Pub Date : 2007-09-01DOI: 10.1109/BCC.2007.4430533
P. de Leon, A.L. Trevizo
Speaker identification (SI) systems based on Gaussian Mixture Models (GMMs) have demonstrated high levels of accuracy when both training and testing signals are acquired in near ideal conditions. These same systems when trained and tested with signals acquired under non-ideal channels such as telephone have been shown to have markedly lower accuracy levels. In this paper, we consider a reverberant test environment and its impact on SI. We measure the degradation in SI accuracy when the system is trained with clean signals but tested with reverberant signals. Next, we propose a method whereby training signals are first filtered with a family of reverberation filters prior to construction of speaker models; the reverberation filters are designed to approximate expected test room reverberation. Reverberant test signals are then scored against the family of speaker models and identification is made. Our research demonstrates that by approximating test room reverberation in the training signals, the channel mismatch problem can be reduced and SI accuracy increased.
{"title":"Speaker Identification in the Presence of Room Reverberation","authors":"P. de Leon, A.L. Trevizo","doi":"10.1109/BCC.2007.4430533","DOIUrl":"https://doi.org/10.1109/BCC.2007.4430533","url":null,"abstract":"Speaker identification (SI) systems based on Gaussian Mixture Models (GMMs) have demonstrated high levels of accuracy when both training and testing signals are acquired in near ideal conditions. These same systems when trained and tested with signals acquired under non-ideal channels such as telephone have been shown to have markedly lower accuracy levels. In this paper, we consider a reverberant test environment and its impact on SI. We measure the degradation in SI accuracy when the system is trained with clean signals but tested with reverberant signals. Next, we propose a method whereby training signals are first filtered with a family of reverberation filters prior to construction of speaker models; the reverberation filters are designed to approximate expected test room reverberation. Reverberant test signals are then scored against the family of speaker models and identification is made. Our research demonstrates that by approximating test room reverberation in the training signals, the channel mismatch problem can be reduced and SI accuracy increased.","PeriodicalId":389417,"journal":{"name":"2007 Biometrics Symposium","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131053424","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2007-09-01DOI: 10.1109/BCC.2007.4430544
P. D. Leon, V. Apsingekar
For large population speaker identification (SID) systems, likelihood computations between an unknown speaker's test feature set and speaker models can be very time-consuming and detrimental to applications where fast SID is required. In this paper, we propose a method whereby speaker models are clustered during the training stage. Then during the testing stage, only those clusters which are likely to contain high-likelihood speaker models are searched. The proposed method reduces the speaker model space which directly results in faster SID. Although there maybe a slight loss in identification accuracy depending on the number of clusters searched, this loss can be controlled by trading off speed and accuracy.
{"title":"Reducing Speaker Model Search Space in Speaker Identification","authors":"P. D. Leon, V. Apsingekar","doi":"10.1109/BCC.2007.4430544","DOIUrl":"https://doi.org/10.1109/BCC.2007.4430544","url":null,"abstract":"For large population speaker identification (SID) systems, likelihood computations between an unknown speaker's test feature set and speaker models can be very time-consuming and detrimental to applications where fast SID is required. In this paper, we propose a method whereby speaker models are clustered during the training stage. Then during the testing stage, only those clusters which are likely to contain high-likelihood speaker models are searched. The proposed method reduces the speaker model space which directly results in faster SID. Although there maybe a slight loss in identification accuracy depending on the number of clusters searched, this loss can be controlled by trading off speed and accuracy.","PeriodicalId":389417,"journal":{"name":"2007 Biometrics Symposium","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121502871","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2007-09-01DOI: 10.1109/BCC.2007.4430530
Yongjin Wang, K. Plataniotis
Changeability, privacy protection, and verification accuracy are important factors for widespread deployment of biometrics based authentication systems. In this paper, we introduce a method for effective combination of biometrics data with user specific secret key for human verification. The proposed approach is based on discretized random orthonormal transformation of biometrics features. It provides attractive properties of zero error rate, and generates revocable and non-invertible biometrics templates. In addition, we also present another scheme where no discretization procedure is involved. The proposed methods are well supported by mathematical analysis. The feasibility of the introduced solutions on a face verification problem is demonstrated using the well known ORL and GT database. Experimentation shows the effectiveness of the proposed methods comparing with existing works.
{"title":"Face Based Biometric Authentication with Changeable and Privacy Preservable Templates","authors":"Yongjin Wang, K. Plataniotis","doi":"10.1109/BCC.2007.4430530","DOIUrl":"https://doi.org/10.1109/BCC.2007.4430530","url":null,"abstract":"Changeability, privacy protection, and verification accuracy are important factors for widespread deployment of biometrics based authentication systems. In this paper, we introduce a method for effective combination of biometrics data with user specific secret key for human verification. The proposed approach is based on discretized random orthonormal transformation of biometrics features. It provides attractive properties of zero error rate, and generates revocable and non-invertible biometrics templates. In addition, we also present another scheme where no discretization procedure is involved. The proposed methods are well supported by mathematical analysis. The feasibility of the introduced solutions on a face verification problem is demonstrated using the well known ORL and GT database. Experimentation shows the effectiveness of the proposed methods comparing with existing works.","PeriodicalId":389417,"journal":{"name":"2007 Biometrics Symposium","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130126125","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2007-09-01DOI: 10.1109/BCC.2007.4430541
J. M. Pascual-Gaspar, Valentín Cardeñoso-Payo
In this paper we present a novel HMM-based automatic signature verification system where the number of states is estimated from the duration of the signatures. This structural user-dependent approach has allowed to obtain high verification rates with a small number of enrolment samples and using only the two basic local x-y geometric features plus their first time derivatives. The proposed system has been tested with the MCYT database reporting EERs of 2.09% with random forgeries and 6.14% with skilled forgeries using only three signatures for enrolment.
{"title":"On-Line Signature Verification Using Hidden Markov Models with Number of States Estimation from the Signature Duration","authors":"J. M. Pascual-Gaspar, Valentín Cardeñoso-Payo","doi":"10.1109/BCC.2007.4430541","DOIUrl":"https://doi.org/10.1109/BCC.2007.4430541","url":null,"abstract":"In this paper we present a novel HMM-based automatic signature verification system where the number of states is estimated from the duration of the signatures. This structural user-dependent approach has allowed to obtain high verification rates with a small number of enrolment samples and using only the two basic local x-y geometric features plus their first time derivatives. The proposed system has been tested with the MCYT database reporting EERs of 2.09% with random forgeries and 6.14% with skilled forgeries using only three signatures for enrolment.","PeriodicalId":389417,"journal":{"name":"2007 Biometrics Symposium","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126170001","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2007-09-01DOI: 10.1109/BCC.2007.4430551
J. Heo, M. Savvides
This paper presents how to remove unwanted occlusions such as eyeglasses in face images. By choosing training images carefully, we can derive a set of basis vectors that can eliminate this impediment for processing face images. In order to handle different poses, we apply multi-view active appearance models (MVAAMs) for fitting and then convert non-frontal images into frontal neutral faces. A set of these frontal-neutral faces is chosen for the basis and used for reconstructing other face images with occlusions. In addition, we are able to correct missing features while converting into frontal faces. The corrected faces are used for inputting to a frontal based face recognition system which can handle non-frontal faces efficiently.
{"title":"Face Pose Correction With Eyeglasses and Occlusions Removal","authors":"J. Heo, M. Savvides","doi":"10.1109/BCC.2007.4430551","DOIUrl":"https://doi.org/10.1109/BCC.2007.4430551","url":null,"abstract":"This paper presents how to remove unwanted occlusions such as eyeglasses in face images. By choosing training images carefully, we can derive a set of basis vectors that can eliminate this impediment for processing face images. In order to handle different poses, we apply multi-view active appearance models (MVAAMs) for fitting and then convert non-frontal images into frontal neutral faces. A set of these frontal-neutral faces is chosen for the basis and used for reconstructing other face images with occlusions. In addition, we are able to correct missing features while converting into frontal faces. The corrected faces are used for inputting to a frontal based face recognition system which can handle non-frontal faces efficiently.","PeriodicalId":389417,"journal":{"name":"2007 Biometrics Symposium","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114847316","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2007-09-01DOI: 10.1109/BCC.2007.4430546
J. Jeffers, A. Arakala
The fuzzy vault is an innovative cryptographic construct that uses error correction techniques to compensate for natural biometric variation. For fingerprints, the fuzzy vault can be used to compensate for the insertion and deletion of minutiae between samples, within the cryptographic framework. However, fingerprint biometrics also suffer from the problem that samples at enrolment and verification cannot be captured and recorded within a universally agreed frame of reference. There is currently no efficient fingerprint pre-alignment technique that also protects the template. In this paper we propose a pre-alignment algorithm that incorporates quantifiable template protection and explore the suitability of three minutiae-based structures for the algorithm. We find that one of the structures is strongly suitable with respect to the goals of our pre-alignment algorithm and its impact on the false non-match rate of an overall system is quantified. Our research also clarifies the key characteristics required from minutiae-based structures for high performance.
{"title":"Fingerprint Alignment for A Minutiae-Based Fuzzy Vault","authors":"J. Jeffers, A. Arakala","doi":"10.1109/BCC.2007.4430546","DOIUrl":"https://doi.org/10.1109/BCC.2007.4430546","url":null,"abstract":"The fuzzy vault is an innovative cryptographic construct that uses error correction techniques to compensate for natural biometric variation. For fingerprints, the fuzzy vault can be used to compensate for the insertion and deletion of minutiae between samples, within the cryptographic framework. However, fingerprint biometrics also suffer from the problem that samples at enrolment and verification cannot be captured and recorded within a universally agreed frame of reference. There is currently no efficient fingerprint pre-alignment technique that also protects the template. In this paper we propose a pre-alignment algorithm that incorporates quantifiable template protection and explore the suitability of three minutiae-based structures for the algorithm. We find that one of the structures is strongly suitable with respect to the goals of our pre-alignment algorithm and its impact on the false non-match rate of an overall system is quantified. Our research also clarifies the key characteristics required from minutiae-based structures for high performance.","PeriodicalId":389417,"journal":{"name":"2007 Biometrics Symposium","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129201272","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2007-09-01DOI: 10.1109/BCC.2007.4430553
Jinyu Zuo, N. Schmid
Daugman's iris recognition algorithm introduced in early 90s and later undergoing continuous refinements remains potentially the most efficient and scalable in iris field. The encoding part of the algorithm relies on application of Gabor wavelets that in terms of their imaging capabilities mimic capabilities of human eye receptor field. In this work, we design and test an algorithm that can be used both individually and as a natural extension scheme to Gabor wavelet-based algorithm. It is based on the local ordinal information extracted from original unfiltered images. This scheme holds a number of promises: (1) it is robust with respect to a number of nonidealities in iris images and (2) because of the binary nature of the local ordinal information this scheme can be flawlessly integrated into the traditional filter-based recognition systems. The proposed scheme was extensively tested individually and when combined with Gabor wavelet-based approach.
{"title":"On a Local Ordinal Binary Extension to Gabor Wavelet-Based Encoding for Improved Iris Recognition","authors":"Jinyu Zuo, N. Schmid","doi":"10.1109/BCC.2007.4430553","DOIUrl":"https://doi.org/10.1109/BCC.2007.4430553","url":null,"abstract":"Daugman's iris recognition algorithm introduced in early 90s and later undergoing continuous refinements remains potentially the most efficient and scalable in iris field. The encoding part of the algorithm relies on application of Gabor wavelets that in terms of their imaging capabilities mimic capabilities of human eye receptor field. In this work, we design and test an algorithm that can be used both individually and as a natural extension scheme to Gabor wavelet-based algorithm. It is based on the local ordinal information extracted from original unfiltered images. This scheme holds a number of promises: (1) it is robust with respect to a number of nonidealities in iris images and (2) because of the binary nature of the local ordinal information this scheme can be flawlessly integrated into the traditional filter-based recognition systems. The proposed scheme was extensively tested individually and when combined with Gabor wavelet-based approach.","PeriodicalId":389417,"journal":{"name":"2007 Biometrics Symposium","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122844307","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2007-07-01DOI: 10.1109/BCC.2007.4430545
I. Nakanishi, Y. Yorikane, Y. Itoh, Y. Fukui
We propose to utilize an electromagnetic wave through a human body as biometrics. The electromagnetic wave (intra-body propagation signal) is generated at a relatively shallow depth in the human body through a pair of electrodes pasted on the human skin. The biological tissue of each individual human being is different from that of others, so that the transfer characteristic of the intra-body propagation signal is also different mutually. By using such a difference, it is expected to authenticate personal identification. In addition, liveness detection can be realized simultaneously using the intra-body propagation signal. It is effective on the detection of spoofing using artificial bodies. In this paper, we examine the individual feature in the intra-body propagation signal based on the spectrum analysis. As a result, the verification rate of 58% is obtained using the similarity of the power spectrum especially in the 30-60 MHz sub-band.
{"title":"Biometric Identity Verification Using Intra-Body Propagation Signal","authors":"I. Nakanishi, Y. Yorikane, Y. Itoh, Y. Fukui","doi":"10.1109/BCC.2007.4430545","DOIUrl":"https://doi.org/10.1109/BCC.2007.4430545","url":null,"abstract":"We propose to utilize an electromagnetic wave through a human body as biometrics. The electromagnetic wave (intra-body propagation signal) is generated at a relatively shallow depth in the human body through a pair of electrodes pasted on the human skin. The biological tissue of each individual human being is different from that of others, so that the transfer characteristic of the intra-body propagation signal is also different mutually. By using such a difference, it is expected to authenticate personal identification. In addition, liveness detection can be realized simultaneously using the intra-body propagation signal. It is effective on the detection of spoofing using artificial bodies. In this paper, we examine the individual feature in the intra-body propagation signal based on the spectrum analysis. As a result, the verification rate of 58% is obtained using the similarity of the power spectrum especially in the 30-60 MHz sub-band.","PeriodicalId":389417,"journal":{"name":"2007 Biometrics Symposium","volume":"150 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115590884","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：