Back to iPRoBE homepage

We are grateful to the following agencies for supporting our research work.

NSF CAREER Award (Program Manager: Maria Zemankova)

• Iris:
  • Multispectral Iris Analysis
  • Iris Segmentation
  • Generating Synthetic Irises
• Fingerprint:
  • Reconstructing Fingerprints From Minutiae
  • Understanding Non-linear Distortions in Fingerprints
  • Fingerprint Sensor Interoperability
  • Hybrid Fingerprint Matcher
  • Fingerprint Mosaicking
• Fusion:
  • Multibiometrics
  • Learning User-specific Parameters in Multibiometrics
  • Score Normalization in Multibiometrics
  • Feature Level Fusion in Biometrics
• Face:
  • Face Mosaicing
  • Texture-based Approach to Face Detection
• Automatic Biometric Template Selection
• Hand Geometry
• Models For User Access Patterns on the Web
• Modeling Saccadic Movement of the Human Eye
• Publications

Iris Segmentation The richness and the stability of the iris texture make it a robust biometric trait for personal authentication. The performance of an automated iris recognition system is affected by the accuracy of the segmentation process used to localize the iris structure. Most segmentation models in the literature assume that the pupillary, limbic and eyelid boundaries are circular or elliptical in shape. Hence, they focus on determining model parameters that best fit these hypotheses. However, it is difficult to segment iris images acquired under non-ideal conditions using such conic models. In this paper, we describe a novel iris segmentation scheme employing Geodesic Active Contours (GAC) to extract the iris from the surrounding structures. Since active contours can (a) assume any shape and (b) segment multiple objects simultaneously, they mitigate some of the concerns associated with the traditional models. The proposed scheme elicits the iris texture in an iterative fashion and is guided by both local and global properties of the image. The matching accuracy of an iris recognition system is observed to improve upon application of the proposed segmentation algorithm. Experimental results on multiple iris databases indicate the efficacy of the proposed technique. Papers: S. Shah and A. Ross, "Iris Segmentation Using Geodesic Active Contours," IEEE Transactions on Information Forensics and Security (TIFS), Vol. 4, Issue 4, pp. 824-836, December 2009. A. Ross and S. Shah, "Segmenting Non-ideal Irises Using Geodesic Active Contours," Proc. of Biometrics Symposium (BSYM), (Baltimore, USA), September 2006.

Generating Synthetic Irises We propose a technique to create digital renditions of iris images that can be used to evaluate the performance of iris recognition algorithms. The proposed scheme is implemented in two stages. In the first stage, a Markov Random Field model is used to generate a background texture representing the global iris appearance. In the next stage a variety of iris features, viz., radial and concentric furrows, collarette and crypts, are generated and embedded in the texture field. The iris images synthesized in this manner are observed to bear close resemblance to real irises. Experiments confirm the potential of this scheme to generate a database of synthetic irises that can be used to evaluate iris recognition algorithms. Papers: S. Shah and A. Ross, " Generating Synthetic Irises By Feature Agglomeration," Proc. of IEEE International Conference on Image Processing (ICIP), (Atlanta, USA), October 2006. S. Makthal and A. Ross, "Synthesis of Iris Images using Markov Random Fields", Proc. of 13th European Signal Processing Conference (EUSIPCO), (Antalya, Turkey), September 2005.

Reconstructing Fingerprints From Minutiae Most fingerprint-based biometric systems store the minutiae template of a user in the database. It has been traditionally assumed that the minutiae template of a user does not reveal any information about the original fingerprint. In this paper, we challenge this notion and show that three levels of information about the parent fingerprint can be elicited from the minutiae template alone, viz., 1) the orientation field information, 2) the class or type information, and 3) the friction ridge structure. The orientation estimation algorithm determines the direction of local ridges using the evidence of minutiae triplets. The estimated orientation field, along with the given minutiae distribution, is then used to predict the class of the fingerprint. Finally, the ridge structure of the parent fingerprint is generated using streamlines that are based on the estimated orientation field. Line Integral Convolution is used to impart texture to the ensuing ridges, resulting in a ridge map resembling the parent fingerprint. The salient feature of this noniterative method to generate ridges is its ability to preserve the minutiae at specified locations in the reconstructed ridge map. Experiments using a commercial fingerprint matcher suggest that the reconstructed ridge structure bears close resemblance to the parent fingerprint. Papers: A. Ross, J. Shah and A. K. Jain, "From Template to Image: Reconstructing Fingerprints From Minutiae Points," IEEE Transactions on Pattern Analysis and Machine Intelligence, Special Issue on Biometrics, Vol. 29, No. 4, pp. 544-560, April 2007. A. Ross, J. Shah, and A. K. Jain, "Towards Reconstructing Fingerprints from Minutiae Points", Proc. of SPIE Conference on Biometric Technology for Human Identification II, (Orlando, USA), pp. 68-80, March 2005.

Understanding Non-linear Distortions in Fingerprints Fingerprint matching is affected by the nonlinear distortion introduced in fingerprint impressions during the image acquisition process. This nonlinear deformation causes fingerprint features such as minutiae points and ridge curves to be distorted in a complex manner. In this work we develop an average deformation model for a fingerprint impression (baseline impression) by observing its relative distortion with respect to several other impressions of the same finger. The deformation is computed using a Thin Plate Spline (TPS) model that relies on minutiae and ridge curve correspondences between image pairs. The estimated average deformation is used to distort the minutiae template of the baseline impression prior to matching. An index of deformation has been proposed to select the average deformation model with the least variability corresponding to a finger. Preliminary results indicate that the average deformation model can improve the matching performance of a fingerprint matcher. Papers: A. Ross, S. Dass and A. K. Jain, "Fingerprint Warping Using Ridge Curve Correspondences", IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 28, No. 1, pp. 19-30, January 2006. Y. Chen, S. Dass, A. Ross and A. K. Jain, "Fingerprint Deformation Models Using Minutiae Locations and Orientations", Proc. of IEEE Workshop on Applications of Computer Vision (WACV), (Colorado, USA), pp. 150-155, January 2005. A. Ross, S. Dass and A. K. Jain, "A Deformable Model for Fingerprint Matching", Pattern Recognition, Vol. 38, No. 1, pp. 95-103, January 2005. A. Ross, S. Dass and A. K. Jain, "Estimating Fingerprint Deformation", Proc. of International Conference on Biometric Authentication (ICBA), (Hong Kong), LNCS vol. 3072, pp. 249-255, Springer Publishers, July 2004.

Fingerprint Sensor Interoperability Biometric sensor interoperability refers to the ability of a system to compensate for the variability introduced in the biometric data of an individual due to the deployment of different sensors. Poor inter-sensor performance has been reported in different biometric domains including fingerprint, face, iris and speech. In the context of fingerprints, variations are observed in the acquired images due to differences in sensor resolution, scanning area, sensing technology, etc. which impact the feature set extracted from these images. The inability of a fingerprint matcher to compensate for these variations introduced by different sensors results in inferior inter-sensor performance. In this work it is demonstrated that a simple non-linear calibration scheme, based on Thin Plate Splines (TPS), is sufficient to facilitate sensor interoperability in the context of fingerprints. In the proposed technique, the geometric variation between the images acquired using two different sensors is modeled using non-linear distortions. Experimental results using multiple datasets confirm the efficacy of the proposed method. Papers: A. Ross and R. Nadgir, "A Thin-plate Spline Calibration Model For Fingerprint Sensor Interoperability," IEEE Transactions on Knowledge and Data Engineering, Special Issue on Intelligence and Security Informatics, Vol. 20, No. 8, pp. 1097-1110, August 2008. A. Ross and R. Nadgir, "A Calibration Model for Fingerprint Sensor Interoperability", Proc. of SPIE Conference on Biometric Technology for Human Identification III, (Orlando, USA), pp. 62020B-1 - 62020B-12, April 2006. A. Ross and A. K. Jain, " Biometric Sensor Interoperability: A Case Study In Fingerprints", Proc. of International ECCV Workshop on Biometric Authentication (BioAW), (Prague), LNCS vol. 3087, pp. 134-145, Springer Publishers, May 2004.

Hybrid Fingerprint Matcher Most fingerprint matching systems rely on the distribution of minutiae on the fingertip to represent and match fingerprints. While the ridge flow pattern is generally used for classifying fingerprints, it is seldom used for matching. This work describes a hybrid fingerprint matching scheme that uses both minutiae and ridge flow information to represent and match fingerprints. A set of 8 Gabor filters, whose spatial frequencies correspond to the average inter-ridge spacing in fingerprints, is used to capture the ridge strength at equally spaced orientations. A square tessellation of the filtered images is then used to construct an eight-dimensional feature map, called the ridge feature map. The ridge feature map along with the minutiae set of a fingerprint image is used for matching purposes. The proposed technique has the following features: (i) the entire image is taken into account in constructing the ridge feature map, and every tessellated cell is equally weighted; (ii) minutiae matching is used to determine the affine transformation parameters relating the query and the template images for ridge feature map extraction; (iii) filtering and ridge feature map extraction are implemented in the frequency domain thereby speeding up the matching process; (iv) filtered query images are cached to greatly increase the one-to-many matching speed. The hybrid matcher s observed to perform better than a minutiae-based fingerprint matching system. Papers: A. Ross, A. K. Jain and J. Reisman, "A Hybrid Fingerprint Matcher", Pattern Recognition, Vol. 36, No. 7, pp. 1661-1673, July, 2003. A. Ross, A. K. Jain and J. Reisman, " A Hybrid Fingerprint Matcher", Proc. of International Conference on Pattern Recognition (ICPR), Quebec City, August 11-15, 2002. A. K. Jain, A. Ross and S. Prabhakar, "Fingerprint Matching Using Minutiae and Texture Features", Proc. of Int'l Conference on Image Processing (ICIP), pp.282-285, Thessaloniki, Greece, Oct 7 - 10, 2001.

Multibiometrics User verification systems that use a single biometric indicator often have to contend with noisy sensor data, restricted degrees of freedom, non-universality of the biometric trait and unacceptable error rates. Attempting to improve the performance of individual matchers in such situations may not prove to be effective because of these inherent problems. Multibiometric systems seek to alleviate some of these drawbacks by providing multiple evidences of the same identity. These systems help achieve an increase in performance that may not be possible using a single biometric indicator. Further, multibiometric systems provide anti-spoofing measures by making it difficult for an intruder to spoof multiple biometric traits simultaneously. However, an effective fusion scheme is necessary to combine the information presented by multiple domain experts. This work addresses the problem of information fusion in biometric systems. Depending upon the sources of information being consolidated, a multibiometric system can be characterized as a multi-sample, multi-sensor, multi-algorithm, multi-instance (or multi-unit), or multimodal system. Furthermore, there are various levels of fusion possible in such a system, viz., raw data level, feature level, score level, rank level, and decision level. Papers: A. Ross, " An Introduction to Multibiometrics," Proc. of the 15th European Signal Processing Conference (EUSIPCO), (Poznan, Poland), September 2007. A. Ross, K. Nandakumar and A. K. Jain, "Handbook of Multibiometrics", Springer Publishers, 1st edition, 2006. ISBN: 0-3872-2296-0. [Springer] [Amazon] A. K. Jain and A. Ross, "Multibiometric Systems", Communications of the ACM, Special Issue on Multimodal Interfaces, Vol. 47, No. 1, pp. 34-40, January 2004. A. Ross and A. K. Jain, " Information Fusion in Biometrics", Pattern Recognition Letters, Vol. 24, Issue 13, pp. 2115-2125, September, 2003. A. Ross, A. K. Jain, and J. Qian, " Information Fusion in Biometrics", Proc. of 3rd Int'l Conference on Audio- and Video-Based Person Authentication (AVBPA), pp. 354-359, Sweden, June 6-8, 2001.

Learning User-specific Parameters in Multibiometrics Biometric systems that use a single biometric trait have to contend with noisy data, restricted degrees of freedom, failureto- enroll problems, spoof attacks, and unacceptable error rates. Multibiometric systems that use multiple traits of an individual for authentication, alleviate some of these problems while improving verification performance. We demonstrate that the performance of multibiometric systems can be further improved by learning user-specific parameters. Two types of parameters are considered here. (i) Thresholds that are used to decide if a matching score indicates a genuine user or an impostor, and (ii) weights that are used to indicate the importance of matching scores output by each biometric trait. User-specific thresholds are computed using the cumulative histogram of impostor matching scores corresponding to each user. The user-specific weights associated with each biometric are estimated by searching for that set of weights which minimizes the total verification error. The tests were conducted on a database of 50 users who provided fingerprint, face and hand geometry data, with 10 of these users providing data over a period of two months. We observed that user-specific thresholds improved system performance by 2%, while user-specific weights improved performance by 3%. Papers: A. K. Jain and A. Ross, " Learning User-specific Parameters in a Multibiometric System", Proc. International Conference on Image Processing (ICIP), pp. 57-60, Rochester, New York, September 22-25, 2002.

Score Normalization in Multibiometrics Although information fusion in a multimodal system can be done at various levels, integration at the match score level is the most common approach due to the ease in accessing and combining the scores generated by different matchers. Since the match scores output by the various modalities are heterogeneous, score normalization is needed to transform these scores into a common domain prior to combining them. In this work we have studied the performance of different normalization techniques and fusion rules in the context of a multimodal biometric system that uses the face, fingerprint and hand-geometry traits of a user. Experiments conducted on a database of 100 users show that the min-max, z-score, and tanh normalization techniques followed by a simple sum of scores fusion method result in better recognition performance compared to other methods. However, experiments also reveal that the min-max and z-score normalization techniques are sensitive to outliers in the data, highlighting the need for a robust and efficient normalization procedure like the tanh normalization. Further, multimodal biometric systems that utilize user-specific weights are found to have better recognition rates compared to systems that assign equal weights to all modalities. Papers: A. K. Jain, K. Nandakumar and A. Ross, "Score Normalization in Multimodal Biometric Systems", Pattern Recognition, Vol. 38, No. 12, pp. 2270-2285, December 2005.

Feature Level Fusion in Biometrics Multimodal biometric systems utilize the evidence presented by multiple biometric sources (e.g., face and fingerprint, multiple fingers of a user, multiple impressions of a single finger, etc.) in order to determine or verify the identity of an individual. Information from multiple sources can be consolidated in several distinct levels While fusion at the match score and decision levels have been extensively studied in the literature, fusion at the feature level is a relatively understudied problem. We present a technique to perform fusion at the feature level by considering two biometric modalities - face and hand geometry. Preliminary results indicate that the proposed technique can lead to substantial improvement in multimodal matching performance. Fusion at the feature level involves the integration of feature sets corresponding to multiple modalities. Since the feature set contains richer information about the raw biometric data than the match score or the final decision, integration at this level is expected to provide better recognition results. However, fusion at this level is difficult to achieve in practice because of the following reasons: (i) the feature sets of multiple modalities may be incompatible (e.g., minutiae set of fingerprints and eigen-coefficients of face); (ii) the relationship between the feature spaces of different biometric systems may not be known; and (iii) concatenating two feature vectors may result in a feature vector with very large dimensionality leading to the 'curse of dimensionality' problem. We describe a technique that utilizes the fused feature vectors of face and hand geometry in order to improve the performance of a multimodal biometric system. Papers: A. Ross and R. Govindarajan, "Feature Level Fusion Using Hand and Face Biometrics", Proc. of SPIE Conference on Biometric Technology for Human Identification II, (Orlando, USA), pp. 196-204, March 2005.

Face Mosaicing Mosaicing entails the consolidation of information represented by multiple images through the application of a registration and blending procedure. We describe a face mosaicing scheme that generates a composite face image during enrollment based on the evidence provided by frontal and semi-profile face images of an individual. Face mosaicing obviates the need to store multiple face templates representing multiple poses of a user's face image. In the proposed scheme, the side profile images are aligned with the frontal image using a hierarchical registration algorithm that exploits neighborhood properties to determine the transformation relating the 2 images. Multiresolution splining is then used to blend the side profiles with the frontal image thereby generating a composite face image of the user. A texture-based face recognition technique that is a slightly modified version of the C2 algorithm proposed by Serre et al. is used to compare a probe face image with the gallery face mosaic. Experiments conducted on three different databases indicate that face mosaicing as described in this paper offers significant benefits by accounting for the pose variations commonly observed in face images. Papers: R. Singh, M. Vatsa, A. Ross and A. Noore, "A Mosaicing Scheme for Pose Invariant Face Recognition," IEEE Transactions on Systems, Mans and Cybernetics - B, Special Issue on Biometrics, Vol. 37, Issue 5, pp. 1212-1225, October 2007. R. Singh, M. Vatsa, A. Ross and A. Noore, "Performance Enhancement of 2D Face Recognition via Mosaicing", Proc. of 4th IEEE Workshop on Automatic Identification Advanced Technologies (AutoID), (Buffalo, USA), pp. 63-68, October 2005. Best Student Paper Award (Singh, Vatsa).

Texture-based Approach to Face Detection Detection of faces in static or video images is an important but challenging problem in computer vision that has applications in image retrieval systems, biometrics, surveillance and law enforcement. It is an essential first-step in face recognition where the goal is to localize the spatial extent of a face in order to determine the identity of an individual in an image. Several methods have been proposed in the literature for detecting faces. Most techniques perform well in constrained environments but perform poorly on noisy images having a cluttered background. Face detection is challenging as faces could occur at different scales, orientations, positions and pose in an image with an uncontrolled background. Texture is an important visual cue that the biological visual system computes preattentively. A face image can be thought of as a symmetric and regular texture pattern. The texture features considered in our work are a set of statistical and multiresolution features which capture the gradient and residual energies of a pattern as well as the directional variations of the pattern. The features extracted from a region are the average, standard deviation, average deviation of gradient magnitude, average residual energy, and average deviation of the horizontal and vertical directional residual of the pixel intensities. The multiresolution features used are the energy, standard deviation and residual energy of the detail subimages. Detection results from several databases show the effectiveness of adding texture features to the face feature space. The algorithm requires fewer training samples than learning-based state-of-the-art methods, and its robustness is demonstrated by training and testing with different face databases. Papers: V. Manian and A. Ross, "Face Detection Using Statistical and Multitresolution Texture Features", Multimedia Cyberscape Journal, Special Issue on Pattern Recognition in Biometrics and Bioinformatics, Vol. 3, No. 3, pp. 1-9, 2005. V. Manian and A. Ross, "A Texture-based Approach to Face Detection", Biometric Consortium Conference (BCC), (Crystal City, VA), September 2004.

Hand Geometry Geometric measurements of the human hand have been used for identity authentication in a number of commercial systems. Yet, there is not much open public literature addressing research issues underlying hand geometry-based identity authentication. This work is our attempt to draw attention to this important biometric by designing a prototype hand geometry-based identity authentication system. We also present our preliminary verification results based on hand measurements of 55 individuals captured over a period of time. The results are encouraging and we plan to address issues to improve the system performance. Papers: A.K. Jain, A. Ross and S. Pankanti, "A Prototype Hand Geometry-based Verification System", Proc. of 2nd Int'l Conference on Audio- and Video-based Biometric Person Authentication (AVBPA) , Washington D.C., pp.166-171, March 22-24, 1999. A. K. Jain, A. Ross, and S. Prabhakar, "Biometrics-based Web Access", MSU Technical Report, TR98-33, 1998. Arun Ross, ""A Prototype Hand Geometry-based Verification System", M.S. Project Report, Michigan State University, 1999.