لایب سان

Knowledge-Based Intelligent Techniques in Character Recognition:An Introduction to Handwritten Character and Word Recognition Previous Table of Contents Next Chapter 1 An Introduction to Handwritten Character and Word Recognition L.C. Jain Knowledge-Based Intelligent Engineering Systems Centre University of South Australia Adelaide, Mawson Lakes, SA 5095, Australia B. Lazzerini Dipartimento di Ingegneria della Informazione Università degli Studi di Pisa Via Diotisalvi 2, 56126 Pisa, Italy There is a worldwide interest in the development of handwritten character and word recognition systems. These systems are used in many situations such as recognition of postcodes, interpretation of amount and verification of signature on bank checks, and law enforcement on public roads. The tremendous advances in the computational intelligence techniques have provided new tools for the development of intelligent character recognition systems. This chapter introduces principles of handwriting recognition for on-line and off-line systems. Some considerations about performance evaluation of a handwriting recognition system are discussed. 1 Introduction In the last few years many academic institutions and industrial companies have been involved in the field of handwriting recognition. The automatic recognition of handwritten text can be extremely useful in many applications where it is necessary to process large volumes of handwritten data, such as recognition of addresses and postcodes on envelopes, interpretation of amounts on bank checks, document analysis, and verification of signatures [1]. Substantial progress has been recently achieved, but the recognition of handwritten text cannot yet approach human performance. The major difficulties descend from the variability of someone's calligraphy over time, the similarity of some characters with each other, and the infinite variety of character shapes and writing styles produced by different writers. Furthermore, the possible low quality of the text image, the unavoidable presence of background noise and various kinds of distortions (such as poorly written, degraded, or overlapping characters) can make the recognition process even more difficult. Therefore, handwriting recognition is still an open and interesting area for research and novel ideas.

Knowledge-Based Intelligent Techniques in Character Recognition:Table of Contents Preface Chapter 1—An Introduction to Handwritten Character and Word Recognition 1 Introduction 2 Knowledge-Based Intelligent Techniques 3 Off-Line Handwriting Recognition 3.1 Preprocessing 3.2 Segmentation 3.3 Recognition 3.3.1 Analytical Methods 3.3.2 Holistic Methods 3.3.3 Recognition-Based Methods 3.3.4 Character Recognition 4 On-Line Handwriting Recognition 5 Some Considerations about Performance Evaluation 6 Summary Chapter 2—Recognition of Handwritten Digits in the Real World by a Neocognitron 1. Introduction 2. Neocognitron 2.1 S-cell Layer 2.2 C-cell Layer 2.3 Detailed Network Architecture of Lower Stages 3. Unsupervised Learning of Feature-Extracting Cells 4. Threshold of S-cells 4.1 Mathematical Notation of an S-cell 4.2 Relationship Between Threshold and Selectivity of an S-cell 4.3 Calculation of Inhibitory Variable Connection 4.4 Network with High Thresholds for S-cells 4.5 Network with a Lower Threshold 5. Different Thresholds in Learning and Recognition 5.1 Threshold in Learning Phase 5.2 Upper Bound of the Threshold 6. The Highest Stage; Classification Layer 7. Results 7.1 Optimization of Thresholds and Recognition Rate file:///F|/htdocs/ICE/temp/20030703/knowledge-based-intelligent-techniques-in-character-recognition/ewtoc.html (1 of 5)2003-11-8 10:36:17 Knowledge-Based Intelligent Techniques in Character Recognition:Table of Contents 8. Summary Chapter 3—Recognition of Rotated Patterns Using a Neocognitron 1 Introduction 2 Rotation-Invariant Neocognitron 2.1 Short Review of Neocognitron 2.2 Structure of Rotation-Invariant Neocognitron 2.3 Unsupervised Learning and Detection of Rotated Patterns 2.4 Robustness for Rotations 3 Simulation 4 Summary Chapter 4—A Soft Computing Approach to Handwritten Numeral Recognition 1. Introduction 2. Description of Data and Preprocessing 3. Extracting Rules from Data 3.1 Essentials of Mass Assignment Theory 3.2 Semantic Unification 3.3 Fuzzy Sets from Data 3.4 Evidential Logic Rule of Combination 4. Choice of Features 4.1 Automatic Feature Discovery 4.2 Compound Features 5. Implementation and Results 6. Summary Chapter 5—Handwritten Character Recognition Using a MLP 1. Introduction 2. The Adopted MLP Description 2.1 The Learning Algorithm 2.2 The Activation Functions 3. The Handwritten Character Recognition Task 3.1 The Preprocessing Phase 3.2 The Character Processing Phase 3.2.1 The Training Phase 3.2.2 The Testing Phase 4. Digital Implementation 5. Experimental Results 6. Summary file:///F|/htdocs/ICE/temp/20030703/knowledge-based-intelligent-techniques-in-character-recognition/ewtoc.html (2 of 5)2003-11-8 10:36:17 Knowledge-Based Intelligent Techniques in Character Recognition:Table of Contents Chapter 6—Signature Verification Based on a Fuzzy Genetic Algorithm 1. Introduction 2. Background 2.1 Problems Associated with Signature Verification 2.2 Feature Selection Algorithms 3. The Fuzzy Genetic Algorithm 4. Feature Selection Problems 5. Experiments with Signature Data 5.1 Original Handwritten Signature Set 5.2 Normalized Handwritten Signature Set 5.3 Facial Signature Set 6. Simulation Results and Discussions 7. Summary Chapter 7—The Application of a Generic Neural Network to Handwritten Digit Classification 1. Introduction 2. The PARADISE Network 2.1 The Feature Extraction Layer 2.2 The Pattern Detection Layer 2.3 The Classification Layer 2.4 Network Parameters 2.5 Network Dynamics 3. Applying the Network 3.1 The Image Data Set 3.2 Application-Specific Details 3.3 Gaussian Feature Extraction Results 3.4 Gabor Feature Extraction Results 3.5 Analysis of Component Patterns 4. Summary Chapter 8—High-Speed Recognition of Handwritten Amounts on Italian Checks 1 Introduction 1.1 Hardware Platform 2 Image Preprocessor 3 Neural Subsystem file:///F|/htdocs/ICE/temp/20030703/knowledge-based-intelligent-techniques-in-character-recognition/ewtoc.html (3 of 5)2003-11-8 10:36:17 Knowledge-Based Intelligent Techniques in Character Recognition:Table of Contents 3.1 Centering Detection 3.2 Pseudo-characters 3.3 Character Recognizer 3.4 Characterization of the Character Recognizer 4 Clustering Subsystem 5 Context Analysis Subsystem 6 Performance Evaluation 7 Summary Chapter 9—Off-Line Handwritten Word Recognition Using Hidden Markov Models 1 Introduction 2 Hidden Markov Models 2.1 The Evaluation Problem 2.2 The Decoding Problem 2.3 The Training Problem 3 Representation of Word Images 3.1 Preprocessing 3.2 Character Segmentation of Words 3.3 Feature Extraction 4 Markovian Modeling of Handwritten Words 4.1 HMM Use in Handwritten Word Recognition 4.2 The Proposed Model 4.3 The Learning Phase 4.4 The Recognition Phase 5 Experiments 6 Rejection Summary Chapter 10—Off-Line Handwriting Recognition with Context-Dependent Fuzzy Rules 1 Methodology and Summary of Previous Work 2 Fuzzy Modeling of Pattern Description Parameters 3 Feature Extraction and Aggregation 3.1 Positional Features 3.2 Discrimination of Straight Lines and Curved Lines 3.3 Class of Fuzzy Curved Lines 3.4 Combination of Features with Fuzzy Aggregation file:///F|/htdocs/ICE/temp/20030703/knowledge-based-intelligent-techniques-in-character-recognition/ewtoc.html (4 of 5)2003-11-8 10:36:17 Knowledge-Based Intelligent Techniques in Character Recognition:Table of Contents 4 Automatic Pattern Description Paradigm 4.1 Multiphased Clustering 4.1.1 Sample Grouping 4.1.2 Reduction of Clustering Features 4.1.3 Multiphased Clustering - the Process 4.1.4 Rule Generation in FOHDEL 4.2 Rule Cross-Checking and Refinement 5 Word Classification with Context Dependence 6 Application and Concluding Discussions Chapter 11—License Plate Recognition 1 Introduction 2 Segmentation and Isolation 2.1 Soft Segmentation of License Plates 2.2 Introducing Size Constraints 2.3 Binarization 2.4 Character Isolation 3 Feature-Based Recognition 3.1 Moments 3.2 Principal Component Analysis (PCA) 3.3 Projection 3.4 Combining Classifiers 4 Other Means of Recognition 4.1 Template Matching (LVQ) 4.2 Syntax Trees 4.3 The Overall Architecture 5 VIPUR Architecture and Development 5.1 System Synthesis 5.2 System Verification 5.3 Evaluation 6 Summary Index Copyright © CRC Press LLC file:///F|/htdocs/ICE/temp/20030703/knowledge-based-intelligent-techniques-in-character-recognition/ewtoc.html (5 of 5)2003-11-8 10:36:17

برای ارسال نظر لطفا وارد شوید یا ثبت نام کنید