingerprint feature in the individual fingerprint feature set can be queried in the corresponding relation between the pre-constructed fingerprint feature and the model parameter; and forming the model parameters corresponding to the individual fingerprint characteristics into the individual parameter set.
It will be appreciated that the fingerprint recognition model has many model parameters, except the personality parameter set, which are common parameter sets common to the source and destination regions, so no adjustment is necessary. Since the personality parameter set is specific to the destination region, the fingerprint identification model of the source region does not conform to the personality parameter set of the destination region.
Step S105: and extracting a personal fingerprint characteristic data set corresponding to the personal fingerprint characteristic set from the fingerprint characteristic data of the user corresponding to the destination area, and constructing a training sample set based on the personal fingerprint characteristic data set.
And extracting only the personalized fingerprint feature data set corresponding to the personalized fingerprint feature set from the fingerprint feature data of the users corresponding to the preset number of the destination area, wherein the rest are the common fingerprint feature data sets.
It will be appreciated that a user corresponds to a personality fingerprint feature data set and that a user corresponds to a user identifier (which may be a digital number); the personalized fingerprint characteristic data set of a user and the user identification are a training sample. The plurality of training samples form a training sample set.
Step S106: and performing supervised training on the fingerprint identification model based on the training sample set, and only adjusting the personalized parameter set in the supervised training process.
And performing supervised training on the fingerprint identification model of the source region by using the training sample set, wherein the individual parameter set is only adjusted in the training process, so that the fingerprint identification model is gradually applicable to the individual parameter set of the destination region.
Step S107: and obtaining a fingerprint identification model suitable for the destination area after reaching the training ending condition.
Through the technical means, the following beneficial effects can be realized:
the application provides a fingerprint identification model migration method, which can identify a personalized parameter set of a fingerprint identification model, wherein a destination area is different from a source area, then acquire a training sample set corresponding to the personalized parameter set, and perform supervised training on the fingerprint identification model of the source area based on the training sample set.
The method has the advantages that a brand new fingerprint identification model does not need to be built and retrained, and migration is performed on the basis of a source region, so that a fingerprint identification model suitable for a destination region is obtained, and the existing fingerprint identification model of a meta region is migrated to the destination region simply and conveniently.
The application provides a second embodiment of a fingerprint identification model migration method, and step S108, step S109 and step S110 are added on the basis of the first embodiment. Referring to fig. 3, the method comprises the steps of:
in step S107, training is performed using a predetermined number of training samples, and a fingerprint identification model of the destination area is obtained, which is already available. For more accuracy, the fingerprint recognition model of the destination area is trimmed again using the fingerprint feature data sets of the plurality of users.
Step S108: and constructing a training sample set based on the fingerprint feature data sets of a plurality of users in the destination area.
A fingerprint feature data set of a plurality of users is obtained from a preset number of destination areas, where the fingerprint feature data set includes both a personality feature data set and a commonality feature data set.
Wherein the fingerprint feature data set comprises a plurality of fingerprint features and corresponding fingerprint feature data; the plurality of fingerprint features includes: the pattern feature, the pattern region feature, the core point feature, the triangular point feature and the pattern number feature.
It will be appreciated that a user corresponds to a fingerprint feature data set and that a user corresponds to a user identification (which may be a digital number); the fingerprint feature data set and the user identification of a user are a training sample. The plurality of training samples form a training sample set.
Step S109: and performing supervised training on the fingerprint identification model based on the training sample set.
And performing supervised training on the fingerprint identification model based on the training sample set, wherein all model parameters can be adjusted in the training process.
Step S110: and obtaining the optimized fingerprint identification model of the destination area after reaching the training ending condition.
Steps S108 to S110 are processes of fine tuning the fingerprint recognition model of the destination area obtained in step S107, and since a small amount of training samples are used, training is relatively simple and convenient.
Through the technical means, the following beneficial effects can be realized:
the application provides a fingerprint identification model migration method, which can identify a personalized parameter set of a fingerprint identification model, wherein a destination area is different from a source area, then acquire a training sample set corresponding to the personalized parameter set, and perform supervised training on the fingerprint identification model of the source area based on the training sample set.
The method has the advantages that a brand new fingerprint identification model does not need to be built and retrained, and migration is performed on the basis of a source region, so that a fingerprint identification model suitable for a destination region is obtained, and the existing fingerprint identification model of a meta region is migrated to the destination region simply and conveniently.
And moreover, the fingerprint identification model of the destination area can be finely adjusted, so that the fingerprint identification model is more accurate.
Referring to fig. 4, the present application provides a fingerprint identification model migration apparatus, including:
a first obtaining unit 41, configured to obtain a fingerprint identification model corresponding to a source region;
a second acquiring unit 42, configured to acquire a fingerprint feature data set of a user corresponding to a source region and a fingerprint feature data set of a user corresponding to a destination region;
a comparing unit 43, configured to compare a fingerprint feature data set of a user corresponding to a source region with a fingerprint feature data set of a user corresponding to a destination region, and determine that the destination region is different from the source region personalized fingerprint feature set;
a determining unit 44, configured to determine a set of personality parameters corresponding to the set of personality fingerprint features in the fingerprint recognition model;
an extracting unit 45, configured to extract a personalized fingerprint feature data set corresponding to a personalized fingerprint feature set from fingerprint feature data of a user corresponding to the destination area, and construct a training sample set based on the personalized fingerprint feature data set;
a training unit 46, configured to perform supervised training on the fingerprint identification model based on the training sample set, and adjust only the personality parameter set during the supervised training;
an obtaining unit 47 is configured to obtain a fingerprint identification model applicable to the destination area after reaching the training end condition.
Wherein the second acquisition unit 42 includes:
a source obtaining unit 421, configured to obtain a fingerprint feature dataset of a preset number of users in the source region;
a destination obtaining unit 422, configured to obtain a fingerprint feature dataset of a preset number of users in the destination area;
wherein the fingerprint feature data set comprises a plurality of fingerprint features and corresponding fingerprint feature data;
the plurality of fingerprint features includes: the pattern feature, the pattern region feature, the core point feature, the triangular point feature and the pattern number feature.
Wherein the comparing unit 43 comprises:
calculating the similarity of the fingerprint feature data of the source region and the fingerprint feature data of the destination region aiming at each fingerprint feature in the fingerprint feature data set;
if the similarity of the fingerprint features is greater than the preset similarity, determining that the fingerprint features are common fingerprint features of the source region and the destination region;
if the similarity of the fingerprint features is not greater than the preset similarity, determining that the fingerprint features are personalized fingerprint features of which the destination area is different from the source area;
and combining the individual fingerprint features into an individual fingerprint feature set of the destination area which is different from the source area.
Wherein the determining unit 44 comprises:
a query unit 441, configured to query each individual fingerprint feature in the individual fingerprint feature set in a corresponding relationship between a pre-constructed fingerprint feature and a model parameter;
a composing unit 442, configured to compose the model parameters corresponding to each individual fingerprint feature into the individual parameter set.
Wherein after obtaining the unit 47, further comprises:
a fine tuning unit 48, configured to construct a training sample set based on fingerprint feature data sets of a plurality of users in the destination area; performing supervised training on the fingerprint identification model based on the training sample set; and obtaining the optimized fingerprint identification model of the destination area after reaching the training ending condition.
Through the technical means, the following beneficial effects can be realized:
the application provides a fingerprint identification model migration method, which can identify a personalized parameter set of a fingerprint identification model, wherein a destination area is different from a source area, then acquire a training sample set corresponding to the personalized parameter set, and perform supervised training on the fingerprint identification model of the source area based on the training sample set.
The method has the advantages that a brand new fingerprint identification model does not need to be built and retrained, and migration is performed on the basis of a source region, so that a fingerprint identification model suitable for a destination region is obtained, and the existing fingerprint identification model of a meta region is migrated to the destination region simply and conveniently.
The functions described in the method of this embodiment, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computing device readable storage medium. Based on such understanding, a part of the present application that contributes to the prior art or a part of the technical solution may be embodied in the form of a software product stored in a storage medium, comprising several instructions for causing a computing device (which may be a personal computer, a server, a mobile computing device or a network device, etc.) to execute all or part of the steps of the method described in the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In this specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, so that the same or similar parts between the embodiments are referred to each other.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (8)
1. A fingerprint identification model migration method, comprising:
acquiring a fingerprint identification model corresponding to a source region;
acquiring a fingerprint characteristic data set of a user corresponding to a source region and a fingerprint characteristic data set of a user corresponding to a destination region;
comparing the fingerprint feature data set of the user corresponding to the source region with the fingerprint feature data set of the user corresponding to the destination region, and determining that the destination region is different from the source region personalized fingerprint feature set;
determining a personalized parameter set corresponding to the personalized fingerprint feature set in the fingerprint identification model;
extracting a personal fingerprint feature data set corresponding to a personal fingerprint feature set from fingerprint feature data of a user corresponding to the destination area, and constructing a training sample set based on the personal fingerprint feature data set;
performing supervised training on the fingerprint identification model based on the training sample set, and only adjusting the personalized parameter set in the supervised training process;
obtaining a fingerprint identification model applicable to the destination area after reaching the training ending condition;
after the fingerprint identification model applicable to the destination area is obtained after the training ending condition is reached, the method further comprises the following steps:
constructing a training sample set based on fingerprint feature data sets of a plurality of users in the destination area, wherein the fingerprint feature data sets comprise a personalized feature data set and a common feature data set;
performing supervised training on the fingerprint identification model based on the training sample set, and adjusting all model parameters in the training process;
and obtaining the optimized fingerprint identification model of the destination area after reaching the training ending condition.
2. The method of claim 1, wherein the acquiring the fingerprint feature dataset of the user corresponding to the source region and the fingerprint feature dataset of the user corresponding to the destination region comprises:
acquiring fingerprint feature data sets of a preset number of users in the source region;
acquiring fingerprint feature data sets of a preset number of users in the destination area;
wherein the fingerprint feature data set comprises a plurality of fingerprint features and corresponding fingerprint feature data;
the plurality of fingerprint features includes: the pattern feature, the pattern region feature, the core point feature, the triangular point feature and the pattern number feature.
3. The method of claim 2, wherein the comparing the fingerprint feature data set of the source region corresponding user with the fingerprint feature data set of the destination region corresponding user, determining that the destination region is different from the source region personalized fingerprint feature set comprises:
calculating the similarity of the fingerprint feature data of the source region and the fingerprint feature data of the destination region aiming at each fingerprint feature in the fingerprint feature data set;
if the similarity of the fingerprint features is greater than the preset similarity, determining that the fingerprint features are common fingerprint features of the source region and the destination region;
if the similarity of the fingerprint features is not greater than the preset similarity, determining that the fingerprint features are personalized fingerprint features of which the destination area is different from the source area;
and combining the individual fingerprint features into an individual fingerprint feature set of the destination area which is different from the source area.
4. The method of claim 1, wherein the determining the set of personality parameters in the fingerprint recognition model that correspond to the set of personality fingerprint features comprises:
inquiring each individual fingerprint feature in the individual fingerprint feature set in the corresponding relation between the pre-constructed fingerprint feature and the model parameter;
and forming the model parameters corresponding to the individual fingerprint characteristics into the individual parameter set.
5. A fingerprint identification model migration apparatus, characterized by comprising:
the first acquisition unit is used for acquiring a fingerprint identification model corresponding to a source region;
the second acquisition unit is used for acquiring a fingerprint characteristic data set of the user corresponding to the source region and a fingerprint characteristic data set of the user corresponding to the destination region;
the comparison unit is used for comparing the fingerprint feature data set of the user corresponding to the source region with the fingerprint feature data set of the user corresponding to the destination region, and determining that the destination region is different from the source region personalized fingerprint feature set;
the determining unit is used for determining a personalized parameter set corresponding to the personalized fingerprint feature set in the fingerprint identification model;
the extraction unit is used for extracting a personalized fingerprint feature data set corresponding to the personalized fingerprint feature set from the fingerprint feature data of the user corresponding to the destination area, and constructing a training sample set based on the personalized fingerprint feature data set;
the training unit is used for performing supervised training on the fingerprint identification model based on the training sample set, and only adjusting the personalized parameter set in the supervised training process;
the obtaining unit is used for obtaining a fingerprint identification model applicable to the destination area after reaching the training ending condition;
after obtaining the unit, further comprising:
the fine tuning unit is used for constructing a training sample set based on fingerprint feature data sets of a plurality of users in the destination area, wherein the fingerprint feature data sets comprise a personalized feature data set and a common feature data set; performing supervised training on the fingerprint identification model based on the training sample set, and adjusting all model parameters in the training process; and obtaining the optimized fingerprint identification model of the destination area after reaching the training ending condition.
6. The apparatus of claim 5, wherein the second acquisition unit comprises:
the source acquisition unit is used for acquiring fingerprint characteristic data sets of a preset number of users in the source region;
the target acquisition unit is used for acquiring fingerprint characteristic data sets of a preset number of users in the target area;
wherein the fingerprint feature data set comprises a plurality of fingerprint features and corresponding fingerprint feature data;
the plurality of fingerprint features includes: the pattern feature, the pattern region feature, the core point feature, the triangular point feature and the pattern number feature.
7. The apparatus of claim 6, wherein the contrast unit comprises:
calculating the similarity of the fingerprint feature data of the source region and the fingerprint feature data of the destination region aiming at each fingerprint feature in the fingerprint feature data set;
if the similarity of the fingerprint features is greater than the preset similarity, determining that the fingerprint features are common fingerprint features of the source region and the destination region;
if the similarity of the fingerprint features is not greater than the preset similarity, determining that the fingerprint features are personalized fingerprint features of which the destination area is different from the source area;
and combining the individual fingerprint features into an individual fingerprint feature set of the destination area which is different from the source area.
8. The apparatus of claim 5, wherein the determining unit comprises:
the query unit is used for querying each individual fingerprint feature in the individual fingerprint feature set in the corresponding relation between the pre-constructed fingerprint feature and the model parameter;
and the composition unit is used for composing the model parameters corresponding to the individual fingerprint characteristics into the individual parameter set.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010360556.7A CN111553429B (en) | 2020-04-30 | 2020-04-30 | Fingerprint identification model migration method, device and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010360556.7A CN111553429B (en) | 2020-04-30 | 2020-04-30 | Fingerprint identification model migration method, device and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111553429A CN111553429A (en) | 2020-08-18 |
CN111553429B true CN111553429B (en) | 2023-11-03 |
Family
ID=72003347
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010360556.7A Active CN111553429B (en) | 2020-04-30 | 2020-04-30 | Fingerprint identification model migration method, device and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111553429B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107506775A (en) * | 2016-06-14 | 2017-12-22 | 北京陌上花科技有限公司 | model training method and device |
CN107832711A (en) * | 2017-11-13 | 2018-03-23 | 常州大学 | A kind of recognition methods again of the pedestrian based on transfer learning |
CN109376699A (en) * | 2018-11-30 | 2019-02-22 | 昆明理工大学 | A kind of fingerprint identification method based on convolutional neural networks |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110598840B (en) * | 2018-06-13 | 2023-04-18 | 富士通株式会社 | Knowledge migration method, information processing apparatus, and storage medium |
-
2020
- 2020-04-30 CN CN202010360556.7A patent/CN111553429B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107506775A (en) * | 2016-06-14 | 2017-12-22 | 北京陌上花科技有限公司 | model training method and device |
CN107832711A (en) * | 2017-11-13 | 2018-03-23 | 常州大学 | A kind of recognition methods again of the pedestrian based on transfer learning |
CN109376699A (en) * | 2018-11-30 | 2019-02-22 | 昆明理工大学 | A kind of fingerprint identification method based on convolutional neural networks |
Also Published As
Publication number | Publication date |
---|---|
CN111553429A (en) | 2020-08-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2016115940A1 (en) | Fingerprint information dynamic updating method and fingerprint recognition apparatus | |
Wang et al. | Toward accurate localization and high recognition performance for noisy iris images | |
US8995729B2 (en) | Accelerated comparison using scores from coarse and fine matching processes | |
CN107301389B (en) | Method, device and terminal for identifying user gender based on face features | |
CN110147710B (en) | Method and device for processing human face features and storage medium | |
CN111125390B (en) | Database updating method and device, electronic equipment and computer storage medium | |
CN110110118A (en) | Dressing recommended method, device, storage medium and mobile terminal | |
CN109002784B (en) | Street view identification method and system | |
CN108875629B (en) | Palm vein identification method based on multi-sample feature fusion | |
US20160253577A1 (en) | Image Clustering Method, Image Clustering System, And Image Clustering Server | |
WO2023273058A1 (en) | Identity identification method, system and apparatus, and computer device and storage medium | |
CN113221086B (en) | Offline face authentication method and device, electronic equipment and storage medium | |
CN111709303A (en) | Face image recognition method and device | |
EP3371739A1 (en) | High speed reference point independent database filtering for fingerprint identification | |
CN110851675B (en) | Data extraction method, device and medium | |
CN111553429B (en) | Fingerprint identification model migration method, device and system | |
CN110889009A (en) | Voiceprint clustering method, voiceprint clustering device, processing equipment and computer storage medium | |
JP7133085B2 (en) | Database update method and device, electronic device, and computer storage medium | |
CN108630208B (en) | Server, voiceprint-based identity authentication method and storage medium | |
CN105224957B (en) | A kind of method and system of the image recognition based on single sample | |
TWI804441B (en) | Method, device, equipment and medium for updating feature vector database | |
CN115984178A (en) | Counterfeit image detection method, electronic device, and computer-readable storage medium | |
CN114639127A (en) | Fingerprint identification method and device, storage medium and electronic equipment | |
WO2021151359A1 (en) | Palm print image recognition method, apparatus and device, and computer readable storage medium | |
CN110348898B (en) | Information pushing method and device based on human body identification |
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 |