KR20040028854A - Verification of Iris Patterns Based on Variable Threshold - Google Patents

Verification of Iris Patterns Based on Variable Threshold Download PDF

Info

Publication number
KR20040028854A
KR20040028854A KR1020040006503A KR20040006503A KR20040028854A KR 20040028854 A KR20040028854 A KR 20040028854A KR 1020040006503 A KR1020040006503 A KR 1020040006503A KR 20040006503 A KR20040006503 A KR 20040006503A KR 20040028854 A KR20040028854 A KR 20040028854A
Authority
KR
South Korea
Prior art keywords
iris
characteristic
image
pattern
similarity
Prior art date
Application number
KR1020040006503A
Other languages
Korean (ko)
Other versions
KR100602526B1 (en
Inventor
조성원
김재민
Original Assignee
김재민
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 김재민 filed Critical 김재민
Priority to KR1020040006503A priority Critical patent/KR100602526B1/en
Publication of KR20040028854A publication Critical patent/KR20040028854A/en
Application granted granted Critical
Publication of KR100602526B1 publication Critical patent/KR100602526B1/en

Links

Classifications

    • GPHYSICS
    • G02OPTICS
    • G02CSPECTACLES; SUNGLASSES OR GOGGLES INSOFAR AS THEY HAVE THE SAME FEATURES AS SPECTACLES; CONTACT LENSES
    • G02C11/00Non-optical adjuncts; Attachment thereof
    • G02C11/02Ornaments, e.g. exchangeable
    • GPHYSICS
    • G02OPTICS
    • G02CSPECTACLES; SUNGLASSES OR GOGGLES INSOFAR AS THEY HAVE THE SAME FEATURES AS SPECTACLES; CONTACT LENSES
    • G02C5/00Constructions of non-optical parts
    • G02C5/008Spectacles frames characterized by their material, material structure and material properties

Landscapes

  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Ophthalmology & Optometry (AREA)
  • Optics & Photonics (AREA)
  • Collating Specific Patterns (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

PURPOSE: A system for verifying an iris pattern in real-time using a variable threshold critical value is provided to lower a false accept rate and a false reject rate by applying a different threshold value according to a characteristic of the iris pattern. CONSTITUTION: Some iris images including an iris area are obtained from a user through a camera. An optimal image suitable for iris recognition is selected by examining a quality of the obtained image. The iris area is extracted from the selected image and is processed to the image suitable for characteristic extraction. A characteristic value is extracted from the processed iris image through a transform process. A characteristic vector is generated by quantizing the characteristic value and a similarity between the characteristic vector and the characteristic vector stored in a database is calculated. If the similarity is larger than the threshold value stored in a characteristic code, the iris is verified at the same iris. If not, the iris is rejected.

Description

가변 임계치을 이용한 실시간 홍채패턴 검증 시스템{Verification of Iris Patterns Based on Variable Threshold}Real-time Iris Pattern Verification System Using Variable Thresholds {Verification of Iris Patterns Based on Variable Threshold}

인간이 지닌 생물학적 특징을 정보로 활용하여 개인의 신원을 판별하고 그 밖에 여러 분야에 응용하려는 기술 개발에 있어 주요 과정은 (1)검증하고자 하는 홍채를 포함하는 눈의 영상을 획득하는 과정, (2)획득한 영상으로부터 홍채영역을 추출하고 추출된 화질을 개선하고 변환하여 홍채특징을 효과적으로 추출할 수 있게 하는 전처리 과정, (3)전처리된 홍채패턴으로부터 특징 코드를 형성하는 과정, (4)추출된 특징 코드와 등록된 특징 코드를 비교 검증하는 과정으로 구성되어있다. 본 연구는 특징 코드와 등록된 특징 코드를 비교 검증과정을 개선하여 오인식률과 오거부율을 낮추어 검증의 정확성을 개선하는 것을 목적으로 한다.The main process in the development of technology to identify the individual's identity and apply it to various other fields by utilizing human's biological characteristics as information is (1) the process of acquiring the image of the eye including the iris to be verified, (2 Preprocessing to extract the iris region from the acquired image and to improve and convert the extracted image quality to effectively extract the iris features, (3) forming a feature code from the preprocessed iris pattern, (4) extracted It consists of comparing and verifying feature codes and registered feature codes. The purpose of this study is to improve the accuracy of verification by reducing the false recognition rate and false rejection rate by improving the verification process of feature code and registered feature code.

본 발명이 속하는 기술 분야는 영상처리와 패턴인식 기술을 생리학, 생물학적 기술에 응용한 생체인식 기술 분야이며, 그중에서도 수정체의 이완 수축을 담당하는 근육체인 홍채의 패턴을 검증과 관련된 기술이다.TECHNICAL FIELD The present invention belongs to the field of biometric technology in which image processing and pattern recognition technology is applied to physiology and biological technology, and among them, a technology related to verifying the pattern of an iris, which is a muscle body responsible for relaxation and contraction of the lens.

홍채패턴의 검증은 검증하고자 하는 홍채를 포함하는 눈의 영상으로부터 홍채영역을 추출하고 추출된 화질을 개선하고 변환하여 홍채특징을 효과적으로 추출할 수 있게 하는 전처리를 하고, 전처리된 홍채패턴으로부터 특징 코드를 추출하고, 추출된 특징 코드와 등록된 특징 코드를 비교하는 것이다.The verification of the iris pattern is performed by preprocessing to extract the iris region from the image of the eye including the iris to be verified, to improve and convert the extracted image quality, and to effectively extract the iris features, and to extract the feature code from the preprocessed iris pattern. The extracted feature code is compared with the registered feature code.

홍채패턴의 검증은 수백 혹은 수백만 명의 다양한 홍채 패턴을 검증에 적용되는 기술로써 효과적인 실시간 검증을 위하여 검증하고자 하는 홍채 패턴과 데이터베이스에 저장된 홍채패턴간의 유사도가 학습과정에서 정한 임계치(threshold value)보다 크면 동일한 홍채패턴으로 검증하는 방법을 사용하고 있다. 이는 동일한 홍채패턴이라 할지라도 환경의 변화에 따른 각종 잡영(noise)으로 인하여 두 특징 코드가 일치하지 않으나, 효과적인 전처리와 특징코드 추출을 통하여 동일한 홍채패턴 간에는 높은 유사도를 가지게 되며, 반면에 홍채패턴의 특성으로 인하여 다른 홍채패턴 간에는 낮은 유사도를 가지는 현상에 바탕을 두고 있다.The verification of the iris pattern is a technique applied to verifying hundreds or millions of various iris patterns. If the similarity between the iris pattern to be verified for effective real-time verification and the iris pattern stored in the database is greater than the threshold value defined in the learning process, It uses a method of verifying with an iris pattern. Although the same iris pattern does not match the two feature codes due to various noises due to environmental changes, the same iris pattern has a high similarity between the iris patterns through effective preprocessing and feature code extraction. Due to its characteristics, it is based on the phenomenon of low similarity between different iris patterns.

학습과정에서 임계치의 설정은 다양한 홍채패턴에서 다양한 잡영을 가지는 동일한 홍채패턴간의 유사도 분포와 다른 홍채패턴간의 유사도의 분포를 계산하고, 두 유사도 분포로부터 오인식률(false accept rate)와 오거부율(false reject rate)의 가중치합을 최소로 하는 임계치을 설정한다.In the learning process, the threshold is calculated by calculating the similarity distribution between the same iris patterns and the iris patterns having different noises in various iris patterns, and the false accept rate and false reject rate from the two similarity distributions. Sets a threshold that minimizes the sum of weights of rates.

다양한 패턴 추출 방법과 유사도 계산 방법을 사용하고 있으나, 모든 패턴의 오인식률 및 오거부률의 가중치 합을 최소로 하는 하나의 임계치를 모든 홍채패턴의 검증에 적용하기 때문에 정확도가 상대적으로 낮다. 반면에 적은 수의 패턴 분류에서 사용되는 신경회로망(neural network), 스포터벡터머신(support vector machine), 가우시안 혼합모델(Gaussian mixture model)에 기반한 베이의 분류기(Bayes classification)등은 수백 혹은 수백만 명의 홍채패턴의 실시간 검증에는 사용할 수 없다. 때문에 실시간으로 다양한 홍채패턴의 검증에 효과적인 기술 방법의 개발이 필요하다.Although various pattern extraction methods and similarity calculation methods are used, the accuracy is relatively low because one threshold is applied to the verification of all iris patterns to minimize the sum of the weights of the false recognition rate and the rejection rate of all patterns. On the other hand, neural networks, support vector machines, and Bayes classifications based on a Gaussian mixture model, which are used in a small number of pattern classifications, are used by hundreds or millions of people. It cannot be used for real-time verification of iris patterns. Therefore, it is necessary to develop an effective technical method for verifying various iris patterns in real time.

본 발명은 (1)각 홍채 패턴을 검증함에 있어 홍채 패턴의 특징에 따라 다른 임계 값(threshold value)을 적용하여 오인식률(false accept rate)과 오거부률(false reject rate)를 낮추는 기술 기반을 제공하는 것이다. (2) 가변 임계 값을 적용할 수 있게 생체패턴의 등록 단계에서 등록하는 패턴의 잡영에 따른 유사도 분포와 등록하는 패턴과 기존 저장된 패턴간의 유사도 분포를 계산하여 함수로 표시하고 이를 바탕으로 최적의 임계값을 계산하는 것이다. (3) 새로운 패턴이 등록될 때 마다 패턴 상호간의 유사도 분포를 갱신하고 이를 바탕으로 각 패턴에따라 설정된 임계치를 갱신하는 것이다.The present invention (1) in verifying each iris pattern by applying a different threshold value in accordance with the characteristics of the iris pattern to reduce the false acceptance rate (false accept rate) and false reject rate (false reject rate) To provide. (2) In order to apply the variable threshold value, the similarity distribution according to the miscellaneous pattern registration in the registration step of the biometric pattern and the similarity distribution between the registered pattern and the existing stored pattern are calculated and displayed as a function and based on the optimal threshold To calculate the value. (3) Each time a new pattern is registered, the similarity distribution between the patterns is updated, and based on this, the threshold set according to each pattern is updated.

제 1 도는 가변 임계치를 이용한 홍채검증시스템의 등록과정 흐름도1 is a flow chart of the registration process of an iris verification system using a variable threshold

제 2 도는 가변 임계치를 이용한 홍채검증시스템의 검정과정 흐름도2 is a flow chart of the verification process of the iris verification system using a variable threshold

제 3 도는 홍채인식 시스템을 위한 영상획득 장비의 구성도3 is a block diagram of the image acquisition equipment for the iris recognition system

제 1 도는 가변 임계치를 이용한 홍채검증시스템의 등록과정의 흐름도이다.1 is a flowchart of a registration process of an iris verification system using a variable threshold.

등록 과정은 다음과 같다.The registration process is as follows.

1) 획득된 영상의 품질을 검사하여 홍채 인식에 적합한 여러 장(N개) 영상을 획득할 때까지 영상을 계속하여 획득한다.1) The quality of the acquired image is examined and the image is continuously acquired until several (N) images suitable for iris recognition are obtained.

2) 선택된 영상에서 영상처리과정을 거쳐 홍채영역을 추출하고, 특징추출에 적합하게 영상을 변환키는 전처리과정을 거친다.2) The iris region is extracted from the selected image through image processing, and the image is transformed according to the feature extraction.

3) 전처리 과정을 거친 홍채영상은 가버변환, 웨이블릿변화. 혹은 라플라시안 변환을 수행하여 특징값을 추출한다.3) The iris image after preprocessing is Gabor transform, wavelet change. Alternatively, the Laplacian transform is performed to extract feature values.

4) 추출된 특징값은 양자화하여 특징벡터를 생성한다.4) The extracted feature values are quantized to generate feature vectors.

5) 2)~4)의 과정을 거쳐 형성된 N개의 특징코드로부터N 2개의 유사도를 계산하고 유사도 분포를 일반화된 가우시안 함수(generalized Gaussian distribution)함수로 모델링한다. 기존 데이터 베이스에 저장된 M개의 홍채 특징 코드와N×M개의 유사도를 계산하고 유사도 분포를 일반화된 가우시안 함수(generalized Gaussian distribution)함수로 모델링한다. 두 유사도 분포함수를 이용하여 오인식률(false accept rate)를 만족하면서 오차를 최소로하는 임계치를 설정한다. 현재 등록하는 홍채의 동일패턴간의N 2개의 유사도를 나타내는 함수의 변수 값을 양자화하여 등록코드에 포함시킨다.5) Calculate N 2 similarities from N feature codes formed through 2) ~ 4) and model the similarity distribution as a generalized Gaussian distribution function. M iris feature codes and N × M similarities stored in the existing database are calculated and the similarity distribution is modeled as a generalized Gaussian distribution function. Two similarity distribution functions are used to set a threshold that minimizes errors while satisfying a false accept rate. By quantizing the value of the variable of a function that represents the N 2 of the same degree of similarity between the pattern of the iris that are currently registered to include the registration code.

제 2 도는 가변 임계치를 이용한 홍채검증시스템의 검증과정의 흐름도이다.2 is a flowchart of a verification process of an iris verification system using a variable threshold.

검증과정은 다음과 같다.The verification process is as follows.

1) 영상획득 장비를 통해 사용자로부터 홍채영역이 포함된 눈 영상 여러 장을 획득한다.1) Acquire several eye images including iris area from user through image acquisition equipment.

2) 획득된 영상의 품질을 검사하여 홍채 인식에 적합한 최적의 영상을 선택한다. 선택된 영상에서 영상처리과정을 거쳐 홍채영역을 추출하고, 특징추출에 적합하게 영상을 변환키는 전처리 과정을 거친다.2) Check the quality of the acquired image and select the optimal image suitable for iris recognition. The iris region is extracted from the selected image through image processing, and the image is transformed to be suitable for feature extraction.

3) 전처리 과정을 거친 홍채영상은 가버변환, 웨이블릿변화. 혹은 라플라시안 변환을 수행하여 특징값을 추출한다.3) The iris image after preprocessing is Gabor transform, wavelet change. Alternatively, the Laplacian transform is performed to extract feature values.

4) 추출된 특징값은 양자화하여 특징벡터를 생성하고, 기존의 데이터베이스에 저장된 특징벡터와 유사도를 계산한다. 계산된 유사도가 특징 코드에 저장된 임계치보다 크면 동일 홍채로 검증되고, 유사도가 임계치보다 작으면 거부된다.4) The extracted feature values are quantized to generate a feature vector, and the similarity with the feature vector stored in the existing database is calculated. If the calculated similarity is greater than the threshold stored in the feature code, it is verified with the same iris; if the similarity is less than the threshold, it is rejected.

오프라인 유사도 갱신과정은 실시간 등록 및 검증을 위하여 새로운 패턴을 등록하거나 검증할 경우에는 제 1도 혹은 제 2도의 과정에 따라 등록 및 검증을 하고, 등록 및 검증이 끝나면 데이터베이스에 저장된 다른 종류 홍채패턴간의 유사도를 계산하고(각 홍채당 1개의 특징 코드벡터가 저장), 이를 바탕으로 각 홍채 패턴과 다른 홍채패턴들 간의 유사도 분포를 나타내는 함수를 갱신한다. 이를 바탕으로 각 홍채마다 설정된 임계치를 갱신하다.The offline similarity update process registers and verifies according to the process of FIG. 1 or 2 when registering or verifying a new pattern for real-time registration and verification, and the similarity between different kinds of iris patterns stored in the database after registration and verification is completed. (1 feature code vector is stored for each iris), and based on this, a function representing a similarity distribution between each iris pattern and other iris patterns is updated. Based on this, the threshold set for each iris is updated.

본 발명의 결과로 획득 가능한 효과는, 각 홍채 패턴을 검증함에 있어 홍채 패턴의 특징에 따라 다른 임계 값(threshold value)을 적용하여 오인식률(false accept rate) 및 오거부률(false reject rate)를 낮추어 홍채검증의 정확성을 높이는 기술 기반을 제공하는 것이다.The effect that can be obtained as a result of the present invention is that, in verifying each iris pattern, different threshold values are applied according to the characteristics of the iris pattern so as to obtain false accept rate and false reject rate. Lower technology provides a technology foundation that increases the accuracy of iris verification.

Claims (4)

가변 임계치를 이용한 홍채패턴을 검증하는 방법.Method of verifying iris pattern using variable threshold. 가변 임계치를 이용한 홍채패턴 검증 방법을 위한 등록 절차과정Registration procedure for iris pattern verification method using variable threshold 가변 임계치를 이용한 홍채패턴 검증 방법을 위한 검정 절차과정Test procedure for iris pattern verification method using variable threshold 오프라인에서 유사도 분포함수를 갱신하고 이를 바탕으로 각 홍채에 설정된 임계치 값을 갱신하는 방법.A method of updating a similarity distribution function offline and updating a threshold value set for each iris based on this.
KR1020040006503A 2004-01-31 2004-01-31 Verification of Iris Patterns Based on Variable Threshold KR100602526B1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
KR1020040006503A KR100602526B1 (en) 2004-01-31 2004-01-31 Verification of Iris Patterns Based on Variable Threshold

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
KR1020040006503A KR100602526B1 (en) 2004-01-31 2004-01-31 Verification of Iris Patterns Based on Variable Threshold

Publications (2)

Publication Number Publication Date
KR20040028854A true KR20040028854A (en) 2004-04-03
KR100602526B1 KR100602526B1 (en) 2006-08-22

Family

ID=37330730

Family Applications (1)

Application Number Title Priority Date Filing Date
KR1020040006503A KR100602526B1 (en) 2004-01-31 2004-01-31 Verification of Iris Patterns Based on Variable Threshold

Country Status (1)

Country Link
KR (1) KR100602526B1 (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100716743B1 (en) * 2004-12-10 2007-05-14 주식회사 아이레보 Fingerprint Authentication System And Method Capable of Adjusting Authentication Reference Level Based on The Number Of Enrolled Fingerprint
KR100778891B1 (en) * 2006-04-05 2007-11-22 (주)니트 젠 Method and apparatus for dynamically changing matching score threshold value in biomass authentication
US8242881B2 (en) 2007-03-28 2012-08-14 Fujitsu Limited Method of adjusting reference information for biometric authentication and apparatus
CN105279492A (en) * 2015-10-22 2016-01-27 北京天诚盛业科技有限公司 Iris identification method and device

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20030070183A (en) * 2002-02-21 2003-08-29 엘지전자 주식회사 Iris recording and recognition method of forged distingction of iris recognition system
KR100933256B1 (en) * 2007-11-27 2009-12-22 주식회사 유니온커뮤니티 Fingerprint registration and authentication method
JP5662157B2 (en) * 2007-12-20 2015-01-28 コーニンクレッカ フィリップス エヌ ヴェ Definition of classification threshold in template protection system
KR101030652B1 (en) * 2008-12-16 2011-04-20 아이리텍 잉크 An Acquisition System and Method of High Quality Eye Images for Iris Recognition
KR101958198B1 (en) * 2015-01-16 2019-03-15 한국정보통신주식회사 Apparatus and method for recoding iris information
KR102215484B1 (en) * 2020-06-23 2021-02-10 주식회사 카카오뱅크 Method for setting threshold value for identity authentication and system for the method

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100716743B1 (en) * 2004-12-10 2007-05-14 주식회사 아이레보 Fingerprint Authentication System And Method Capable of Adjusting Authentication Reference Level Based on The Number Of Enrolled Fingerprint
KR100778891B1 (en) * 2006-04-05 2007-11-22 (주)니트 젠 Method and apparatus for dynamically changing matching score threshold value in biomass authentication
US8242881B2 (en) 2007-03-28 2012-08-14 Fujitsu Limited Method of adjusting reference information for biometric authentication and apparatus
CN105279492A (en) * 2015-10-22 2016-01-27 北京天诚盛业科技有限公司 Iris identification method and device
CN105279492B (en) * 2015-10-22 2018-11-09 北京天诚盛业科技有限公司 The method and apparatus of iris recognition

Also Published As

Publication number Publication date
KR100602526B1 (en) 2006-08-22

Similar Documents

Publication Publication Date Title
Yuan et al. Fingerprint liveness detection using an improved CNN with image scale equalization
CN107145842B (en) Face recognition method combining LBP characteristic graph and convolutional neural network
KR20160098581A (en) Method for certification using face recognition an speaker verification
CN113221086B (en) Offline face authentication method and device, electronic equipment and storage medium
Dey et al. Improved feature processing for iris biometric authentication system
KR100602526B1 (en) Verification of Iris Patterns Based on Variable Threshold
Lei et al. Finger-vein image enhancement based on pulse coupled neural network
Setiawan et al. Biometric recognition based on palm vein image using learning vector quantization
Ali et al. Half iris Gabor based iris recognition
Tallapragada et al. Iris recognition based on combined feature of GLCM and wavelet transform
Latha et al. A novel method for person authentication using retinal images
Pandey et al. IRIS based human identification: Analogizing and exploiting PSNR and MSE techniques using MATLAB
Thiyaneswaran et al. Iris Recognition using Left and Right Iris Feature of the Human Eye for Biometric Security System
CN106940786B (en) Iris reconstruction method using iris template based on LLE and PSO
Choudhary et al. A Statistical Approach for Iris Recognition Using K-NN Classifier
CN111898533B (en) Gait classification method based on space-time feature fusion
Elangovan et al. A review: person identification using retinal fundus images
Ganorkar et al. Iris recognition system and analysis using neural networks
Chirchi et al. Modified circular fuzzy segmentor and local circular encoder to iris segmentation and recognition
Radouane et al. Fusion of Gabor filter and steerable pyramid to improve iris recognition system
Wibawa et al. Iris Grid Image Classification using Naive Bayes for Human Biometric System
Sarhan A WPD scanning technique for iris recognition
Tayal et al. A multimodal biometric authentication system using decision theory, iris and speech recognition
Zhou et al. Eye localization based on face alignment
Gupta et al. Iris recognition system for smart environments

Legal Events

Date Code Title Description
A201 Request for examination
N231 Notification of change of applicant
N231 Notification of change of applicant
E902 Notification of reason for refusal
E701 Decision to grant or registration of patent right
GRNT Written decision to grant
FPAY Annual fee payment

Payment date: 20120810

Year of fee payment: 7

FPAY Annual fee payment

Payment date: 20130529

Year of fee payment: 8

FPAY Annual fee payment

Payment date: 20150810

Year of fee payment: 10

FPAY Annual fee payment

Payment date: 20160809

Year of fee payment: 11

FPAY Annual fee payment

Payment date: 20170809

Year of fee payment: 12

FPAY Annual fee payment

Payment date: 20180809

Year of fee payment: 13

FPAY Annual fee payment

Payment date: 20190809

Year of fee payment: 14