Experiments with an Improved Iris Segmentation Algorithm

The texture of the iris is commonly represented as an iris code in iris recognition systems. While several approaches have been presented for generating iris codes, relatively few comparison techniques have been proposed. In this paper, we take advantage of the availability of several frames from an iris video to create a single optimized iris code. This is achieved by performing both row-wise and column-wise optimization of iris codes. Inconsistent bits are accurately detected and masked in the final iris code. Our experiments demonstrate that by exploiting variations within the comparison scores of different rows and columns of N frames, we are able to derive the number of consistent bits in the final iris code thereby resulting in significant improvement in recognition performance. We compare our algorithm with well-known methods, namely, Fragile bit masking, Signal fusion and, two Score Fusion techniques. Experimental results on a dataset of 986 iris videos show that the proposed method is encouraging and comparable to the best algorithms in the current literature. To our knowledge, this is the first work that makes use of the best rows and columns from different frames in an iris video to improve performance.

Iris recognition, the ability to recognize and distinguish individuals by their iris pattern, is the most reliable biometric in terms of recognition and identification performance. However, performance of these systems is affected by poor quality imaging. In this work, we extend previous research efforts on iris quality assessment by analyzing the effect of seven quality factors: defocus blur, motion blur, off-angle, occlusion, specular reflection, lighting, and pixel-counts on the performance of traditional iris recognition system. We have concluded that defocus blur, motion blur, and off-angle are the factors that affect recognition performance the most. We further designed a fully automated iris image quality evaluation block that operates in two steps. First each factor is estimated individually, then the second step involves fusing the estimated factors by using Dempster-Shafer theory approach to evidential reasoning. The designed block is tested on two datasets, CASIA 1.0 and a dataset collected at WVU. Considerable improvement in recognition performance is demonstrated when removing poor quality images evaluated by our quality metric. The upper bound on processing complexity required to evaluate quality of a single image is O(<i>n</i>² log <i>n</i>), that of a 2D-Fast Fourier Transform.

This paper presents the iris recognition system for biometric personal identification using neural network. Personal identification consists of localization of the iris region and generation of a data set of iris images followed by iris pattern recognition. In this paper, a fast algorithm is proposed for the localization of the inner and outer boundaries of the iris region. Located iris is extracted from an eye image, and, after normalization and enhancement, it is represented by a data set. Using this data set a Neural Network (NN) is used for the classification of iris patterns. The adaptive learning strategy is applied for training of the NN. The results of simulations illustrate the effectiveness of the neural system in personal identification.

The role of biometrics, as an integral component of security infrastructures, has been firmly established in today's world. Its ability to accurately identify people from a distance has served as an important life-saving tool in the arsenalof our men and women in uniform, out in war theaters; it has also served as an effective forensics tool for our police force and off-late, is being increasingly adopted to secure sensitive transactions such as those with financial institutions.Traditionally, fingerprint patterns and facial photographs have held sway over identity veri fication tools; the former, because of its uniqueness and the latter because of its ubiquity. However, over the past decade, iris patterns have alsocome to the fore because of the combined advantage of uniqueness across the populace as well as the ability to image these patterns from a distance. Today most commercial iris acquisition systems report recognition accuracies upwards of 90%, over entire populations. However, the caveat is that these systems areconstrained to operate very close to the user, owing to the challenges involved in achieving sufficient resolution over the iride texture from large distances. This dissertation aims to push the boundaries of iris imaging devices and in-troduces theoretical and practical considerations in designing a novel long range iris recognition system. The aim is to design a system capable of acquiring enrollment quality iris images, similar in quality to those captured by short range systems such as IriShield, LG, PIER, HIIDE systems, at much larger stand-off distances. The challenges involved in acquiring images with sufficient spatialresolution, when subjects are too far out from the optimal acquisition range, are explored and potential solutions are described. First, we describe the design methodology behind creating a long range iris system, to aid the reader inunderstanding the key design considerations when using available commercial o -the-shelf hardware. Competing objectives such as achieving high magnifi cation while maintaining sufficient depth of field, using a sensor with large pixel pitch for low noise performance while resolving the required patterns, etc. are discussed. As a result of this work, the rst long range iris acquisition system of its kind, capable of acquiring enrollment iris images from up to 9mand matching up to 12m, has been designed and built. We show the system, present lessons learned during the design process as well as empirical results for the imaging performance. The latter half of the dissertation develops aniris texture enhancement algorithm for restoring degraded iris images - specifically noisy, low-resolution and out-of-focus/blurred images, which are often encountered in the real world. The method is developed and evaluated usingsynthesized degraded images from the NIST Iris Challenge Evaluation (ICE) database. A signifi cant improvement in verification performance is also seen, when using this enhancement method, on actual degraded images captured with the proposed long range iris system. The proposed enhancement method is modeled as a linear regression problem that exploits the sparsity of iris textures inan over-complete dictionary. Improvements in iris recognition performance using the enhanced textures are studied in the latter half of this dissertation. In summary, this work enhances and exceeds the capabilities of current iris imaging systems for the explicit purpose of accurately identifying threats using iride patterns, from distances larger than are currently possible. This will serve as anecessary tool in the arsenal of security establishments to identify threats, save lives and to protect our freedoms for generations to come.

In this study, the authors evaluated a processing algorithm aimed at improving speech recognition via the telephone among older adults with sensorineural hearing loss (SNHL). Thirty older adults with SNHL participated. Speech recognition was measured in quiet using the Modified Rhyme Test (MRT; Kreul et al., 1968) and the Speech Perception in Noise (SPIN; Bilger et al., 1984) sentences, and in noise using the Quick Speech in Noise (QSIN; Killion et al., 2004) test. Each test was presented via the telephone with and without processing. Significant improvements in recognition performance due to processing were observed for the SPIN and QSIN. The improvement on the QSIN was significantly greater than on the MRT and SPIN, likely because the MRT and SPIN sentences were presented in quiet, whereas the QSIN was presented in noise. Significant improvements in recognition performance were observed for both an offline version and a real-time version of the algorithm relative to the unprocessed condition, although no difference was noted between the 2 versions. Results indicate that preprocessing the acoustic signal is a viable method of improving speech recognition via the telephone. The algorithm has the potential to benefit older adults with SNHL who struggle to communicate via the telephone with or without hearing aids.

This report describes the large-scale experimental results from the Face Recognition Vendor Test (FRVT) 2006 and the Iris Challenge Evaluation (ICE) 2006. The FRVT 2006 looks at recognition from high-resolution still images and three-dimensional (3D) face images, and measures performance for still images taken under controlled and uncontrolled illumination. The ICE 2006 reports iris recognition performance from left and right iris images. The FRVT 2006 results from controlled still images and 3D images document an order-of-magnitude improvement in recognition performance over the FRVT 2002. This order-of-magnitude improvement was one of the goals of the preceding technology development effort, the Face Recognition Grand Challenge (FRGC). The FRVT 2006 and the ICE 2006 compared recognition performance from very-high resolution still face images, 3D face images, and single-iris images. On the FRVT 2006 and the ICE 2006 datasets, recognition performance was comparable for all three biometrics. In an experiment comparing human and algorithm performance, the best-performing face recognition algorithms were more accurate than humans. These and other results are discussed in detail.

Combined speech enhancement and auditory modelling for robust distributed speech recognition

Traditional iris segmentation methods give good results when the iris images are taken under ideal imaging conditions. However, the segmentation accuracy of an iris recognition system significantly influences its performance especially in nonideal iris images. This paper proposes a novel segmentation method for nonideal iris images. Two algorithms are proposed for pupil segmentation in iris images which are captured under visible and near infrared light. Then, a fusion of an expanding and a shrinking active contour is developed for iris segmentation by integrating a new pressure force to the active contour model. Thereafter, a noncircular iris normalization scheme is adopted to effectively unwrap the segmented iris. In addition, a novel method for closed eye detection is proposed. The proposed scheme is robust in finding the exact iris boundary and isolating the eyelids of the iris images. Experimental results on CASIA V4.0, MMU2, UBIRIS V1, and UBIRIS V2 iris databases indicate a high level of accuracy using the proposed technique. Moreover, the comparison results with the state-of-the-art iris segmentation algorithms revealed considerable improvement in segmentation accuracy and recognition performance while being computationally more efficient.

Iris recognition systems are among the most accurate of all biometric technologies with immense potential for use in worldwide security applications. This study examined the effect of eye pathology on iris recognition and in particular whether eye disease could cause iris recognition systems to fail. The experiment involved a prospective cohort of 54 patients with anterior segment eye disease who were seen at the acute referral unit of the Princess Alexandra Eye Pavilion in Edinburgh. Iris camera images were obtained from patients before treatment was commenced and again at follow-up appointments after treatment had been given. The principal outcome measure was that of mathematical difference in the iris recognition templates obtained from patients' eyes before and after treatment of the eye disease. Results showed that the performance of iris recognition was remarkably resilient to most ophthalmic disease states, including corneal oedema, iridotomies (laser puncture of iris) and conjunctivitis. Problems were, however, encountered in some patients with acute inflammation of the iris (iritis/anterior uveitis). The effects of a subject developing anterior uveitis may cause current recognition systems to fail. Those developing and deploying iris recognition should be aware of the potential problems that this could cause to this key biometric technology.

Among biometrics, iris and ocular recognition systems are the methods that recognize eye features in an image. Such iris and ocular regions must have a certain image resolution to achieve a high recognition performance; otherwise, the risk of performance degradation arises. This is even more critical in the case of iris recognition where detailed patterns are used. In cases where such low-resolution images are acquired and the acquisition apparatus and environment cannot be improved, recognition performance can be enhanced by obtaining high-resolution images with methods such as super-resolution reconstruction. However, previous survey papers have mainly summarized studies on high-resolution iris and ocular recognition, but do not provide detailed summaries of studies on low-resolution iris and ocular recognition. Therefore, we investigated high-resolution iris and ocular recognition methods and introduced in detail the low-resolution iris and ocular recognition methods and methods of solving the low-resolution problem. Furthermore, since existing survey papers have focused on and summarized studies on traditional handcrafted feature-based iris and ocular recognition, this survey paper also introduced the latest deep learning-based methods in detail.

We have developed a hybrid speech recognition system which uses a multilayer perceptron (MLP) to estimate the observation likelihoods associated with the states of a HMM. In this paper, we propose two schemes for incorporating distinctive speech features (sonorant, fricative, nasal, vocalic, and voiced) into the MLP component of our system. We show a small improvement in recognition performance on a 160-word speaker-independent continuous-speech Japanese conference room reservation database. Further experiments simulating an improved distinctive feature classifier indicate that this approach can potentially lead to substantial performance improvements.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

A mathematical model for stimulus recognition has been developed that incorporates two processes, and has been applied to the recognition of words in noise. The activation process uses the stimulus information to restrict the possible alternatives to a subset of the total stimulus set. This process employs concepts analogous to signal detection theory but generalized to handle recognition. The decision process assumes that a stimulus is selected from the subset with some biased sampling probability, and that the name of the selected stimulus is then emitted as the recognition response. This process assumes that Luce's choice axiom applies to the sampling probabilities. The model is capable of factoring a confusion matrix into two matrices of sensitivity and bias parameters corresponding to the two processes. Tests with word-recognition data suggest that the model offers a reasonable explanation for the improvement in recognition performance with reduction in stimulus set size and with increase in stimulus probability. The latter is closely related to the word-frequency effect. [Research supported in part by U. S. Public Health Service Grant. The author is a U. S. PHS Predoctoral Fellow.]

Accurately modelling pronunciation variability in conversational speech is an important component of an automatic speech recognition system. We describe some of the projects undertaken in this direction during and after WS97, the Fifth LVCSR Summer Workshop, held at Johns Hopkins University, Baltimore, in July-August, 1997. We first illustrate a use of hand-labelled phonetic transcriptions of a portion of the Switchboard corpus, in conjunction with statistical techniques, to learn alternatives to canonical pronunciations of words. We then describe the use of these alternate pronunciations in an automatic speech recognition system. We demonstrate that the improvement in recognition performance from pronunciation modelling persists as the system is enhanced with better acoustic and language models.

The performance of a speech recognition system may be degraded even without any background noise because of the linear or non-linear distortions incurred by recording devices or reverberations. One of the well-known approaches to reduce this channel distortion is feature mapping which maps the distorted speech feature to its clean counterpart. The feature mapping rule is usually trained based on a set of stereo data which consists of the simultaneous recordings obtained in both the reference and target conditions. In this paper, we propose a novel approach to speech feature sequence mapping based on the switching linear dynamic transducer (SLDT). The proposed algorithm enables us a sequence-to-sequence mapping in a systematic way, instead of the traditional vector-to-vector mapping. The proposed approach is applied to compensate channel distortion in speech recognition and shows improvement in recognition performance.

Speech recognition in digital assistants such as Google Assistant can potentially benefit from the use of conversational context consisting of user queries and responses from the agent. We explore the use of recurrent, Long Short-Term Memory (LSTM), neural language models (LMs) to model the conversations in a digital assistant. Our proposed methods effectively capture the context of previous utterances in a conversation without modifying the underlying LSTM architecture. We demonstrate a 4% relative improvement in recognition performance on Google Assistant queries when using the LSTM LMs to rescore recognition lattices.