CN108596112A - A kind of choice of dynamical method of extensive iris feature identification matching threshold - Google Patents
A kind of choice of dynamical method of extensive iris feature identification matching threshold Download PDFInfo
- Publication number
- CN108596112A CN108596112A CN201810389947.4A CN201810389947A CN108596112A CN 108596112 A CN108596112 A CN 108596112A CN 201810389947 A CN201810389947 A CN 201810389947A CN 108596112 A CN108596112 A CN 108596112A
- Authority
- CN
- China
- Prior art keywords
- similarity
- matching threshold
- iris
- enrollment
- distribution map
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/18—Eye characteristics, e.g. of the iris
- G06V40/197—Matching; Classification
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Ophthalmology & Optometry (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention discloses a kind of choice of dynamical methods that extensive iris feature identifies matching threshold, include the following steps：S1. the similarity between all classes of iris enrollment is calculated；S2. the similarity counted in S1 forms similarity distribution map；S3. according to the false acceptance rate set in advance, the similarity distribution map in S2 is integrated, determines matching threshold T；S4. often increase N number of iris enrollment, repeat step S1 S3, new similarity distribution map is obtained, to obtain new matching threshold T.The present invention is according to the distribution situation and quantity of enrollment, it is dynamically determined matching threshold, the choice of dynamical method of matching threshold provided by the invention is identified suitable for extensive iris feature, on the one hand it ensure that the safety of iris feature system, on the other hand the percent of pass for improving iris feature system as far as possible, to further improve the experience sense of user.
Description
Technical field
The present invention relates to iris recognition technology fields more particularly to a kind of extensive iris feature to identify the dynamic of matching threshold
State choosing method.
Background technology
With popularizing for biometrics identification technology, more and more people pay close attention to biometrics identification technology and simultaneously make correlation
Product largely occurs, and in actual scene, with the substantial increase of registration user, is proposed to living creature characteristic recognition system new
Challenge registers the increase of user along with the increase for accidentally knowing probability.In order to reduce false acceptance rate, general there are two types of methods：One
It is the more robust algorithm of exploitation, makes to be spaced further apart between class in class, second is that properly increasing matching threshold.First method pair
For a fixed system, it cannot change, second method can use in actual scene, but if matching threshold tune
Excessively high, false rejection rate can increase, and user experience can also decline, so how to determine that new matching threshold is one and is worth grinding
The project studied carefully.
Invention content
To solve the above-mentioned problems, the object of the present invention is to provide a kind of choice of dynamical method of matching threshold, this method
It is matching threshold to be determined according to the distribution situation of enrollment, and the matching threshold can also be with the distribution situation of enrollment
Dynamic change occurs.
The present invention is achieved by the following technical solutions：A kind of dynamic choosing of extensive iris feature identification matching threshold
Method is taken, is included the following steps：
S1. the similarity between all classes of iris enrollment is calculated；
S2. the similarity counted in S1 forms similarity distribution map；
S3. according to preset false acceptance rate ε, the similarity distribution map in S2 is integrated, determines matching threshold T,
There is following relational expression：
Wherein, x is the expectation and variance yields of similarity, μ and σ for similarity distribution map；
S4. often increase N number of iris enrollment, repeat step S1-S3, new similarity distribution map is obtained, to obtain
New matching threshold T.
Further, the matching threshold T in step S3 is obtained by following steps：
S31. remember that similarity distribution map Normal Distribution, the normal distyribution function are：
Wherein, x is the expectation and variance yields of similarity, μ and σ for similarity distribution map；
S32. remember that preset false acceptance rate is ε, there is following relational expression：
S is obtained according to formula (2) solution；
S33. note matching threshold T is：T=s.
Further, similarity is obtained by following steps between the class described in step S1：
S11. iris enrollment is respectively F1 and F2 between remembering two classes, then normalization Hamming distance between the two is：
Wherein, XOR is exclusive or operator, and the length of F1 and F2 is identical, and Length (F1) is the length of F1；
S12. the similarity that iris registration between two classes is calculated according to the normalization Hamming distance in step S11, calculates public
Formula is as follows：
Sim (F1, F2)=1-HD (F1, F2),
Wherein, Sim (F1, F2) is the similarity of iris enrollment between two classes, and HD (F1, F2) is iris between two classes
Normalization Hamming distance between enrollment；
S13. step S11 and step S12 is repeated, until iris enrollment similarity calculation finishes between all classes two-by-two.
The present invention is dynamically determined matching threshold, matching provided by the invention according to the distribution situation and quantity of enrollment
The choice of dynamical method of threshold value is identified suitable for extensive iris feature, on the one hand ensure that the safety of iris feature system,
On the other hand the percent of pass for improving iris feature system as far as possible, to further improve the experience sense of user.
Description of the drawings
Fig. 1 is the flow diagram of the present invention.
Specific implementation mode
Following further describes the present invention with reference to the drawings.
The choice of dynamical method of matching threshold provided by the invention is suitable for different application scene, including but not limited to terminal
Unlocking screen, using unlock and the scenes such as mobile payment.Living things feature recognition can generally use three indexs in evaluation model,
It is false rejection rate (False Rejection Rate, FRR, also referred to as refuse sincere), false acceptance rate (False respectively
Acceptance Rate, FAR, also referred to as accuracy of system identification) and etc. error rates (EER-Equal Error Rate).False rejection rate and
False acceptance rate can be adjusted by the biological characteristic quantity compared in biometric identification process, for pair of biological characteristic
Than, it is general using by biological characteristic from big feature to small feature again to the progressive contrastive pattern of minutia, the feature of comparison is got over
More, false rejection rate is higher, and false acceptance rate is lower, and matching threshold is also higher, and the probability of successful match is lower；The feature of comparison
Fewer, false rejection rate is lower, and false acceptance rate is higher, and matching threshold is also lower, and the probability of successful match is higher.In existing skill
In art, once once manufacturer terminal is to the debugging of false rejection rate and false acceptance rate success, basic just relatively more fixed or wave
Dynamic very little, cannot be satisfied the demand of different application scene, the present invention is according to preset false acceptance rate and extensive rainbow
The distribution situation and quantity of film enrollment adjust matching threshold into Mobile state, to cope with a large amount of of iris enrollment quantity
Increase.
The choice of dynamical method of extensive iris feature identification matching threshold as shown in Figure 1, includes the following steps：
S1. the similarity x between all classes of iris enrollment is calculated；
S2. the similarity counted in S1 forms similarity distribution map；
S3. according to the false acceptance rate ε set in advance, the similarity distribution map in S2 is integrated, determines matching
Threshold value T；
S4. often increase N number of iris enrollment, repeat step S1-S3, new similarity distribution map is obtained, to obtain
New matching threshold T.
Similarity can be calculated by following formula between class described in step S1：
S11. iris enrollment is respectively F1 and F2 between remembering two classes, then normalization Hamming distance between the two is：
Wherein, XOR is exclusive or operator, and the length of F1 and F2 is identical, and Length (F1) is the length of F1；
S12. the similarity that iris registration between two classes is calculated according to the normalization Hamming distance in step S31, calculates public
Formula is as follows：
Sim (F1, F2)=1-HD (F1, F2),
Wherein, Sim (F1, F2) is the similarity of iris enrollment between two classes, and HD (F1, F2) is iris between two classes
Normalization Hamming distance between enrollment；
S13. step S11 and step S12 is repeated, until iris enrollment similarity calculation finishes between all classes two-by-two.
The present invention is identified suitable for extensive iris feature, when iris enrollment quantity is larger, such as reach ten million with
On, bi-distribution approximation be in normal distribution, the present invention in registered template number need to reach 10,000,000 or more；John
G.Daugman obey bi-distribution and (be specifically shown in " High between verified iris enrollment class by normalization Hamming distance
Confidence Visual Recognition of Persons by a Test of Statistical
Independence ", John G.Daugman, IEEE TRANSACTIONS ON PATTERN ANALYSIS AND
MACHINE INTELLIGENCE, VOL.15, NO.11, NOVEMBER 1993, P1148-1161), and similarity and normalization
Hamming distance is in a linear relationship, therefore also obeys bi-distribution.According to person in servitude's Mo Fo-laplace's theorem：
If stochastic variable Xn~B (n, p) n=1,2 ... then has arbitrary x ∈ R：
When n is bigger, more approximate Normal Distribution.In the present invention, above formula n refers to comparing between iris enrollment class
To number, XnRefer to that similarity between iris enrollment class, p refer to identifying successful probability, q=1-p.The present invention is ground
What is studied carefully is extensive matching problem, therefore similarity Normal Distribution between extensive iris enrollment class, so in step
In S3, the method for determining matching threshold T is as follows：
S31. remember that similarity distribution map Normal Distribution, the normal distyribution function are：
Wherein, x is the expectation and variance yields of similarity, μ and σ for similarity distribution map；
S32. remember that preset false acceptance rate is ε, there is following relational expression：
S is obtained according to formula (2) solution；
S33. note matching threshold T is：T=s.
In step s 4, if system detectio increases to iris enrollment quantity, often increase N (General N >=5000) a rainbow
Film enrollment returns to step S1-S3, will be recalculated together with chartered template and the N number of template being newly added between class
Similarity forms new normal distribution, and the expectation of distribution map and variance change, to form new matching threshold T, with suitable
Answer the increase of iris enrollment.
Under the same application scene of the same terminal, false acceptance rate ε is substantially identical, with iris enrollment
The increase of quantity, similarity distribution map change, and matching threshold T also occurs to change accordingly；In different application scenarios,
Preset false acceptance rate ε is different, and obtained matching threshold T is also different, so, matching threshold T becomes with false acceptance rate ε dynamics
Change, in different actual scene applications, for example, under the application scenarios of terminal screen unlock, it is successful that user pursues unlock
Efficiency, hence, it can be determined that a lower matching threshold, iris recognition will be easier successful match, thus quick release,
Promote the experience sense of user；If application scenarios are in terms of mobile payment, the safety that user pursues fund therefore can be true
Surely a higher matching threshold is used, the feature that iris recognition compares is more, although being not easy successful match, can improve
The safety of fund.
The present invention is dynamically determined matching threshold according to the distribution situation and quantity of enrollment, can effectively avoid matching threshold
Value T is chronically at the false matching rate of lower state and lifting system, to effectively improve the safety of authentication result, also
It can effectively avoid matching threshold T and be chronically at higher state and the matching rate that reduces system, to effectively promote user experience
Sense.
Embodiment described above only expresses one or more embodiments of the present invention, and description is more specific and detailed
Carefully, but it cannot be construed as a limitation to the scope of the present invention.It should be pointed out that for the common skill of this field
For art personnel, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to this hair
Bright protection domain.
Claims (3)
1. a kind of choice of dynamical method of extensive iris feature identification matching threshold, which is characterized in that include the following steps：
S1. the similarity x between all iris enrollment classes is calculated；
S2. similarity forms similarity distribution map between counting the class in S1；
S3. according to preset false acceptance rate ε, the similarity distribution map in S2 is integrated, determines matching threshold T, just like
Lower relational expression：
Wherein, x is the expectation and variance yields of similarity, μ and σ for similarity distribution map；
S4. often increase N number of iris enrollment, repeat step S1-S3, obtain new similarity distribution map, it is new to obtain
Matching threshold T.
2. the choice of dynamical method of extensive iris feature identification matching threshold according to claim 1, which is characterized in that
Matching threshold T in step S3 is obtained by following steps：
S31. remember that similarity distribution map Normal Distribution, the normal distyribution function are：
Wherein, x is the expectation and variance yields of similarity, μ and σ for similarity distribution map；
S32. remember that preset false acceptance rate is ε, there is following relational expression：
S is obtained according to formula (2) solution；
S33. note matching threshold T is：T=s.
3. the choice of dynamical method of extensive iris feature identification matching threshold according to claim 1, which is characterized in that
Similarity is obtained by following steps between class described in step S1：
S11. iris enrollment is respectively F1 and F2 between remembering two classes, then normalization Hamming distance between the two is：
Wherein, XOR is exclusive or operator, and the length of F1 and F2 is identical, and Length (F1) is the length of F1；
S12. the similarity that iris enrollment between two classes is calculated according to the normalization Hamming distance in step S11 calculates public
Formula is as follows：
Sim (F1, F2)=1-HD (F1, F2),
Wherein, Sim (F1, F2) similarities between the class of two iris enrollments, HD (F1, F2) are two iris enrollments
Between normalization Hamming distance；
S13. step S11 and step S12 is repeated, until iris enrollment similarity calculation finishes between all classes two-by-two.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810389947.4A CN108596112A (en) | 2018-04-27 | 2018-04-27 | A kind of choice of dynamical method of extensive iris feature identification matching threshold |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810389947.4A CN108596112A (en) | 2018-04-27 | 2018-04-27 | A kind of choice of dynamical method of extensive iris feature identification matching threshold |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108596112A true CN108596112A (en) | 2018-09-28 |
Family
ID=63610498
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810389947.4A Pending CN108596112A (en) | 2018-04-27 | 2018-04-27 | A kind of choice of dynamical method of extensive iris feature identification matching threshold |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108596112A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113469015A (en) * | 2021-06-29 | 2021-10-01 | 浙江大华技术股份有限公司 | Face recognition method and device, electronic equipment and computer storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101894256A (en) * | 2010-07-02 | 2010-11-24 | 西安理工大学 | Iris identification method based on odd-symmetric 2D Log-Gabor filter |
| CN103136533A (en) * | 2011-11-28 | 2013-06-05 | 汉王科技股份有限公司 | Face recognition method and device based on dynamic threshold value |
| CN104537334A (en) * | 2014-07-08 | 2015-04-22 | 珠海易胜电子技术有限公司 | Method for improving iris recognition property in non-ideal environment |
| CN105868731A (en) * | 2016-04-15 | 2016-08-17 | 山西天地科技有限公司 | Binocular iris characteristic obtaining method, binocular iris characteristic obtaining device, identity identification method and identity identification system |
| CN105975960A (en) * | 2016-06-16 | 2016-09-28 | 湖北润宏科技有限公司 | Iris identification method based on texture-direction energy characteristic |
| CN106022032A (en) * | 2015-03-30 | 2016-10-12 | 欧姆龙株式会社 | Individual identification device, and identification threshold setting method |
| CN106096539A (en) * | 2016-06-07 | 2016-11-09 | 广东欧珀移动通信有限公司 | A kind of method carrying out bio-identification and terminal |
| EP3285206A1 (en) * | 2016-08-18 | 2018-02-21 | Fujitsu Limited | Evaluation device, evaluation method and evaluation program for biometric data |
-
2018
- 2018-04-27 CN CN201810389947.4A patent/CN108596112A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101894256A (en) * | 2010-07-02 | 2010-11-24 | 西安理工大学 | Iris identification method based on odd-symmetric 2D Log-Gabor filter |
| CN103136533A (en) * | 2011-11-28 | 2013-06-05 | 汉王科技股份有限公司 | Face recognition method and device based on dynamic threshold value |
| CN104537334A (en) * | 2014-07-08 | 2015-04-22 | 珠海易胜电子技术有限公司 | Method for improving iris recognition property in non-ideal environment |
| CN106022032A (en) * | 2015-03-30 | 2016-10-12 | 欧姆龙株式会社 | Individual identification device, and identification threshold setting method |
| CN105868731A (en) * | 2016-04-15 | 2016-08-17 | 山西天地科技有限公司 | Binocular iris characteristic obtaining method, binocular iris characteristic obtaining device, identity identification method and identity identification system |
| CN106096539A (en) * | 2016-06-07 | 2016-11-09 | 广东欧珀移动通信有限公司 | A kind of method carrying out bio-identification and terminal |
| CN105975960A (en) * | 2016-06-16 | 2016-09-28 | 湖北润宏科技有限公司 | Iris identification method based on texture-direction energy characteristic |
| EP3285206A1 (en) * | 2016-08-18 | 2018-02-21 | Fujitsu Limited | Evaluation device, evaluation method and evaluation program for biometric data |
Non-Patent Citations (8)
| Title |
|---|
| D. M. MONRO 等: "An effective human iris code with low complexity", 《IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING 2005》 * |
| J. DAUGMAN: "Probing the Uniqueness and Randomness of IrisCodes: Results From 200 Billion Iris Pair Comparisons", 《PROCEEDINGS OF THE IEEE》 * |
| JOHN G .DAUGMAN: "High Confidence Visual Recognition of Persons by a Test of Statistical Independence", 《IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE》 * |
| W. DONG 等: "Quality-based dynamic threshold for iris matching", 《2009 16TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)》 * |
| 柯呈通 等: "基于虹膜识别技术的身份验证模型", 《计算机时代》 * |
| 汪良会: "虹膜识别关键技术的研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 田启川: "《虹膜识别原理及算法》", 31 December 2010 * |
| 田启川: "不完美虹膜的定位分割、 特征提取与分类识别", 《中国优秀博硕士学位论文全文数据库 (博士) 信息科技辑》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113469015A (en) * | 2021-06-29 | 2021-10-01 | 浙江大华技术股份有限公司 | Face recognition method and device, electronic equipment and computer storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Sun et al. | Improving iris recognition accuracy via cascaded classifiers | |
| Tsai et al. | Iris recognition using possibilistic fuzzy matching on local features | |
| Hamd et al. | Biometric system design for iris recognition using intelligent algorithms | |
| Sundaram et al. | Neural network based Iris recognition system using Haralick features | |
| Latha et al. | Efficient approach to normalization of multimodal biometric scores | |
| Choudhary et al. | A survey: Feature extraction methods for iris recognition | |
| Zhang et al. | Palmprint verification using complex wavelet transform | |
| US9183440B2 (en) | Identification by iris recognition | |
| Harakannanavar et al. | An extensive study of issues, challenges and achievements in iris recognition | |
| Azizi et al. | Efficient iris recognition through improvement of feature extraction and subset selection | |
| Choras | Multimodal biometrics for person authentication | |
| CN108596112A (en) | A kind of choice of dynamical method of extensive iris feature identification matching threshold | |
| Ali et al. | Half iris Gabor based iris recognition | |
| Doublet et al. | Contactless hand recognition based on distribution estimation | |
| Habeeb | Comparison between physiological and behavioral characteristics of biometric system | |
| Chen et al. | Automatic online signature verification based on dynamic function features | |
| Lee et al. | A New Similarity Measure Based on Intraclass Statistics for Biometrie Systems | |
| Mohammed et al. | Developing iris recognition system based on enhanced normalization | |
| CN102147862A (en) | Face feature extracting method based on survival exponential entropy | |
| Li et al. | Multi-feature based score fusion method for fingerprint recognition accuracy boosting | |
| Proença et al. | A method for the identification of inaccuracies in pupil segmentation | |
| Murugan et al. | Feature extraction on half iris for personal identification | |
| Choi et al. | Facial fraud discrimination using detection and classification | |
| Tsai et al. | A new matching approach for local feature based iris recognition systems | |
| Telgad et al. | Development of an efficient secure biometric system by using iris, fingerprint, face |
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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 100081 Room 204, building 3, Daliushu Fuhai center, Haidian District, Beijing Applicant after: Beijing wanlihong Technology Co.,Ltd. Address before: 100081 Room 204, building 3, Daliushu Fuhai center, Haidian District, Beijing Applicant before: BEIJING SUPERRED TECHNOLOGY Co.,Ltd. |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180928 |