Search

In many areas in machine learning, decision trees play a crucial role in classification and regression. When a decision tree based classifier is hosted as a service in a critical application with the need for privacy protection of the service as well as the user data, fully homomorphic encrypted can be employed. However, a decision node in a decision tree can’t be directly implemented in FHE. In this paper, we describe an end-to-end approach to support privacyenhanced decision tree classification using IBM supported open-source library HELib. Using several options for building a decision node and employing oblivious computations coupled with an argmax function in FHE we show that a highly secure and trusted decision tree service can be enabled.

Several successful adversarial attacks have demonstrated the vulnerabilities of deep learning algorithms. These attacks are detrimental in building deep learning based dependable AI applications. Therefore, it is imperative to build a defense mechanism to protect the integrity of deep learning models. In this paper, we present a novel "defense layer" in a network which aims to block the generation of adversarial noise and prevents an adversarial attack in black-box and gray-box settings. The parameter-free defense layer, when applied to any convolutional network, helps in achieving protection against attacks such as FGSM, L 2 , Elastic-Net, and DeepFool. Experiments are performed with different CNN architectures, including VGG, ResNet, and DenseNet, on three databases, namely, MNIST, CIFAR-10, and PaSC. The results showcase the efficacy of the proposed defense layer without adding any computational overhead. For example, on the CIFAR-10 database, while the attack can reduce the accuracy of the ResNet-50 model to as low as 6.3%, the proposed "defense layer" retains the original accuracy of 81.32%.

Several computer vision applications such as object detection and face recognition have started to completely rely on deep learning based architectures. These architectures, when paired with appropriate loss functions and optimizers, produce state-of-the-art results in a myriad of problems. On the other hand, with the advent of "blockchain", the cybersecurity industry has developed a new sense of trust which was earlier missing from both the technical and commercial perspectives. Employment of cryptographic hash as well as symmetric/asymmetric encryption and decryption algorithms ensure security without any human intervention (i.e., centralized authority). In this research, we present the synergy between the best of both these worlds. We first propose a model which uses the learned parameters of a typical deep neural network and is secured from external adversaries by cryptography and blockchain technology. As the second contribution of the proposed research, a new parameter tampering attack is proposed to properly justify the role of blockchain in machine learning.

Recent work shows unequal performance of commercial face classification services in the gender classification task across intersectional groups defined by skin type and gender. Accuracy on dark-skinned females is significantly worse than on any other group. We provide initial evidence that skin type alone is not the driver for this disparity by conducting novel stability experiments that vary an image's skin type via color-theoretic methods, namely luminance mode-shift and optimal transport. We evaluate the effect of skin type change on the gender classification decision of a pair of state-of-the-art commercial and open-source gender classifiers. The results raise the possibility that broader differences in ethnicity, as opposed to the skin type alone, are what contribute to unequal gender classification accuracy in face images.

While deep learning is a valuable tool for solving many tough problems in computer vision, the success of deep learning models is typically determined by: (i) availability of sufficient training data, (ii) access to extensive computational resources, and (iii) expertise in selecting the right model and hyperparameters for the selected task. Often, the availability of data is the hard part due to compliance, legal, and privacy constraints. Cryptographic techniques such as fully homomorphic encryption (FHE) offer a potential solution by enabling processing on encrypted data. While prior work has been done on using FHE for inferencing, training a deep neural network in the encrypted domain is an extremely challenging task due to the computational complexity of the operations involved. In this paper, we evaluate the feasibility of training neural networks on encrypted data in a completely non-interactive way. Our proposed system uses the open-source FHE toolkit HElib to implement a Stochastic Gradient Descent (SGD)-based training of a neural network. We show that encrypted training can be made more computationally efficient by (i) simplifying the network with minimal degradation of accuracy, (ii) choosing appropriate data representation and resolution, and (iii) packing the data elements within the ciphertext in a smart way so as to minimize the number of operations and facilitate parallelization of FHE computations. Based on the above optimizations, we demonstrate that it is possible to achieve more than 50x speed up while training a fully-connected neural network on the MNIST dataset while achieving reasonable accuracy (96%). Though the cost of training a complex deep learning model from scratch on encrypted data is still very high, this work establishes a solid baseline and paves the way for relatively simpler tasks such as fine-tuning of deep learning models based on encrypted data to be implemented in the near future.

Mobile biometric approaches provide the convenience of secure authentication with an omnipresent technology. However, this brings an additional challenge of recognizing biometric patterns in an unconstrained environment including variations in mobile camera sensors, illumination conditions, and capture distance. To address the heterogeneous challenge, this research presents a novel heterogeneity aware loss function within a deep learning framework. The effectiveness of the proposed loss function is evaluated for periocular biometrics using the CSIP, IMP and VISOB mobile periocular databases. The results show that the proposed algorithm yields state-of-the-art results in a heterogeneous environment and improves generalizability for cross-database experiments.

Existing research in the field of face recognition with variations due to disguises focuses primarily on images captured in controlled settings. Limited research has been performed on images captured in unconstrained environments, primarily due to the lack of corresponding disguised face datasets. In order to overcome this limitation, this work presents a novel Disguised Faces in the Wild (DFW) dataset, consisting of over 11,000 images for understanding and pushing the current state-of-the-art for disguised face recognition. To the best of our knowledge, DFW is a first-of-a-kind dataset containing images pertaining to both obfuscation and impersonation for understanding the effect of disguise variations. A major portion of the dataset has been collected from the Internet, thereby encompassing a wide variety of disguise accessories and variations across other covariates. As part of CVPR2018, a competition and workshop are organized to facilitate research in this direction. This paper presents a description of the dataset, the baseline protocols and performance, along with the phase-I results of the competition.

Classifier fusion is a well-studied problem in which decisions from multiple classifiers are combined at the score, rank, or decision level to obtain better results than a single classifier. Subsequently, various techniques for combining classifiers at each of these levels have been proposed in the literature. Many popular methods entail scaling and normalizing the scores obtained by each classifier to a common numerical range before combining the normalized scores using the sum rule or another classifier. In this research, we explore an alternative method to combine classifiers at the score level. The Pool Adjacent Violators (PAV) algorithm has traditionally been utilized to convert classifier match scores to confidence values that model posterior probabilities for test data. The PAV algorithm and other score normalization techniques have studied the same problem without being aware of each other. In this first ever study to combine the two, we propose the PAV algorithm for classifier fusion on publicly available NIST multi-modal biometrics score dataset. We observe that it provides several advantages over existing techniques and find that the interpretation learned by the PAV algorithm is more robust than the scaling learned by other popular normalization algorithms such as min-max. Moreover, the PAV algorithm enables the combined score to be interpreted as confidence and is able to further improve the results obtained by other approaches. We also observe that utilizing traditional normalization techniques first for individual classifiers and then normalizing the fused score using PAV offers a performance boost compared to only using the PAV algorithm.

Biometric systems can be attacked in several ways and the most common being spoofing the input sensor. Therefore, anti-spoofing is one of the most essential prerequisite against attacks on biometric systems. For face recognition it is even more vulnerable as the image capture is non-contact based. Several anti-spoofing methods have been proposed in the literature for both contact and non-contact based biometric modalities often using video to study the temporal characteristics of a real vs. spoofed biometric signal. This paper presents a novel multi-feature evidence aggregation method for face spoofing detection. The proposed method fuses evidence from features encoding of both texture and motion (liveness) properties in the face and also the surrounding scene regions. The feature extraction algorithms are based on a configuration of local binary pattern and motion estimation using histogram of oriented optical flow. Furthermore, the multi-feature windowed videolet aggregation of these orthogonal features coupled with support vector machine-based classification provides robustness to different attacks. We demonstrate the efficacy of the proposed approach by evaluating on three standard public databases: CASIA-FASD, 3DMAD and MSU-MFSD with equal error rate of 3.14%, 0%, and 0%, respectively.

Often deep learning methods are associated with huge amounts of training data. The deeper the network gets, the larger is the need for training data. A large amount of labeled data helps the network learn about the variations it needs to handle in the prediction stage. It is not easy for everyone to get access to huge amounts of labeled data leaving a few to have the luxury to design very deep networks. In this paper, we propose to flatten the disparity by using the modeling methods to minimize the need for huge amounts of data for training a deep network. Using face recognition as an example, we demonstrate how limited labeled data can be leveraged to obtain near state of the art performance with generalization capability across multiple databases. In addition, we show that the normalization in the overall network can improve the speed and resource requirement for the prediction/inferencing stage.

A personal or enterprise collection of a large set of face images may contain many types of tags used for querying the collection. Often the tags have many irrelevant content that may not reflect the image content in terms of the facial characteristics. In this paper, we propose a data curation method to filter out the irrelevant face images using a face recognition based subgraph identification. Results on retrievals from the Internet using popular celebrities show the efficacy of our approach after we cleanse the images collection retrieved and applying our algorithm to the collection.

It has been shown that significant age difference between a probe and gallery face image can decrease the matching accuracy. If the face images can be normalized in age, there can be a huge impact on the face verification accuracy and thus many novel applications such as matching driver's license, passport and visa images with the real person's images can be effectively implemented. Face progression can address this issue by generating a face image for a specific age. Many researchers have attempted to address this problem focusing on predicting older faces from a younger face. In this paper, we propose a novel method for robust and automatic face progression in totally unconstrained conditions. Our method takes into account that faces belonging to the same age-groups share age patterns such as wrinkles while faces across different age-groups share some common patterns such as expressions and skin colors. Given training images of K different age-groups the proposed method learns to recover K low-rank age and one low-rank common components. These extracted components from the learning phase are used to progress an input face to younger as well as older ages in bidirectional fashion. Using standard datasets, we demonstrate that the proposed progression method outperforms state-of-the-art age progression methods and also improves matching accuracy in a face verification protocol that includes age progression.

Biologically inspired model (BIM) is proven to be an effective feature representation approach for visual object categorization. In BIM, two successive S(simple)-to-C(complex) hierarchical layers are performed to simulate the visual perception process of primate visual cortex. However, the intensive computational cost above C1 layer in BIM extremely limits its application in real-time object recognition tasks. This paper proposes to use a set of improved early biologically inspired features (EBIF, including S1 and C1) for face recognition, in which pyramidal statistics of mean and standard deviation rather than MAX pooling are used for scale-tolerant feature condensation and local normalization is performed on C1 layer. Incremental PCA (Principal Component Analysis) and LDA (Linear Discriminant Analysis) are then combined to efficiently learn a discriminant subspace for feature dimensionality reduction. In the matching stage, Cosine similarity is adopted as the distance metric for a given face pair. Experimental results on two public face datasets and a mobile face dataset show the effectiveness of the proposed method.

Iris recognition has gained popularity due to factors such as its perceived high accuracy, significant usability advantages attributed to its non-contact acquisition method, and the availability of low cost sensors due to improvements in technology. However, non-contact biometrics authentication systems are vulnerable to different types of attacks than contact-type biometrics, such as fingerprints, for which there are a number of simple techniques to guard against attacks. In particular, the fashion industry has developed designer contact lenses with patterns that range from a simple change in eye color to the imposition of stars or other festive decorations. As these lenses are readily available and can be personalized at a very affordable price, their use in thwarting or spoofing iris-based authentication systems becomes plausible. Given the high security nature of many of these systems, there is a urgent need for a some countermeasure to this type of attack. In this paper, we describe a novel method to detect the presence of fake iris patterns, such as designer contact lenses, during the image acquisition stage to further enhance the basic security value of iris biometrics. Exploiting the anatomy and geometry of the human eye, we present a structured light projection method to detect the presence of artificial items obscuring the real iris. The detection principle has been verified using an inexpensive experimental setup consisting of a miniature projector and an offset camera. We also describe a novel algorithm to process the acquired images to find patterned contact lenses, and measure its performance using data collected with our apparatus. We argue that the addition of the proposed system and algorithm to existing iris biometrics based authentication systems will significantly improve their security.

Face recognition systems, trained in controlled environment, often fail to efficiently match low resolution images with high resolution images. In this research, a co-transfer learning framework is proposed in which knowledge learnt in controlled high resolution environment is transferred for matching low resolution probe images with high resolution gallery. The proposed framework seamlessly combines transfer learning and co-training to perform knowledge transfer by updating classifier's decision boundary with low resolution probe instances. Experiments are performed on the CMU-Multi-PIE and SCface database with gallery images of size 72 × 72 and size of probe images varying from 48 × 48 to 16 × 16. The results show that, in terms of rank-1 identification accuracy, the proposed algorithm outperforms existing approaches by at least 5%.

In the multibiometric systems, various matcher/modality scores are fused together to provide better performance than the individual matcher scores. In [1] the authors have proposed a likelihood ratio test (LRT) based fusion technique for the biometric verification task that outperformed several other classifiers. They model the genuine and the imposter densities by the finite Gaussian mixture models (GMM, a generative model) whose parameters are estimated using the maximum likelihood (ML) criteria. Lately, the discriminative training methods and models have been shown to provide additional accuracy gains over the generative models, in multiple applications such as the speech recognition, verification and text analytics[5, 7]. These gains are based on the fact that the discriminative models are able to partially compensate for the unavoidable mismatch, which is always present between the specified statistical model (GMM in this case) and the true distribution of the data which is unknown. In this paper, we propose to use a discriminative method to estimate the GMM density parameters using the maximum accept and reject (MARS) criteria[8]. The test results using the proposed method on the NIST-BSSRI multimodal dataset indicate improved verification performance over a very competitive maximum likelihood (ML) trained system proposed in [1].

While biometrics-based identification is a key technology in many critical applications such as searching for an identity in a watch list or checking for duplicates in a citizen ID card system, there are many technical challenges in building a solution because the size of the database can be very large (often in 100s of millions) and the intrinsic errors with the underlying biometrics engines. Often multi-modal biometrics is proposed as a way to improve the underlying biometrics accuracy performance. In this paper, we propose a score based fusion scheme tailored for identification applications. The proposed algorithm uses a gradient descent method to learn weights for each modality such that weighted sum of genuine scores is larger than the weighted sum of all the impostor scores. During the identification phase, top K candidates from each modality are retrieved and a super-set of identities is constructed. Using the learnt weights, we compute the weighted score for all the candidates in the superset. The highest scoring candidate is declared as the top candidate for identification. The proposed algorithm has been tested using NIST BSSR-1 dataset and results in terms of accuracy as well as the speed (execution time) are shown to be far superior than the published results on this dataset.

In many identification problems, finding a duplicate using biometrics is very challenging task because of the size of the database and the errors related to the core biometrics engine. Fusion of different modes of biometric system can lead to improved recognition accuracy. In a multi classifier fusion scenario, the confidence of one classifier should not only depend on its own decision confidence but also on the diffidence of other classifiers. We propose a novel solution which uses a machine learning approach to generate confidence scores for each system for every instance of decision by implicitly modelling the redundancy (all classifiers making the same decision) and diversity (each classifier making a different decision and only a subset of classifiers is right at one time)from the training data. These confidence scores are used as weights for votes and the final decision is made using weighted sum of votes. Experimental results are provided by comparing this method with more conventional methods like majority voting, and majority voting with confidence. The performance of the proposed method on NIST BSSR-1 dataset and FERET face dataset shows the efficacy of our approach measured by the accuracy improvements we are able to achieve by implicitly modelling the redundancy and diversity measures.

In biometrics authentication systems, it has been shown that fusion of more than one modality (e.g., face and finger) and fusion of more than one classifier (two different algorithms) can improve the system performance. Often a score level fusion is adopted as this approach doesn't require the vendors to reveal much about their algorithms and features. Many score level transformations have been proposed in the literature to normalize the scores which enable fusion of more than one classifier. In this paper, we propose a novel score level transformation technique that helps in fusion of multiple classifiers. The method is based on two components: quantile transform of the genuine and impostor score distributions and a power transform which further changes the score distribution to help linear classification. After the scores are normalized using the novel quantile power transform, several linear classifiers are proposed to fuse the scores of multiple classifiers. Using the NIST BSSR-1 dataset, we have shown that the results obtained by the proposed method far exceed the results published so far in the literature.

Privacy and security are essential requirements in practical biometric systems. In order to prevent the theft of biometric patterns, it is desired to modify them through revocable and non invertible transformations called Cancelable Biometrics. In this paper, we propose an efficient algorithm for generating a Cancelable Iris Biometric based on Sectored Random Projections. Our algorithm can generate a new pattern if the existing one is stolen, retain the original recognition performance and prevent extraction of useful information from the transformed patterns. Our method also addresses some of the drawbacks of existing techniques and is robust to degradations due to eyelids and eyelashes.

There have been claims of very high information content in iris texture, higher even than in fingerprints. This makes iris attractive for large scale identification systems with possibly millions of people. However, some systems operate by performing N 1:1 matches of the probe against the database. This can get prohibitively expensive in terms of computation as N grows large. Note that for identification systems the per-match time dominants system performance, unlike verification where feature extraction time is the primary component. In this paper we show how to use a short-length iris code to pre-screen a large database and thereby reduce the number of full comparisons needed to a fraction of the total. Since the screening code is much smaller than the full iris code, the time to process the whole database is greatly reduced. As an added benefit, we show that we can use the alignment inferred from the short code to greatly restrict the range of alignments searched for the full code, which further speeds up the system. As we demonstrate in experiments, the two stage approach can reduce the cost and/or time needed by an order of magnitude with very little impact on identification performance.

The texture in a human iris has been shown to have good individual distinctiveness and thus is suitable for use in reliable identification. A conventional iris recognition system unwraps the iris image and generates a binary feature vector by quantizing the response of selected filters applied to the rows of this image. Typically there are 360 angular sectors, 64 radial rings, and 2 filter responses. This produces a full-length iris code (FLIC) of about 5760 bytes. In contrast, this paper seeks to shrink the representation by finding those regions of the iris that contain the most descriptive potential. We show through experiments that the regions close to the pupil and sclera contribute least to discrimination, and that there is a high correlation between adjacent radial rings. Using these observations we produce a short-length iris code (SLIC) of only 450 bytes. The SLIC is an order of magnitude smaller the FLIC and yet has comparable performance as shown by results on the MMU2 database. The smaller sized representation has the advantage of being easier to store as a barcode, and also reduces the matching time per pair.

A person only has two irises - if his pattern is stolen he quickly runs out of alternatives. Thus methods that protect the true iris pattern need to be adopted in practical biometric applications. In particular, it is desirable to have a system that can generate a new unique pattern if the one being used is lost, or generate different unique patterns for different applications to prevent cross-matching. For backwards compatibility, these patterns should look like plausible irises so they can be handled with the same processing tools. However, they should also non-invertibly hide the true biometric so it is never exposed, or even stored. In this paper four such ldquocancelablerdquo biometric methods are proposed that work with conventional iris recognition systems, either at the unwrapped image level or at the binary iris code level.

Cancelable biometric systems are gaining in popularity for use in person authentication for applications where the privacy and security of biometric templates are important considerations. A variety of approaches have been proposed in the literature. In this work, we have chosen two (a registration based and a registration free) techniques and performed a comparative study focusing on template representation size, useful dataset coverage, system accuracy and transform strength. Results show that both systems have their own advantages that are suited for use in specific applications.

Iris segmentation is an important first step for high accuracy iris recognition. A robust iris segmentation procedure should be able to handle noise, occlusion and non-uniform lighting. It also impacts system accuracy - high FAR or FRR values may come directly from bad or wrong segmentations. In this paper a simple new approach for iris segmentation is proposed that tries to integrate quality evaluation ideas directly into the segmentation algorithm. By cutting out all the bad areas, the fraction of the iris that remains can be used as a comprehensive quality measure. This eliminates images with high occlusion (e.g. by the eyelids) as well as images with other quality problems (e.g. low contrast), all using the same mechanism. The proposed method has been tested on a medium-sized (450 image) public database (MMU1) and the score distribution investigated. We also show that, as expected, overall matching accuracy can be improved by rejecting images which have a low quality assessment, thus validating the utility of this measure.

We present a face reconstruction approach for revocable face matching. The proposed approach generates photometrically valid cancelable face images by following the image formation process. Given a face image, the approach estimates facial albedo followed by a subject-specific key based photometric deformation to generate a cancelable face image. The proposed approach allows for using any available face matcher to perform verification or recognition in the transformed domain, a capability missing from most existing works on cancelable face matching. Experiments are performed to evaluate the performance, privacy and cancelable aspects of the face images reconstructed using the approach. Results obtained are very promising and make a strong case for such backward compatible cancelable face representations that can seamlessly make use of advancements in automatic face recognition research.

One of the main challenges in building an efficient and scalable automatic fingerprint identification system is to identify features which are highly discriminative and are reproducible across different prints of the same finger. Most existing fingerprint matching approaches rely on minutiae geometry. Relatively, little effort has gone into analyzing ridge flow patterns present in the fingerprint, partly due to difficulty in extracting robust discriminative features from the fingerprint images. In this paper, we analyze the usefulness of ridge curvature information for fingerprint matching and classification applications. Specifically, for an indexing framework, we explore whether the curvature information can be utilized along with the existing minutiae geometry-based features for further reducing the number of potential candidates for fingerprint identification. Experimental results indicate the robustness of the proposed curvature-based characterization and its usefulness in improving the efficiency of existing fingerprint-based identification systems.

Iris-based human recognition is very attractive because of the high accuracy achievable. However, existing encoding methods are unable to handle iris images acquired when the ambient lighting is non-uniform. In this paper we propose a novel encoding technique which can handle images acquired under such conditions. The method is based on 2D median filters, which are commonly used as de-noising tools in image processing domain, and uses their nonlinear characteristics to generate binary codes from gray scale iris images. The application of this method for robust iris recognition is discussed and compared with two more traditional iris encoding techniques: the 2D Gabor filter based method and the 1D log-Gabor method. Our algorithm is usable for the whole range of conditions encountered in the MMU1 iris database, performs well in a synthetic uneven illumination study, and gets comparable results on other tests.

Though a lot of research has been done to match fingerprints, most existing approaches rely on locations of minutiae features for matching tasks. Relatively, little effort has gone into utilizing textural information present in fingerprints as distinguishing characteristic. In this paper, we propose a novel gradient-based approach to characterize textural information in fingerprints for the task of biometric matching. In particular, the proposed approach uses histograms of oriented gradients (HOGs) to represent minutiae neighborhoods. The minutiae neighborhoods are divided into several regions to make the computed histograms distinguishing and robust at the same time. Experimental results are provided to show the efficacy of the proposed characterization.

Cancelable biometrics bridges the gap between the convenience of biometric authentication and security offered by cryptographic systems. There have been several prior attempts to construct cancelable templates out of fingerprint biometrics. However, existing approaches require pre-alignment of fingerprint images and therefore, are not robust to image registration errors. In this paper, we propose a provably secure, registration-free construction of cancelable fingerprint templates. Firstly, we present a new fingerprint representation based on localized, self-aligned texture features. Secondly, we demonstrate how to construct cancelable templates from them and also show how to harden the security of the proposed construction further. And finally, we present arguments for it resilience against various modes of adversarial attack.

The security, lack of anonymity and revocability of the biometric template are critical issues that need to be addressed in order to vindicate the viability of biometric based authentication systems. Several methods have been proposed to address these problems. However, most of these methods offer lower accuracies than the base system where the template is insecure. This is because in most systems the gain in the security is achieved as a result of loss in non-redundant information. In this paper, we describe tradeoff between accuracy of an anonymous system and the security of the biometrics system. As a case study we start with a highly secure representation of a fingerprint. Then we describe several methods and show experimental results proving that every time we add more information to the secure representation, the accuracy increases, however valuable information is revealed to an adversary. We propose, via a k-trial attack model, how the trade-off can be complemented by another security policy such as an account lockout after a given number of attempts.

Biometric identification has numerous advantages over conventional ID and password systems; however, the lack of anonymity and revocability of biometric templates is of concern. Several methods have been proposed to address these problems. Many of the approaches require a precise registration before matching in the anonymous domain. We introduce binary string representations of fingerprints that obviates the need for registration and can be directly matched. We describe several techniques for creating anonymous and revocable representations using these binary string representations. The match performance of these representations is evaluated using a large database of fingerprint images. We prove that given an anonymous representation, it is computationally infeasible to invert it to the original fingerprint, thereby preserving privacy. To the best of our knowledge, this is the first linear, anonymous and revocable fingerprint representation that is implicitly registered.

A majority of the minutiae based fingerprint verification algorithms rely on explicit or implicit alignment of the minutiae points for matching the two prints. With no prior knowledge about point correspondences, this becomes a combinatorial problem. Global features of the fingerprints such as the core and delta points represent intrinsic points of reference that can be used to align the two prints and reduce the computational complexity of the matcher. However, automatic extraction of singular points is usually error prone and is therefore not used by existing matchers. But, a systematic study of the impact on matching performance when core/delta points are available has not been done to date. In this paper, we explore the effects of the availability of reliable core and delta points on speed and accuracy of a matching algorithm. Towards this end, we present significant improvements to core and delta point detection algorithm based on complex filtering principles originally proposed by Nilsson et al., (2005). We also present a modified graph based matching algorithm that can run in O(n) time when the reference points are available. We analyse the resulting improvement in computational complexity and present experimental evaluation over FVC2002 database. We show that there is upto 43% improvement (70.2 ms to 39.8 ms) in average verification time and almost no loss in accuracy when reliable core and delta points are used.

The cumulative match curve (CMC) is used as a measure of 1: m identification system performance. It judges the ranking capabilities of an identification system. The receiver operating characteristic curve (ROC curve) of a verification system, on the other hand, expresses the quality of a 1:1 matcher. The ROC plots the false accept rate (FAR) of a 1:1 matcher versus the false reject rate (FRR) of the matcher. We show that the CMC is also related to the FAR and FRR of a 1:1 matcher, i.e., the matcher that is used to rank the candidates by sorting the scores. This has as a consequence that when a 1:1 matcher is used for identification, that is, for sorting match scores from high to low, the CMC does not offer any additional information beyond the FAR and FRR curves. The CMC is just another way of displaying the data and can be computed from the FAR and FRR.

Biometrics offers usability advantages over traditional token and password based authentication schemes, but raises privacy and security concerns. When compromised, credit cards and passwords can be revoked or replaced while biometrics are permanently associated with a user and cannot be replaced. Cancelable biometrics attempt to solve this by constructing revocable biometric templates. We present several constructs for cancelable templates us- ing feature domain transformations and empirically exam- ine their efficacy. We also present a method for accurate registration which is a key step in building cancelable trans- forms. The overall approach has been tested using large databases and our results demonstrate that without losing much accuracy, we can build a large number of cancelable transforms for fingerprints.

A majority of the existing fingerprint recognition algorithms are based on matching minutia features. Therefore, minutiae extraction is one of the critical steps in fingerprint verification algorithms. Poor quality fingerprint images lead to missing and spurious minutiae that degrade the performance of the matching system. We propose two new techniques for minutiae verification based on non-trivial gray level features. The proposed features intuitively represent the structural properties of the minutiae neighborhood leading to better classification. We use directionally selective steerable wedge filters to differentiate between minutiae and non-minutiae neighborhoods. We also propose a second technique based on Gabor expansion that results in even better discrimination. We present an objective evaluation of both the algorithms.

Biometrics-based authentication and identification systems have to handle images acquired in noisy and hostile environments. The signal quality is assessed to decide if there is sufficient signal strength to process further. Poor quality signals require "enhancement" before further processing of the input signal. Often enhancement implies creating a more visibly pleasing image. However, biometrics signals need to improve the image quality for machine processability. This means that the enhancement algorithm should have some weak model about the sample (image) formation process. Enhancement is then some type of "normalization" or "beautification". We present a weak model-based image enhancement algorithm for fingerprint images. The results of the proposed algorithm are presented in terms of the improvements in the overall system performance measured in terms of a receiver operating characteristics curve.

Parallel implementations of two computer vision algorithms on distributed cluster platforms are described. The first algorithm is a square-error data clustering method whose parallel implementation is based on the well-known sequential CLUSTER program. The second algorithm is a motion parameter estimation algorithm used to determine correspondence between two images taken of the same scene. Both algorithms have been implemented and tested on cluster platforms using the PVM package. Performance measurements demonstrate that it is possible to attain good performance in terms of execution time and speedup for large-scale problems, provided that adequate memory; swap space, and I/O capacity are available at each node.

A hybrid Evolutionary algorithm using Evolutionary Programming (EP) and Genetic algorithms (GA), is proposed for the design of hardware efficient discrete coefficient modified filter banks. The proposed hybrid algorithm has improved convergence compared to the convergence provided by the two techniques individually. The filter coef-ficients are expressed as a sum of powers of two. Design examples are used to illustrate the effectiveness of the proposed algorithm