A wide variety of systems requires reliable personal recognition schemes to either confirm or determine the identity an individual requesting their services. The purpose such is ensure that rendered services are accessed only by a legitimate user and no one else. Examples applications include secure access buildings, computer systems, laptops, cellular phones, ATMs. In absence robust schemes, these vulnerable wiles impostor. Biometric recognition, or, simply, biometrics, refers automatic...

A large number of algorithms have been proposed for feature subset selection. Our experimental results show that the sequential forward floating selection algorithm, by Pudil et al. (1994), dominates other tested. We study problem choosing an optimal set land use classification based on SAR satellite images using four different texture models. Pooling features derived from models, followed a in substantial improvement accuracy. also illustrate dangers small sample size situations.

In order to ensure that the performance of an automatic fingerprint identification/verification system will be robust with respect quality input images, it is essential incorporate a enhancement algorithm in minutiae extraction module. We present fast algorithm, which can adaptively improve clarity ridge and valley structures images based on estimated local orientation frequency. have evaluated image using goodness index extracted accuracy online verification system. Experimental results...

This paper proposes an unsupervised algorithm for learning a finite mixture model from multivariate data. The adjective "unsupervised" is justified by two properties of the algorithm: 1) it capable selecting number components and 2) unlike standard expectation-maximization (EM) algorithm, does not require careful initialization. proposed method also avoids another drawback EM fitting: possibility convergence toward singular estimate at boundary parameter space. novelty our approach that we...

Human face detection plays an important role in applications such as video surveillance, human computer interface, recognition, and image database management. We propose a algorithm for color images the presence of varying lighting conditions well complex backgrounds. Based on novel compensation technique nonlinear transformation, our method detects skin regions over entire then generates candidates based spatial arrangement these patches. The constructs eye, mouth, boundary maps verifying...

We consider a texture to be stochastic, possibly periodic, two-dimensional image field. A model is mathematical procedure capable of producing and describing textured image. explore the use Markov random fields as models. The binomial model, where each point in has distribution with parameter controlled by its neighbors ``number tries'' equal number gray levels, was taken basic for analysis. method generating samples from given, followed theoretical practical analysis method's convergence....

The effects of sample size on feature selection and error estimation for several types classifiers are discussed. focus is the two-class problem. Classifier design in context small explored. rates under test given. Sample Recommendations choice learning sizes In addition to surveying prior work this area, an emphasis placed giving practical advice designers users statistical pattern recognition systems.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML"...

Fingerprint verification is one of the most reliable personal identification methods. However, manual fingerprint incapable meeting today's increasing performance requirements. An automatic system (AFIS) needed. This paper describes design and implementation an online which operates in two stages: minutia extraction matching. improved version algorithm proposed by Ratha et al. (1995), much faster more reliable, implemented for extracting features from input image captured with inkless...

With identity fraud in our society reaching unprecedented proportions and with an increasing emphasis on the emerging automatic personal identification applications, biometrics-based verification, especially fingerprint-based identification, is receiving a lot of attention. There are two major shortcomings traditional approaches to fingerprint representation. For considerable fraction population, representations based explicit detection complete ridge structures difficult extract...

We explore the idea of evidence accumulation (EAC) for combining results multiple clusterings. First, a clustering ensemble--a set object partitions, is produced. Given data (n objects or patterns in d dimensions), different ways producing partitions are: 1) applying algorithms and 2) same algorithm with values parameters initializations. Further, combinations representations (feature spaces) can also provide multitude significantly partitionings. propose simple framework extracting...

Establishing identity is becoming critical in our vastly interconnected society. Questions such as "Is she really who claims to be?," this person authorized use facility?," or he the watchlist posted by government?" are routinely being posed a variety of scenarios ranging from issuing driver's license gaining entry into country. The need for reliable user authentication techniques has increased wake heightened concerns about security and rapid advancements networking, communication,...

Biometrics offers greater security and convenience than traditional methods of personal recognition. In some applications, biometrics can replace or supplement the existing technology. others, it is only viable approach. But how secure biometrics? And what are privacy implications?.

Fingerprint verification is an important biometric technique for personal identification. We describe the design and implementation of a prototype automatic identity-authentication system that uses fingerprints to authenticate identity individual. have developed improved minutiae-extraction algorithm faster more accurate than our earlier (1995). An alignment-based minutiae-matching has been proposed. This capable finding correspondences between input minutiae stored template without...