[...]... Image Recognition 4 .1 Wavelets 4.2 Wavelet Packet Analysis 4.3 Feature Extraction by Wavelet Packet Analysis 4.4 Hidden Markov Models 4.5 Feature Classification by Hidden Markov Models 4.6 Questions to Ponder 4.7 Experimental Results 5 Discussion References ix 17 2 17 3 17 7 17 7 17 8 18 0 18 0 18 1 18 3 18 4 18 5 18 9 19 1 19 2 19 2 19 4 19 4 19 5 19 8 19 8 2 01 2 01 202 202 203 204 204 205 205 206 206 207 210 218 218 220... (1. 5c) 2 (1. 5d) M30 + M12 − 3 M 21 + M03 2 2 + 3M12 − M03 M 21 + M03 ∗ 3 M30 + M12 2 − M 21 + M03 6 = M20 − M02 M30 + M12 2 − M 21 + M03 2 2 (1. 5f) + 4M 11 M30 + M12 M 21 + M03 7 = 3M 21 − M03 M30 + M12 (1. 5e) M30 + M12 2 − 3 M 21 + M03 2 + 3 M 21 − M03 M 21 + M03 ∗ 3 M30 + M12 2 − M 21 + M03 2 (1. 5g) Recognition Strategy 11 Table 1. 2 Moment invariants represented by a 4 × 4 array M 0 1 0 1 2 3 X X √ √ X √ √ X... 1 2 3 X X √ √ X √ √ X 2 √ √ 3 √ X X X X X Table 1. 3 Moment invariant values for some characters ¨ 1 2 3 4 5 6 7 1. 180 −4.027 −7.293 −6.459 13 .338 −8.576 13 .17 5 1. 422 −5.823 −9.088 −8.6 21 17 .739 11 .578 17 .404 1. 595 −4.790 −7.642 10 .19 0 19 .18 1 12 .584 −20. 016 1. 536 −6.244 −8.935 13 .024 −24.368 16 .325 −23.9 81 0.375 0.636 −2.668 −2.804 −5.5 41 −2.582 −9.058 These functions can be normalized... (1. 1) and (1. 2) are approximated by the summations: m n Mpq = x−x p y − y q f x y dxdy (1. 3) x=0 y=0 m n M pq = xp yq f x y dxdy (1. 4) x=0 y=0 where m and n are dimensions of the image The set of moment invariants that has been used by Hu are given by: 1 = M20 + M02 2 = M20 − M02 3 = M30 − 3M12 2 + 3M 21 − M03 4 = M30 + M12 2 + M 21 + M03 5 = M30 − 3M12 M30 + M12 (1. 5a) 2 2 + 4M 11 (1. 5b) 2 (1. 5c) 2 (1. 5d)... Complex MLP and Quaternionic MLP xiii 379 379 380 3 81 382 382 382 384 384 384 385 385 385 386 386 387 389 389 390 392 392 392 393 394 395 397 397 398 400 400 4 01 403 407 408 408 411 411 412 412 413 414 414 415 416 417 xiv Contents 5 Clifford-valued Feed-forward Neural Networks 5 .1 The Activation Function 5.2 The Geometric Neuron 5.3 Feed-forward Clifford-valued Neural Networks 6 Learning Rule 6 .1 Multidimensional... 249 249 250 2 51 252 253 253 253 253 254 254 257 257 259 2 61 263 265 268 2 71 272 273 275 275 276 280 280 280 2 81 2 81 2 81 Contents 6 Region Identification Based on Fuzzy Logic 6 .1 Experimental Results 7 Conclusions References 16 Feature Extraction and Compression with Discriminative and Nonlinear Classifiers and Applications in Speech Recognition Xuechuan Wang (Canada) 1 Introduction 2 Standard Feature... Measures and Their Extraction 4 .1 IR Measures 4.2 GPR Measures 4.3 MD Measures 5 Modeling of Measures in Terms of Belief Functions and Their Discounting 5 .1 IR Measures 5.2 GPR A-scan and Preprocessed C-scan Measures 5.3 GPR B-scan (Hyperbola) Measures 5.4 MD Measures 5.5 Discounting Factors xi 285 289 293 294 297 298 300 300 3 01 3 01 3 01 303 304 304 306 307 307 307 308 309 311 316 317 319 319 320 320 3 21. .. academia and industry The twenty-four chapters within this book will cover computer-aided intelligent recognition techniques, applications, systems and tools The book is planned to be of most use to researchers, computer scientists, practising engineers, and many others who seek state-of-the-art techniques, applications, systems and tools for intelligent recognition It will also be equally and extremely... between man Computer-Aided Intelligent Recognition Techniques and Applications © 2005 John Wiley & Sons, Ltd Edited by M Sarfraz 2 Offline Arabic Character Recognition and machine Some of the applications of OCR include automatic processing of data, check verification and a large variety of banking, business and scientific applications OCR provides the advantage of little human intervention and higher... character ‘ ’ Recognition Strategy 13 database The extracted character, after normalization, is matched with all the characters in the database using a Hamming distance approach This approach is shown in Equations (1. 7) and (1. 8) nrows ncols mismatchi j (1. 7) i =1 j =1 where mismatchi j = 1 0 if originali j = extractedi j if originali j = extractedi j (1. 8) where nrows and ncols are the number of rows and columns . Verification 11 2 4. Proposed Online Signature Verification Applications 11 3 4 .1 System Password Authentication 11 3 4.2 Internet E-commerce Application 11 4 5. Conclusions 11 6 References 11 6 8. Hybrid. Arabia COMPUTER-AIDED INTELLIGENT RECOGNITION TECHNIQUES AND APPLICATIONS COMPUTER-AIDED INTELLIGENT RECOGNITION TECHNIQUES AND APPLICATIONS Edited by Muhammad Sarfraz King Fahd University of Petroleum and Minerals,. Fingerprint Recognition using Minutiae and Shape 11 9 Asker Bazen, Raymond Veldhuis and Sabih Gerez (The Netherlands) 1. Introduction 11 9 2. Elastic Deformations 12 0 3. Elastic Minutiae Matching 12 2 3.1