Shishoo Textiles in Sport

1. Auflage 2005
ISBN: 978-1-84569-088-5
Verlag: Elsevier Reference Monographs
Format: PDF
Kopierschutz: Adobe DRM (»Systemvoraussetzungen)

E-Book, Englisch, 376 Seiten

The technical developments in the sports clothing industry has resulted in the use of functional textiles for highly-specialised performances in different sports. Developments include thermal and functional properties and coated and laminated clothes. With bio- and smart materials providing such a strong focus in the textile industry generally, companies are going for 'value-added' textiles, such as in-built sensors which monitor performance. In-built wear comfort is a growing market trend and includes clothing which improves the skin's performance. Written by a distinguished editor and a team of authors from the cutting edge of textile research, Textiles in sport discusses high-performance, high-function and intelligent textiles for sportswear.
Invaluable for a broad range of readersDiscusses high-performance, high-function and intelligent textiles for sportswear

Autoren/Hrsg.

Shishoo, Roshan

Inhaltsverzeichnis

1;Front Cover;1
2;Image Processing and Pattern Recognition;4
3;Copyright Page;5
4;Contents;6
5;Contributors;14
6;Preface;16
7;Chapter 1. Pattern Recognition;22
7.1;I. Introduction;22
7.2;II. Pattern Recognition Problem;24
7.3;III. Neural Networks in Feature Extraction;32
7.4;IV. Classification Methods: Statistical and Neural;41
7.5;V. Neural Network Applications in Pattern Recognition;59
7.6;VI. Summary;73
7.7;References;74
8;Chapter 2. Comparison of Statistical and Neural Classifiers and Their Applications to Optical Character Recognition and Speech Classification;82
8.1;I. Introduction;82
8.2;II. Applications;84
8.3;III. Data Acquisition and Preprocessing;85
8.4;IV. Statistical Classifiers;86
8.5;V. Neural Classifiers;95
8.6;VI. Literature Survey;100
8.7;VII. Simulation Results;102
8.8;VIII. Conclusions;106
8.9;References;107
9;Chapter 3. Medical Imaging;110
9.1;I. Introduction;110
9.2;II. Review of Artificial Neural Network Applications in Medical Imaging ;116
9.3;III. Segmentation of Arteriograms;120
9.4;IV. Back-Propagation Artificial Neural Network for Arteriogram Segmentation: A Supervised Approach;122
9.5;V. Self-Adaptive Artificial Neural Network for Arteriogram Segmentation: An Unsupervised Approach;128
9.6;VI. Conclusions;145
9.7;References;150
10;Chapter 4. Paper Currency Recognition;154
10.1;I. Introduction;154
10.2;II. Small-Size Neuro-Recognition Technique Using the Masks;155
10.3;III. Mask Determination Using the Genetic Algorithm;164
10.4;IV. Development of the Neuro-Recognition Board Using the Digital Signal Processor;173
10.5;V. Unification of Three Core Techniques;177
10.6;VI. Conclusions;179
10.7;References;180
11;Chapter 5. Neural Network Classification Reliability: Problems and Applications;182
11.1;I. Introduction;182
11.2;II. Classification Paradigms;185
11.3;III. Neural Network Classifiers;188
11.4;IV. Classification Reliability;193
11.5;V. Evaluating Neural Network Classification Reliability;195
11.6;VI. Finding a Reject Rule;199
11.7;VII. Experimental Results;206
11.8;VIII. Summary;217
11.9;References;218
12;Chapter 6. Parallel Analog Image Processing: Solving Regularization Problems with Architecture Inspired by the Vertebrate Retinal Circuit;222
12.1;I. Introduction;222
12.2;II. Physiological Background;223
12.3;III. Regularization Vision Chips;242
12.4;IV. Spatio-Temporal Stability of Vision Chips;285
12.5;References;304
13;Chapter 7. Algorithmic Techniques and Their Applications;308
13.1;I. Introduction;308
13.2;II. Quasi-Newton Methods for Neural Network Training;310
13.3;III. Selecting the Number of Output Units;316
13.4;IV. Determining the Number of Hidden Units;317
13.5;V. Selecting the Number of Input Units;324
13.6;VI. Determining the Network Connections by Pruning;330
13.7;VII. Applications of Neural Networks to Data Mining;334
13.8;VIII. Summary;337
13.9;References;338
14;Chapter 8. Learning Algorithms and Applications of Principal Component Analysis;342
14.1;I. Introduction;342
14.2;II. Adaptive Learning Algorithm;345
14.3;III. Simulation Results;356
14.4;IV. Applications;364
14.5;V. Conclusion;370
14.6;VI. Appendix;371
14.7;References;372
15;Chapter 9. Learning Evaluation and Pruning Techniques;374
15.1;I. Introduction;374
15.2;II. Complexity Regularization;378
15.3;III. Sensitivity Calculation;383
15.4;IV. Optimization through Constraint Satisfaction;389
15.5;V. Local and Distributed Bottlenecks;393
15.6;VI. Interactive Pruning;395
15.7;VII. Other Pruning Methods;397
15.8;VIII. Concluding Remarks;399
15.9;References;399
16;Index;404

Leseproben

Preface
Cornelius T. Leondes Inspired by the structure of the human brain, artificial neural networks have been widely applied to fields such as pattern recognition, optimization, coding, control, etc., because of their ability to solve cumbersome or intractable problems by learning directly from data. An artificial neural network usually consists of a large number of simple processing units, i.e., neurons, via mutual interconnection. It learns to solve problems by adequately adjusting the strength of the interconnections according to input data. Moreover, the neural network adapts easily to new environments by learning, and can deal with information that is noisy, inconsistent, vague, or probabilistic. These features have motivated extensive research and developments in artificial neural networks. This volume is probably the first rather comprehensive treatment devoted to the broad areas of algorithms and architectures for the realization of neural network systems. Techniques and diverse methods in numerous areas of this broad subject are presented. In addition, various major neural network structures for achieving effective systems are presented and illustrated by examples in all cases. Numerous other techniques and subjects related to this broadly significant area are treated. The remarkable breadth and depth of the advances in neural network systems with their many substantive applications, both realized and yet to be realized, make it quite evident that adequate treatment of this broad area requires a number of distinctly titled but well-integrated volumes. This is the fifth of seven volumes on the subject of neural network systems and it is entitled Image Processing and Pattern Recognition. The entire set of seven volumes contains Volume 1: Algorithms and Architectures Volume 2: Optimization Techniques Volume 3: Implementation Techniques Volume 4: Industrial and Manufacturing Systems Volume 5: Image Processing and Pattern Recognition Volume 6: Fuzzy Logic and Expert Systems Applications Volume 7: Control and Dynamic Systems The first contribution to this volume is “Pattern Recognition,” by Jouko Lampinen, Jorma Laaksonen, and Erkki Oja. Pattern recognition (PR) is the science and art of giving names to the natural objects in the real world. It is often considered part of artificial intelligence. However, the problem here is even more challenging because the observations are not in symbolic form and often contain much variability and noise. Another term for PR is artificial perception. Typical inputs to a PR system are images or sound signals, out of which the relevant objects have to be found and identified. The PR solution involves many stages such as making the measurements, preprocessing and segmentation, finding a suitable numerical representation for the objects we are interested in, and finally classifying them based on these representations. Presently, there are a growing number of applications for pattern recognition. A leading motive from the very start of the field has been to develop user-friendly and flexible user interfaces that understand speech and handwriting. Only recently have these goals become possible with the highly increased computing power of workstations. Document processing is emerging as a major application. In industrial problems as well as in biomedicine, automatic analysis of images and signals can be achieved with PR techniques. Remote sensing is routinely using automated recognition techniques, too. This contribution is a rather comprehensive presentation of the techniques and methods of neural network systems in pattern recognition. Several substantive examples are included. It is also worth noting as a valuable feature of this contribution that almost 200 references, which have been selectively culled from the literature, are included in the reference list. The next contribution is “Comparison of Statistical and Neural Classifiers and Their Applications to Optical Character Recognition and Speech Classification,” by Ethem Alpaydin and Fikret Giirgen. Improving person–machine communication leads to wider use of advanced information technologies. Toward this aim, character recognition and speech recognition are two applications whose automatization allows easier interaction with a computer. As they are the basic means of person-to-person communication, they are known by everyone and require no special training. Speech in particular is the most natural form of human communication and writing is the tool by which humanity has stored and transferred its knowledge for millennia. In a typical pattern recognition system, the first step is the acquisition of data. These raw data are preprocessed to suppress noise and normalize input. Features are those parts of the signal that carry information salient to its identity, and their extraction is an abstraction operation where the important information is extracted and the irrelevant is discarded. Classification is assignment of the input as an element of one of a set of predefined classes. The rules for classification are generally not known exactly and thus are estimated. A classifier is written as a parametric model whose parameters are computed using a given training sample to optimize particular error criterion. Approaches for classification differ in their assumptions about the model, in the way parameters are computed, or in the error criterion they optimize. This contribution treats what are probably the two principle approaches to classifiers as embodied by neural and statistical classifiers, and applies them to the major areas of optical character recognition and speech recognition. Illustrative examples are included as well as the literature for the two application categories. The next contribution is “Medical Imaging,” by Ying Sun and Reza Nekovei. The history of medical imaging began a century ago. The landmark discovery of X-rays by Wilhelm Conrad Röntgen in 1895 ushered in the development of noninvasive methods for visualization of internal organs. The birth of the digital computer in 1946 brought medical imaging into a new era of computer-assisted imagery. During the second half of the 20th century, medical imaging technologies have diversified and advanced at an accelerating rate. Today, clinical diagnostics rely heavily on the various medical imaging systems. In addition to conventional X-ray radiography, computer-assisted tomography and magnetic resonance imaging produce two-dimensional cross sections and three-dimensional imagery of the internal organs that drastically improve our capability to diagnose various diseases. X-ray angiography used in cardiac catheterization laboratories allows us to detect stenoses in the coronary arteries and guide treatment procedures such as balloon angioplasty and cardiac ablation. Ultrasonography has become a routine procedure for fetal examination. Two-fetal dimensional echocardiography combined with color Doppler flow imaging has emerged as a powerful and convenient tool for diagnosing heart valve abnormalities and for assessing cardiac functions. In the area of nuclear medicine, the scintillation gamma camera provides two-dimensional images of pharmaceuticals labeled by radioactive isotopes. Single photon emission computed tomography and positron emission tomography further allow for three-dimensional imaging of radioactive tracers. This contribution is a rather in-depth treatment of the important role neural network system techniques can play in the greatly significant area of medical imaging systems. Two major application areas are treated, i.e., detection of blood vessels in angiograms and image segmentation. The next contribution is “Paper Currency Recognition,” by Fumiaki Takeda and Sigeru Omatu. Three core techniques are presented. The first is the small size neurorecognition technique using masks. The second is the mask determination technique using the genetic algorithm. The third is the neurorecognition board technique using the digital signal processor. Unification of these three techniques demonstrates that realization of neurorecognition machines capable of transacting various kinds of paper currency is feasible. The neurosystem technique enables acceleration in the commercialization of a new type of banking machine in a short period and in a few trials. Furthermore, this technique will be effective for various kinds of recognition applications owing to its high recognition ability, high speed transaction, short developing period, and reasonable cost. It can be presumed that it is so effective that it applies not only to paper currency and coins, but also to handwritten symbols such as electron systems or questionnaires. The next contribution is “Neural Network Classification Reliability: Problems and Applications,” by Luigi P. Cordelia, Carlo Sansone, Francesco Tortorella, and Claudio De Stefano. Classification is a process according to which an entity is attributed to one of a finite set of classes or, in other words, it is recognized as belonging to a set of equal or similar entities, possibly identified by a name. In the framework of signal and image analysis, this process is generally considered part of a more complex process referred to as pattern recognition. In its simplest and still most commonly followed approach, a pattern recognition system is made of two distinct parts: 1. A description unit, whose input is the entity to be recognized, represented in...