Sin Ka Wai. / Thesis (M.Sc.)--Chinese University of Hong Kong, 1992. / Includes bibliographical references (leaves [81]-[82]). / PREFACE / ABSTRACT / CONTENT / Chapter §1. --- INTRODUCTION / Chapter §1.1 --- The Chinese language --- p.1 / Chapter §1.2 --- Chinese information processing system --- p.2 / Chapter §1.3 --- Chinese character recognition --- p.4 / Chapter §1.4 --- Multi-stage tree classifier Vs Single-stage tree classifier in Chinese character recognition --- p.6 / Chapter §1.5 --- Decision Tree / Chapter §1.5.1 --- Basic Terminology of a decision tree --- p.7 / Chapter §1.5.2 --- Structure design of a decision tree --- p.10 / Chapter §1.6 --- Motivation of the project --- p.12 / Chapter §1.7 --- Objects of the project --- p.14 / Chapter §1.8 --- Development environment --- p.14 / Chapter §2. --- APPROACH 1 - UNSUPERVISED LEARNING / Chapter §2.1 --- Idea --- p.15 / Chapter §2.2 --- Feature Extraction / Chapter §2.2.1 --- Feature selection criteria --- p.15 / Chapter §2.2.2 --- 4C code --- p.20 / Chapter §2.2.3 --- Regional code --- p.22 / Chapter §2.2.4 --- Walsh Transform --- p.24 / Chapter §2.2.5 --- Black dot density projection profile --- p.26 / Chapter §2.2.6 --- Corner features --- p.28 / Chapter §2.3 --- Clustering Method -K-MEANS & Other Algorithms --- p.32 / Chapter §2.4 --- Pros & Cons --- p.35 / Chapter §2.5 --- Decision Table --- p.37 / Chapter §2.6 --- The Optimum Classifier & its Implemen- tation difficulties --- p.39 / Chapter §3. --- APPROACH 2 - SUPERVISED LEARNING --- p.43 / Chapter §4. --- POSSIBLE IMPROVEMENT / Chapter §4.1 --- Training and Test Sample Reduction --- p.46 / Chapter §4.2 --- Noise Filtering --- p.46 / Chapter §4.3 --- Decision with Overlapping --- p.52 / Chapter §4.4 --- Back Tracking for Holes --- p.56 / Chapter §4.5 --- Fuzzy Decision with Tolerance Limit --- p.59 / Chapter §4.6 --- Different Tree Architecture --- p.63 / Chapter §4.7 --- Building Decision Tree by Entropy Reduction Method --- p.65 / Chapter §5. --- EXPERIMENTAL RESULTS & THE IMPROVED MULTISTAGE CLASSIFIER / Chapter §5.1 --- Experimental Results --- p.70 / Chapter §5.2 --- Conclusion --- p.81 / Chapter §6. --- IMPROVED MULTISTAGE TREE CLASSIFIER / Chapter §6.1 --- The Optimal Multistage Tree Classifier --- p.83 / Chapter §6.2 --- Performance Analysis --- p.84 / Chapter §7. --- FURTHER DISCRIMINATION BY CONTEXT CONSIDERATION --- p.87 / Chapter §8. --- CONCLUSION / Chapter §8.1 --- Advantage of the Classifier --- p.89 / Chapter §8.2 --- Limitation of the Classifier --- p.90 / Chapter §9. --- AREA OF FUTURE RESEARCH AND IMPROVEMENT / Chapter §9.1 --- Detailed Analysis at Each Terminal Node --- p.91 / Chapter §9.2 --- Improving the Noise Filtering Technique --- p.92 / Chapter §9.3 --- The Use of 4 Corner Code --- p.93 / Chapter §9.4 --- Increase in the Dimension of the Feature Space --- p.95 / Chapter §9.5 --- 1-Tree Protocol with Entropy Reduction --- p.96 / Chapter §9.6 --- The Use of Human Intelligence --- p.97 / APPENDICES / Chapter A.1 --- K-MEANS / Chapter A.2 --- Maximum Distance Algorithm & ISODATA Algorithm / Chapter A.3 --- Approach Two - Supervised Learning / Chapter A.4 --- Theories on Statistical Discriminant Analysis / Chapter A.5 --- An Example of Misclassification Table / Chapter A.6 --- "Listing of the Program ""CHDIS.C""" / Chapter A.7 --- Further Discrimination by Context Consideration / Chapter A.8 --- Passage used in Testing the Performance of the Classifier with Context Consideration / Chapter A.9 --- A Partial List of Semantically Related Chinese Characters / REFERENCE

by Kong Shao-hua. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1992. / Includes bibliographical references (leaves 113-116). / Chapter CHAPTER 1 --- INTRODUCTION --- p.1 / Chapter CHAPTER 2 --- REVIEW OF 3-D OBJECT RECOGNITION SYSTEMS --- p.8 / Chapter CHAPTER 3 --- FEATURE EXTRACTION AND OBJECT REPRESEN- TATION --- p.16 / Chapter 3.1 --- Preprocessing --- p.18 / Chapter 3.2 --- Extraction of Characteristic Points --- p.20 / Chapter 3.3 --- Characterization of Surface Patches --- p.28 / Chapter 3.4 --- Object Representation --- p.37 / Chapter 3.5 --- Model Formation --- p.42 / Chapter CHAPTER 4 --- OBJECT RECOGNITION AND OBJECT LOCATION AND ORIENTATION DETERMINATION --- p.45 / Chapter 4.1 --- RBM-Matching --- p.48 / Chapter 4.1.1 --- Rigid body model (RBM) --- p.48 / Chapter 4.1.2 --- RBM-matching --- p.55 / Chapter 4.2 --- Estimation of the Transformation Parameters --- p.63 / Chapter 4.3 --- Recognition Decision Making --- p.72 / Chapter CHAPTER 5 --- EXPERIMENTATION --- p.80 / Chapter 5.1 --- Automatic Model Building --- p.82 / Chapter 5.2 --- Recognition of Single Objects --- p.88 / Chapter 5.3 --- Recognition of Multiple Objects with Occlusion --- p.103 / Chapter CHAPTER 6 --- CONCLUSION AND DISCUSSION --- p.109 / REFERENCES --- p.113

by Yingli Tian. / Thesis (Ph.D.)--Chinese University of Hong Kong, 1996. / Includes bibliographical references (leaves 202-222). / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Physics-Based Shape Recovery Techniques --- p.3 / Chapter 1.2 --- Proposed Approaches to Shape Recovery in this Thesis --- p.9 / Chapter 1.3 --- Thesis Outline --- p.13 / Chapter 2 --- Camera Model in Color Vision --- p.15 / Chapter 2.1 --- Introduction --- p.15 / Chapter 2.2 --- Spectral Linearization --- p.17 / Chapter 2.3 --- Image Balancing --- p.21 / Chapter 2.4 --- Spectral Sensitivity --- p.24 / Chapter 2.5 --- Color Clipping and Blooming --- p.24 / Chapter 3 --- Extended Light Source Models --- p.27 / Chapter 3.1 --- Introduction --- p.27 / Chapter 3.2 --- A Spherical Light Model in 2D Coordinate System --- p.30 / Chapter 3.2.1 --- Basic Photometric Function for Hybrid Surfaces under a Point Light Source --- p.32 / Chapter 3.2.2 --- Photometric Function for Hybrid Surfaces under the Spher- ical Light Source --- p.34 / Chapter 3.3 --- A Spherical Light Model in 3D Coordinate System --- p.36 / Chapter 3.3.1 --- Radiance of the Spherical Light Source --- p.36 / Chapter 3.3.2 --- Surface Brightness Illuminated by One Point of the Spher- ical Light Source --- p.38 / Chapter 3.3.3 --- Surface Brightness Illuminated by the Spherical Light Source --- p.39 / Chapter 3.3.4 --- Rotating the Source-Object Coordinate to the Camera- Object Coordinate --- p.41 / Chapter 3.3.5 --- Surface Reflection Model --- p.44 / Chapter 3.4 --- Rectangular Light Model in 3D Coordinate System --- p.45 / Chapter 3.4.1 --- Radiance of a Rectangular Light Source --- p.45 / Chapter 3.4.2 --- Surface Brightness Illuminated by One Point of the Rect- angular Light Source --- p.47 / Chapter 3.4.3 --- Surface Brightness Illuminated by a Rectangular Light Source --- p.47 / Chapter 4 --- Shape Recovery from Specular Reflection --- p.54 / Chapter 4.1 --- Introduction --- p.54 / Chapter 4.2 --- Theory of the First Method --- p.57 / Chapter 4.2.1 --- Torrance-Sparrow Reflectance Model --- p.57 / Chapter 4.2.2 --- Relationship Between Surface Shapes from Different Images --- p.60 / Chapter 4.3 --- Theory of the Second Method --- p.65 / Chapter 4.3.1 --- Getting the Depth of a Reference Point --- p.65 / Chapter 4.3.2 --- Recovering the Depth and Normal of a Specular Point Near the Reference Point --- p.67 / Chapter 4.3.3 --- Recovering Local Shape of the Object by Specular Reflection --- p.69 / Chapter 4.4 --- Experimental Results and Discussions --- p.71 / Chapter 4.4.1 --- Experimental System and Results of the First Method --- p.71 / Chapter 4.4.2 --- Experimental System and Results of the Second Method --- p.76 / Chapter 5 --- Shape Recovery from One Sequence of Color Images --- p.81 / Chapter 5.1 --- Introduction --- p.81 / Chapter 5.2 --- Temporal-color Space Analysis of Reflection --- p.84 / Chapter 5.3 --- Estimation of Illuminant Color Ks --- p.88 / Chapter 5.4 --- Estimation of the Color Vector of the Body-reflection Component Kl --- p.89 / Chapter 5.5 --- Separating Specular and Body Reflection Components and Re- covering Surface Shape and Reflectance --- p.91 / Chapter 5.6 --- Experiment Results and Discussions --- p.92 / Chapter 5.6.1 --- Results with Interreflection --- p.93 / Chapter 5.6.2 --- Results Without Interreflection --- p.93 / Chapter 5.6.3 --- Simulation Results --- p.95 / Chapter 5.7 --- Analysis of Various Factors on the Accuracy --- p.96 / Chapter 5.7.1 --- Effects of Number of Samples --- p.96 / Chapter 5.7.2 --- Effects of Noise --- p.99 / Chapter 5.7.3 --- Effects of Object Size --- p.99 / Chapter 5.7.4 --- Camera Optical Axis Not in Light Source Plane --- p.102 / Chapter 5.7.5 --- Camera Optical Axis Not Passing Through Object Center --- p.105 / Chapter 6 --- Shape Recovery from Two Sequences of Images --- p.107 / Chapter 6.1 --- Introduction --- p.107 / Chapter 6.2 --- Method for 3D Shape Recovery from Two Sequences of Images --- p.109 / Chapter 6.3 --- Genetics-Based Method --- p.111 / Chapter 6.4 --- Experimental Results and Discussions --- p.115 / Chapter 6.4.1 --- Simulation Results --- p.115 / Chapter 6.4.2 --- Real Experimental Results --- p.118 / Chapter 7 --- Shape from Shading for Non-Lambertian Surfaces --- p.120 / Chapter 7.1 --- Introduction --- p.120 / Chapter 7.2 --- Reflectance Map for Non-Lambertian Color Surfaces --- p.123 / Chapter 7.3 --- Recovering Non-Lambertian Surface Shape from One Color Image --- p.127 / Chapter 7.3.1 --- Segmenting Hybrid Areas from Diffuse Areas Using Hue Information --- p.127 / Chapter 7.3.2 --- Calculating Intensities of Specular and Diffuse Compo- nents on Hybrid Areas --- p.128 / Chapter 7.3.3 --- Recovering Shape from Shading --- p.129 / Chapter 7.4 --- Experimental Results and Discussions --- p.131 / Chapter 7.4.1 --- Simulation Results --- p.131 / Chapter 7.4.2 --- Real Experimental Results --- p.136 / Chapter 8 --- Shape from Shading under Multiple Extended Light Sources --- p.142 / Chapter 8.1 --- Introduction --- p.142 / Chapter 8.2 --- Reflectance Map for Lambertian Surface Under Multiple Rectan- gular Light Sources --- p.144 / Chapter 8.3 --- Recovering Surface Shape Under Multiple Rectangular Light Sources --- p.148 / Chapter 8.4 --- Experimental Results and Discussions --- p.150 / Chapter 8.4.1 --- Synthetic Image Results --- p.150 / Chapter 8.4.2 --- Real Image Results --- p.152 / Chapter 9 --- Shape from Shading in Unknown Environments by Neural Net- works --- p.167 / Chapter 9.1 --- Introduction --- p.167 / Chapter 9.2 --- Shape Estimation --- p.169 / Chapter 9.2.1 --- Shape Recovery Problem under Multiple Rectangular Ex- tended Light Sources --- p.169 / Chapter 9.2.2 --- Forward Network Representation of Surface Normals --- p.170 / Chapter 9.2.3 --- Shape Estimation --- p.174 / Chapter 9.3 --- Application of the Neural Network in Shape Recovery --- p.174 / Chapter 9.3.1 --- Structure of the Neural Network --- p.174 / Chapter 9.3.2 --- Normalization of the Input and Output Patterns --- p.175 / Chapter 9.4 --- Experimental Results and Discussions --- p.178 / Chapter 9.4.1 --- Results for Lambertian Surface under One Rectangular Light --- p.178 / Chapter 9.4.2 --- Results for Lambertian Surface under Four Rectangular Light Sources --- p.180 / Chapter 9.4.3 --- Results for Hybrid Surface under One Rectangular Light Sources --- p.190 / Chapter 9.4.4 --- Discussions --- p.190 / Chapter 10 --- Summary and Conclusions --- p.191 / Chapter 10.1 --- Summary Results and Contributions --- p.192 / Chapter 10.2 --- Directions of Future Research --- p.199 / Bibliography --- p.202

A fundamental problem in machine vision is the classifcation of objects which may have unknown position, orientation, or a combination of these and other transformations. The massive amount of data required to accurately form an appearance-based model of an object under all values of shift and rotation transformations has discouraged the incorporation of the combination of both transformations into a single model representation. This Master's Thesis documents the theory and implementation of a hierarchical classifier, named the Information Theoretic Decision Tree system, which has the demonstrated ability to form appearance-based models of objects which are shift and rotation invariant which can be searched with a great reduction in evaluations over a linear sequential search. Information theory is utilized to obtain a measure of information gain in a feature space recursive segmentation algorithm which positions hyperplanes to local information gain maxima. This is accomplished dynamically through a process of local optimization based on a conjugate gradient technique enveloped by a simulated annealing optimization loop. Several target model training strategies have been developed for shift and rotation invariance, notably the method of exemplar grouping, in which any combination of rotation and translation transformations of target object views can be simulated and folded into the appearance-based model. The decision tree structure target models produced as a result of this process effciently represent the voluminous training data, according rapid test-time classification of objects.

In recent years there has been great interest in implementing object recognition frame work on mobile phones. This has stemmed from the fact the advances in object recognition algorithm and mobile phone capabilities have built a congenial ecosystem. Application developers on mobile platforms are trying to utilize the object recognition technology to build better human computer interfaces. This approach is in the nascent phase and proper application framework is required. In this thesis, we propose a framework to overcome design challenges and provide an evaluation methodology to assess the system performance. We use the emerging Android mobile platform to implement and test the framework. We performed a case study using the proposal and reported the test result. This assessment will help developers make wise decisions about their application design. Furthermore, the Android API developers could use this information to provide better interfaces to the third party developers. The design and evaluation methodology could be extended to other mobile platforms for a wider consumer base. / by Vivek Kumar Tyagi. / Thesis (M.S.C.S.)--Florida Atlantic University, 2010. / Includes bibliography. / Electronic reproduction. Boca Raton, Fla., 2010. Mode of access: World Wide Web.

This work explores the process of model-based classification of speech audio signals using low-level feature vectors. The process of extracting low-level features from audio signals is described along with a discussion of established techniques for training and testing mixture model-based classifiers and using these models in conjunction with feature selection algorithms to select optimal feature subsets. The results of a number of classification experiments using a publicly available speech database, the Berlin Database of Emotional Speech, are presented. This includes experiments in optimizing feature extraction parameters and comparing different feature selection results from over 700 candidate feature vectors for the tasks of classifying speaker gender, identity, and emotion. In the experiments, final classification accuracies of 99.5%, 98.0% and 79% were achieved for the gender, identity and emotion tasks respectively. / by Chris Thoman. / Thesis (M.S.C.S.)--Florida Atlantic University, 2009. / Includes bibliography. / Electronic reproduction. Boca Raton, Fla., 2009. Mode of access: World Wide Web.

by Lin Feng. / "June 2003." / Thesis (Ph.D.)--Chinese University of Hong Kong, 2003. / Includes bibliographical references (p. [114]-130). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. Ann Arbor, MI : ProQuest Information and Learning Company, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Mode of access: World Wide Web. / Abstracts in English and Chinese.

Basic technologies for fingerprint feature extraction and matching have been improved to such a stage that they can be embedded into commercial Automatic Fingerprint Verification Systems (AFVSs). However, the reliability of AFVSs has kept attracting concerns from the society since AFVSs do fail occasionally due to difficulties like problematic fingers, changing environments, and malicious attacks. Furthermore, the absence of a solid theoretical foundation for evaluating AFVSs prevents these failures from been predicted and evaluated. Under the traditional empirical AFVS evaluation framework, repeated verification experiments, which can be very time consuming, have to be performed to test whether an update to an AFVS can really lead to an upgrade in its performance. Also, empirical verification results are often unable to provide deeper understanding of AFVSs. To solve these problems, we propose a novel statistical evaluation model for minutiae-based AFVSs based on the understanding of fingerprint minutiae patterns. This model can predict the verification performance metrics as well as their confidence intervals. The analytical power of our evaluation model, which makes it superior to empirical evaluation methods, can assist system developers to upgrade their AFVSs purposefully. Also, our model can facilitate the theoretical analysis of the advantages and disadvantages of various fingerprint verification techniques. We verify our claims through different and extensive experiments. / Chen, Jiansheng. / "November 2006." / Adviser: Yiu-Sang Moon. / Source: Dissertation Abstracts International, Volume: 68-08, Section: B, page: 5343. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2006. / Includes bibliographical references (p. 110-122). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. [Ann Arbor, MI] : ProQuest Information and Learning, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstracts in English and Chinese. / School code: 1307.

by Jian-Zhuang Liu. / Thesis (Ph.D.)--Chinese University of Hong Kong, 1997. / Includes bibliographical references (p. 183-196). / Microfiche. Ann Arbor, Mich.: UMI, 1998. 3 microfiches ; 11 x 15 cm.

by Law Tak Ming. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1996. / Includes bibliographical references (leaves 91-96). / by Law Tak Ming. / Chapter 1. --- INTRODUCTION --- p.1 / Chapter 1.1 --- The Structure of Chinese Characters --- p.3 / Chapter 1.1.1 --- Pixels (像素) --- p.4 / Chapter 1.1.2 --- Strokes (筆劃) --- p.4 / Chapter 1.1.3 --- Basic Stroke Types (Segment Type)基本筆劃（筆段) --- p.4 / Chapter 1.1.4 --- Compound-segment Stroke (複合筆劃) --- p.5 / Chapter 1.1.5 --- Total Stroke types --- p.6 / Chapter 1.1.6 --- Stroke Sequence (筆順) --- p.6 / Chapter 1.1.7 --- Segments as Basic Features --- p.7 / Chapter 1.1.8 --- Geographic Structure of Components --- p.7 / Chapter 1.2 --- Stroke Distribution of Chinese Characters --- p.10 / Chapter 1.3 --- Radical --- p.10 / Chapter 1.4 --- The Comparison between ON-line and Off-line Chinese Character Recognition Approach --- p.11 / Chapter 1.5 --- Commercial Product Comparison --- p.14 / Chapter 1.6 --- Related Works --- p.17 / Chapter 1.7 --- Objectives --- p.29 / Chapter 2. --- PREPROCESSING --- p.31 / Chapter 2.1 --- Smoothing and Sampling --- p.32 / Chapter 2.2 --- Interpolation --- p.34 / Chapter 2.3 --- DEHOOKING --- p.37 / Chapter 2.4 --- Stroke Segmentation --- p.39 / Chapter 3. --- DATA LEARNING --- p.41 / Chapter 3.1 --- Definition of Terms --- p.41 / Chapter 3.2 --- Definition of Direction type --- p.42 / Chapter 3.3 --- Data Base Structure --- p.43 / Chapter 3.4 --- Learning Algorithms of Segments --- p.45 / Chapter 3.4.1 --- Learning of the Coordinates --- p.48 / Chapter 3.4.2 --- Learning of Direction Type --- p.48 / Chapter 3.4.3 --- Learning of Slope Angle --- p.50 / Chapter 3.5 --- Learning of the Tolerance of Coordinate --- p.50 / Chapter 3.6 --- Stroke Relation Coding --- p.51 / Chapter 4. --- PRECLASSIFICATION --- p.54 / Chapter 4.1 --- Decision Path Classification --- p.56 / Chapter 4.2 --- First-Two-Ending-One Classification Method --- p.57 / Chapter 4.3 --- Stroke Type Matching Algorithm --- p.61 / Chapter 5. --- RECOGNITION STAGE --- p.64 / Chapter 5.1 --- Connected Strokes Handling --- p.65 / Chapter 5.2 --- Stroke Sequence Free Matching Algorithm --- p.70 / Chapter 5.3 --- Preliminary Character Distance Measure --- p.72 / Chapter 5.4 --- Detailed Matching Techniques --- p.74 / Chapter 5.5 --- Segments Sequence Within a Compound-segment Stroke Compatibility --- p.75 / Chapter 5.5.1 --- Length and Slope Orientation Similarities --- p.78 / Chapter 5.5.2 --- Segment Similarity Measure Function --- p.79 / Chapter 5.6 --- Stroke Relation Influences --- p.79 / Chapter 5.7 --- Final Character Similarity Measure --- p.81 / Chapter 6. --- RESULTS AND CONCLUSIONS --- p.83 / Chapter 6.1 --- Experiment Results --- p.83 / Chapter 6.2 --- Analysis --- p.85 / Chapter 6.3 --- Conclusion --- p.87