- Recognition The recognition as a separate criminalistic method serves to identification of persons, things or animals linked to a criminally relevant event. The recognition is also procedural act regulated in a separate section of Criminal Procedure Code called Some specific methods of evidence. This classification, which has come into effect in 2002, shows that the importance and significance of the recognition can't be underestimated and in addition to the theoreticians it comes to the forefront of the legislature's attention. The thesis is divided into six chapters and each of them deals with the recognition from another point of view. In total, they cover the most important aspects of the recognition. The first chapter is devoted to a theoreticall introduction to the topic of recognition, its definiton, description of its essence as well as to an individual types of recognition, defining the basic differences between all mentioned types. This chapter also includes the theoreticall inclusion of this institute into a criminalistic science. The second chapter includes the constitution of the recognition from the oldest sources to the current legislation in the Criminal Procedure Code - mainly the section 104b and its eight paragraphs which constitute the most fundamental rules which must be...

Chiang Kong Wa. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2006. / Includes bibliographical references (leaves 74-79). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 2 --- Technology Overview --- p.4 / Chapter 2.1 --- Components of RFID Systems --- p.5 / Chapter 2.1.1 --- Tag --- p.6 / Chapter 2.1.2 --- Reader --- p.9 / Chapter 2.1.3 --- Software systems --- p.10 / Chapter 2.1.4 --- Communication infrastructure --- p.11 / Chapter 2.2 --- Frequency Regulations and Standards --- p.11 / Chapter 2.2.1 --- RFID frequency bands --- p.11 / Chapter 2.2.2 --- Standards --- p.12 / Chapter 2.3 --- Advantages and Limitations of RFID Technology --- p.14 / Chapter 2.4 --- Applications --- p.17 / Chapter 3 --- Background of Research --- p.20 / Chapter 3.1 --- Anti-collision methods for RFID systems --- p.22 / Chapter 3.1.1 --- Stochastic Anti-collision Protocols --- p.25 / Chapter 3.1.2 --- Deterministic Anti-collision Protocols --- p.27 / Chapter 4 --- Even-Odd Binary Tree Protocol --- p.30 / Chapter 4.1 --- Protocol Description --- p.31 / Chapter 4.2 --- Time Complexity Analysis --- p.34 / Chapter 4.3 --- Performance Evaluation --- p.37 / Chapter 4.4 --- Summary --- p.41 / Chapter 5 --- Prefix-Randomized Query-Tree Protocol --- p.44 / Chapter 5.1 --- Tag Identification - Known Tag Set Size --- p.45 / Chapter 5.1.1 --- Protocol Description --- p.45 / Chapter 5.1.2 --- Time Complexity Analysis --- p.47 / Chapter 5.1.3 --- Optimal Initial Prefix Length --- p.50 / Chapter 5.1.4 --- Optimal Number of Level-1 Nodes --- p.52 / Chapter 5.2 --- Tag Identification - Unknown Tag Set Size --- p.53 / Chapter 5.2.1 --- Initial Prefix Length Adaptation Algorithm --- p.54 / Chapter 5.2.2 --- Computing r*Δ(l) --- p.55 / Chapter 5.2.3 --- Optimal Choice of Step Size Δ --- p.56 / Chapter 5.3 --- Performance Evaluation --- p.59 / Chapter 5.4 --- Summary --- p.64 / Chapter 6 --- Conclusion and Future Work --- p.68 / Chapter 6.1 --- Conclusion --- p.68 / Chapter 6.2 --- Future Work --- p.70 / Bibliography --- p.74

Wei Dacheng. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2006. / Includes bibliographical references. / Abstracts in English and Chinese. / Table of Contents --- p.VIII / List of Tables --- p.XI / List of Figures --- p.XII / List of Abbreviations --- p.XV / Chapter Chapter 1 --- Introduction / Chapter 1.1 --- Introduction to RFID system --- p.1 / Chapter 1.2 --- Why we choose Active RFID system --- p.4 / Chapter 1.3 --- Objective of the research --- p.6 / Chapter 1.3.1 --- Requirement analysis --- p.7 / Chapter 1.3.2 --- Selection of RFID system and standard --- p.8 / Chapter 1.4 --- Original contribution of this dissertation --- p.9 / Chapter 1.5 --- Organization of the dissertation --- p.9 / Reference --- p.10 / Chapter Chapter 2 --- Implementation of An Active RFID System / Chapter 2.1 --- RFID System hardware design and related protocol --- p.1 / Chapter 2.2 --- Introduction to ISO 18000-7 --- p.7 / Chapter 2.3 --- Microcontroller specification --- p.12 / Chapter 2.4 --- RF model specifications --- p.14 / Chapter 2.5 --- Communication between a PC and a Reader --- p.15 / Chapter 2.6 --- Programming --- p.16 / Chapter 2.6.1 --- Procedure sequences of Reader and Tag --- p.17 / Chapter 2.6.2 --- Sequence of data transmission and reception --- p.24 / Chapter 2.6.3 --- CRC implementation --- p.28 / Chapter 2.7 --- Testing result --- p.31 / Reference --- p.35 / Chapter Chapter 3 --- Novel Power Saving Methods for an Active RFID System / Chapter 3.1 --- Some drawbacks of the existing Active RFID protocol --- p.1 / Chapter 3.1.1 --- Power consumption problem --- p.1 / Chapter 3.1.2 --- Multi-Reader problem --- p.9 / Chapter 3.2 --- Solutions of the Multi-Reader problem and power saving problem --- p.10 / Chapter 3.2.1 --- A solution to the power saving problem --- p.11 / Chapter 3.2.2 --- A solution to the Multi-Reader problem --- p.16 / Reference --- p.21 / Chapter Chapter 4 --- A Probe-fed Compact Half-wave Length Dipole Antenna for Active RFID System / Chapter 4.1 --- Requirement of an antenna for Active RFID system --- p.1 / Chapter 4.2 --- A probe-fed half-wave length dipole EE shape antenna for metallic object application --- p.2 / Chapter 4.3 --- Electromagnetic simulation results --- p.5 / Chapter 4.4 --- Operating principle analysis --- p.9 / Chapter 4.5 --- Using V shape structure to increase the bandwidth --- p.19 / Chapter 4.6 --- Prototyping and measurement results --- p.22 / Chapter 4.7 --- Conclusion --- p.28 / Reference --- p.29 / Chapter Chapter 5 --- Conclusion

說話人識別(Speaker Recognition) 主要利用聲音來檢測說話人的身份，是一項重要且極具挑戰性的生物認證研究課題。通常來說，針對語音信號的文本內容差別，說話人識別可以分成文本相關和文本無關兩類。另外，說話人識別有兩類重要應用，第一類是說話人確認，主要是通過給定話者聲音信息對說話人聲稱之身份進行二元判定。另一類是說話人辨識，其主要是從待選說話人集中判斷未知身份信息的話者身份。 / 在先進的說話人識別系統中，每個說話人模型是通過給定的說話人數據進行特徵統計分佈估計由生成模型訓練得到。這類方法由於需要逐帧進行概率或似然度計算而得出最終判決，會耗費大量系統資源並降低實時性性能。採用子空間降維技術，我們不僅避免選取冗餘高維度數據，同時能夠有效删除於識別中無用之數據。為克服上述生成性模型的不足並獲得不同說話人間的區分邊界，本文提出了利用區分性子空間方法訓練模型並採用有效的距離測度作為最終的建模識別新算法。 / 在本篇論文中，我們將先介紹並分析各類產生性說話人識別方法，例如高斯混合模型及聯合因子分析。另外，為了降低特徵空間維度和運算時間，我們也對子空間分析技術做了調研。除此之外，我們提出了一種取名為Fishervoice 基於非參數分佈假定的新穎說話人識別框架。所提出的Fishervoice 框架的主要目的是為了降低噪聲干擾同時加重分類信息，而能夠加強在可區分性的子空間內對聲音特徵建模。採用上述Fishervoice 框架，說話人識別可以簡單地通過測試樣本映射到Fishervoice 子空間並計算其簡單歐氏距離而實現。為了更好得降低維度及提高識別率，我們還對Fishervocie 框架進行多樣化探索。另外，我們也在低維度的全變化空間(Total Variability) 對各類多種子空間分析模型進行調比較。基於XM2VTS 和NIST 公開數據庫的實驗驗證了本文提出的算法的有效性。 / Speaker Recognition (SR), which uses the voice to determine the speaker’s identity, is an important and challenging research topic for biometric authentication. Generally speaking, speaker recognition can be divided into text-dependent and text-independent methods according to the verbal content of the speech signal. There are two major applications of speaker recognition: the first is speaker verification, also referred to speaker authentication, which is used to validate the identity of a speaker according to the voice and it involves a binary decision. The second is speaker identification, which is used to determine an unknown speaker’s identity. / In a state-of-art speaker recognition system, the speaker training model is usually trained by generative methods, which estimate feature distribution of each speaker among the given data. These generative methods need a frame-based metric (e.g. probability, likelihoods) calculation for making final decision, which consumes much computer resources, slowing down the real-time responses. Meanwhile, lots of redundant data frames are blindly selected for training without efficient subspace dimension reduction. In order to overcome disadvantages of generative methods and obtain boundary information between individual speakers, we propose to apply the discriminative subspace technique for model training and employ simple but efficient distance metrics for decision score calculation. / In this thesis, we shall present an overview of both conventional and state-of-the-art generative speaker recognition methods (e.g. Gaussian Mixture Model and Joint Factor Analysis) and analyze their advantages and disadvantages. In addition, we have also made an investigation of the application of subspace analysis techniques to reduce feature dimensions and computation time. After that, a novel speaker recognition framework based on the nonparametric Fisher’s discriminant analysis which we name Fishervoice is proposed. The objective of the proposed Fishervoice algorithm is to model the intrinsic vocal characteristics in a discriminant subspace for de-emphasizing unwanted noise variations and emphasizing classification boundaries information. Using the proposed Fishervoice framework, speaker recognition can be easily realized by mapping a test utterance to the Fishervoice subspace and then calculating the score between the test utterance and its reference. Besides, we explore the proposed Fishervoice framework with several extensions for further dimensionality reduction and performance improvement. Furthermore, we investigate various subspace analysis techniques in a total variability-based low-dimensional space for fast computation. Extensive experiments on two large speaker recognition corpora (XM2VTS and NIST) demonstrate significant improvements of Fishervoice over standard, state-of-the-art approaches for both speaker identification and verification systems. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Jiang, Weiwu. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2012. / Includes bibliographical references (leaves 127-135). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstract also in Chinese. / Abstract --- p.i / Acknowledgements --- p.vi / Contents --- p.xiv / List of Figures --- p.xvii / List of Tables --- p.xxiii / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Overview of Speaker Recognition Systems --- p.1 / Chapter 1.2 --- Motivation --- p.4 / Chapter 1.3 --- Outline of Thesis --- p.6 / Chapter 2 --- Background Study --- p.7 / Chapter 2.1 --- Generative Gaussian Mixture Model (GMM) --- p.7 / Chapter 2.1.1 --- Basic GMM --- p.7 / Chapter 2.1.2 --- The Gaussian Mixture Model-Universal Background Model (GMM-UBM) System --- p.9 / Chapter 2.2 --- Discriminative Subspace Analysis --- p.12 / Chapter 2.2.1 --- Principal Component Analysis --- p.12 / Chapter 2.2.2 --- Linear Discriminant Analysis --- p.16 / Chapter 2.2.3 --- Heteroscedastic Linear Discriminant Analysis --- p.17 / Chapter 2.2.4 --- Locality Preserving Projections --- p.18 / Chapter 2.3 --- Noise Compensation --- p.20 / Chapter 2.3.1 --- Eigenvoice --- p.20 / Chapter 2.3.2 --- Joint Factor Analysis --- p.24 / Chapter 2.3.3 --- Probabilistic Linear Discriminant Analysis --- p.26 / Chapter 2.3.4 --- Nuisance Attribute Projection --- p.30 / Chapter 2.3.5 --- Within-class Covariance Normalization --- p.32 / Chapter 2.4 --- Support Vector Machine --- p.33 / Chapter 2.5 --- Score Normalization --- p.35 / Chapter 2.6 --- Summary --- p.39 / Chapter 3 --- Corpora for Speaker Recognition Experiments --- p.41 / Chapter 3.1 --- Corpora for Speaker Identification Experiments --- p.41 / Chapter 3.1.1 --- XM2VTS Corpus --- p.41 / Chapter 3.1.2 --- NIST Corpora --- p.42 / Chapter 3.2 --- Corpora for Speaker Verification Experiments --- p.45 / Chapter 3.3 --- Summary --- p.47 / Chapter 4 --- Performance Measures for Speaker Recognition --- p.48 / Chapter 4.1 --- Performance Measures for Identification --- p.48 / Chapter 4.2 --- Performance Measures for Verification --- p.49 / Chapter 4.2.1 --- Equal Error Rate --- p.49 / Chapter 4.2.2 --- Detection Error Tradeoff Curves --- p.49 / Chapter 4.2.3 --- Detection Cost Function --- p.50 / Chapter 4.3 --- Summary --- p.51 / Chapter 5 --- The Discriminant Fishervoice Framework --- p.52 / Chapter 5.1 --- The Proposed Fishervoice Framework --- p.53 / Chapter 5.1.1 --- Feature Representation --- p.53 / Chapter 5.1.2 --- Nonparametric Fisher’s Discriminant Analysis --- p.55 / Chapter 5.2 --- Speaker Identification Experiments --- p.60 / Chapter 5.2.1 --- Experiments on the XM2VTS Corpus --- p.60 / Chapter 5.2.2 --- Experiments on the NIST Corpus --- p.62 / Chapter 5.3 --- Summary --- p.64 / Chapter 6 --- Extension of the Fishervoice Framework --- p.66 / Chapter 6.1 --- Two-level Fishervoice Framework --- p.66 / Chapter 6.1.1 --- Proposed Algorithm --- p.66 / Chapter 6.2 --- Performance Evaluation on the Two-level Fishervoice Framework --- p.70 / Chapter 6.2.1 --- Experimental Setup --- p.70 / Chapter 6.2.2 --- Performance Comparison of Different Types of Input Supervectors --- p.72 / Chapter 6.2.3 --- Performance Comparison of Different Numbers of Slices --- p.73 / Chapter 6.2.4 --- Performance Comparison of Different Dimensions of Fishervoice Projection Matrices --- p.75 / Chapter 6.2.5 --- Performance Comparison with Other Systems --- p.77 / Chapter 6.2.6 --- Fusion with Other Systems --- p.78 / Chapter 6.2.7 --- Extension of the Two-level Subspace Analysis Framework --- p.80 / Chapter 6.3 --- Random Subspace Sampling Framework --- p.81 / Chapter 6.3.1 --- Supervector Extraction --- p.82 / Chapter 6.3.2 --- Training Stage --- p.83 / Chapter 6.3.3 --- Testing Procedures --- p.84 / Chapter 6.3.4 --- Discussion --- p.84 / Chapter 6.4 --- Performance Evaluation of the Random Subspace Sampling Framework --- p.85 / Chapter 6.4.1 --- Experimental Setup --- p.85 / Chapter 6.4.2 --- Random Subspace Sampling Analysis --- p.87 / Chapter 6.4.3 --- Comparison with Other Systems --- p.90 / Chapter 6.4.4 --- Fusion with the Other Systems --- p.90 / Chapter 6.5 --- Summary --- p.92 / Chapter 7 --- Discriminative Modeling in Low-dimensional Space --- p.94 / Chapter 7.1 --- Discriminative Subspace Analysis in Low-dimensional Space --- p.95 / Chapter 7.1.1 --- Experimental Setup --- p.96 / Chapter 7.1.2 --- Performance Evaluation on Individual Subspace Analysis Techniques --- p.98 / Chapter 7.1.3 --- Performance Evaluation on Multi-type of Subspace Analysis Techniques --- p.105 / Chapter 7.2 --- Discriminative Subspace Analysis with Support Vector Machine --- p.115 / Chapter 7.2.1 --- Experimental Setup --- p.116 / Chapter 7.2.2 --- Performance Evaluation on LDA+WCCN+SVM --- p.117 / Chapter 7.2.3 --- Performance Evaluation on Fishervoice+SVM --- p.118 / Chapter 7.3 --- Summary --- p.118 / Chapter 8 --- Conclusions and Future Work --- p.120 / Chapter 8.1 --- Contributions --- p.120 / Chapter 8.2 --- Future Directions --- p.121 / Chapter A --- EM Training GMM --- p.123 / Bibliography --- p.127

Fingerprints in blood indicate a threshold of violence has been surpassed in crime scenarios - making the crime resolution more urgent. There exist multiple processes that enhance a blood fingerprint in its original position, or in-situ, with reliability so that an image can be obtained. However, blood fingerprint evidence that cannot directly be transported to a laboratory for further analysis, due to the size or mobility of the substrate, calls for portability. In 2010 Zar-Pro Fluorescent Blood Lifting Strips were patented by Jessica Zarate as a "fluorogenic method for lifting, enhancing, and preserving blood impression evidence". The lifted prints are also inherently fluorescent to further increase enhancement and contrast of the print. There are currently no studies comparing Zar-Pro results with the results of other laboratory enhancement methods. This experiment compared Zar-Pro to other non-portable and frequently used alternatives - blood peak absorption and Hungarian Red enhancement to determine if Zar-Pro gives better blood fingerprint enhancement results than other non-portable alternatives - ALS visualization and Hungarian Red enhancement. In this study, Zar-Pro methods produced more reliable and reproducible results over the Hungarian Red and blood peak absorption methods on white and black ceramic tile. From this study, one can also conclude that ALS peak absorption is better suited for the location of blood prints on a light-colored item of evidence, rather than an enhancement method of blood prints.

The objective of this dissertation is to offer a series of analyses that contribute to the validation of the Aurora Battery, a cognitive assessment based on Robert J. Sternberg's theory of Successful Intelligence that is currently under development. Convergent validity will be examined by exploring how objective and subjective measures of the battery converge, through the novel application of the Correlated Trait Correlated Method-1, a specialized confirmatory factor analysis model that allows subjective measures to be compared against an objective measure. The predictive validity of Aurora will be shown by highlighting Aurora's ability to help predict students' school grade point average through latent growth curve models that are extended into path models. Divergent validity will be demonstrated by establishing sensitivity and specificity between the Aurora Battery and the TerraNova tests. Finally, the current state of the field of giftedness and possible future directions will be discussed.

O problema de identificação de parâmetros consiste em se determinar parâmetros que minimizem a diferença entre valores medidos e calculados de determinadas grandezas. Certamente, essa identificação é realizada para parâmetros que apresentam razoável grau de incerteza nos seus valores. Neste trabalho apresentam-se os principais conceitos e fundamentos matemáticos envolvidos no assunto, desenvolve-se um procedimento de identificação de parâmetros com base matemática sólida e aplica-se esse procedimento em problemas de interesse prático da engenharia. São estudados o Túnel de Hudvudsta e a barragem de Machadinho, nos quais são identificados parâmetros relacionados com as ações ou com as propriedades físicas dos materiais, considerando modelos hierárquicos para representar as estruturas. Utilizam-se os principais critérios de identificação para a definição das funções objetivo e métodos do tipo Newton para a minimização dessas funções. / The parameter identification problem consists of determining the values of the parameters that minimize the difference between measured and calculated values of some variables. Indeed, this identification is performed to parameters that present some uncertainty on their values. In this work the main mathematical concepts and fundaments related to back analysis are presented. A procedure for parameter identification with a consistent mathematical basis is developed and applied in practical engineering problems. The Hudvudsta tunnel and the Machadinho dam are studied to identify parameters related to loads or material physical properties by using hierarchical models to represent the structures. The objetive functions are defined with the main identification criteria and minimized with Newton´s methods.

Due to the recent advances in semiconductor technology, the Radio Frequency Identification (RFID) technology is approaching the critical point for inventory application in global supply chain logistics. Its unique advantage to identify multiple tags simultaneously can remove large amount of labor-intensive scanning for inventory control and checkout process. To realize multiple-tag-identification, the tag collision problem, which is due to the signal collision of simultaneous transmission by multiple tags, needs to be solved. / Over the years, many different RFID systems and anti-collision algorithms were proposed, but a uniform theory which fully analyzes the anti-collision system is still lacking. Most of the previous work treated the RFID system as a special kind of Random Access (RA) system and designed anti-collision algorithms to maximize the temporary throughput. In this thesis, we provide a theoretical framework for the RFID anti-collision system. We differentiate the RFID system with the RA system and propose a general model for all types of the RFID systems. We also provide a general method for algorithm optimization in different RFID systems. As special examples, we analyze some popular RFID systems and derive the optimal algorithms under the system constraints. These optimal results are verified by computer simulation. / Zhu, lei. / Adviser: Tak-shing Peter Yum. / Source: Dissertation Abstracts International, Volume: 73-03, Section: B, page: . / Thesis (Ph.D.)--Chinese University of Hong Kong, 2010. / Includes bibliographical references (leaves 152-156). / 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, [201-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstract also in Chinese.

Finally, the research of classifying and identifying individuals through their walking patterns is introduced. Alive biometrical features in dynamic human gait are adopted in the intelligent shoe system. Since gait data are dynamic, non-linear, stochastic, time-varying, noisy and multi-channel, we must select a modeling framework capable of dealing with these expected complexities in the data. Using the proposed machine learning methods, support vector machine (SVM) and hidden Markov models (HMMs), we build up probabilistic models that take the information of human walking patterns into account, and compare the overall similarity among human walking patterns of several wearers. / In this thesis, we will build intelligent shoes under the framework for capturing and analyzing dynamic human gait. Existing MEMS technology makes it possible to integrate all the sensors and circuits inside a small module. In designing our intelligent shoe system, we require the following key characteristics in our system: (1) It should be convenient to wear and socially acceptable. Thus, the sensors and electronic hardware installed should not substantially change the weight and weight balance of a typical shoe, lest it alters how an individual normally walks. (2) We want to analyze a user's motion in real-time through a wireless interface to a remote laptop or other computer; we will also incorporate on-shoe data logging hardware for off-line analysis. (3) Sensors that monitor gait motion conditions may need to be attached to the insoles, in closer proximity to the foot of users. In order to investigate the problem of capturing power parasitically from normal human-body-motion for use in personal electronics applications, we also plan to develop an electromechanical generator embedded within the shoe for parasitic power collection from heel strike. / Next, we can encode specific motions to control external devices through a wireless interface. This same system architecture that allows us to classify broad categories of motion also allows the intelligent shoe to act as a programmable, low-data rate control interface. We apply the system to several successful tasks based on this platform, especially the Shoe-Mouse. By using this interface, we can operate a device with our feet. / Then, we present potential use of machine learning techniques, in particular support vector machine (SVM), and the intelligent shoe platform to detect discrete stages in the cyclic motion of dynamic human gait, and construct an identifier of five discrete events that occur in a cyclic process for precise control of functional electrical stimulation (FES). With the information of when the legs are in each phase of a gait, the timing of specific gait phase can be assessed. / Huang, Bufu. / "September 2007." / Adviser: Yangsheng Xu. / Source: Dissertation Abstracts International, Volume: 69-08, Section: B, page: 4931. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2007. / Includes bibliographical references (p. 122-131). / 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.

Among the these proposed methods (i.e., random subspace, bagging, and pairwise classification), the pairwise classification method produces the highest accuracy at the expense of more computation time for training classifiers. The random subspace method and bagging approach have similar performance. To recognize a testing plankton pattern, the computational costs of the these methods are alike. / Due to the complexity of plankton recognition problem, it is difficult to pursue a single optimal classifier to meet all the requirements. In this work, instead of developing a single sophisticated classifier, we propose an ensemble learning framework based on the random sampling techniques including random subspace and bagging. In the random subspace method, a set of low-dimensional subspaces are generated by randomly sampling on the feature space, and multiple classifiers constructed from these random subspaces are combined to yield a powerful classifier. In the bagging approach, a number of independent bootstrap replicates are generated by randomly sampling with replacement on the training set. A classifier is trained on each replicate, and the final result is produced by integrating all the classifiers using majority voting. Using random sampling, the constructed classifiers are stable and multiple classifiers cover the entire feature space or the whole training set without losing discriminative information. Thus, good performance can be achieved. Experimental results demonstrate the effectiveness of the random sampling techniques for improving the system performance. / On the other hand, in previous approaches, normally the samples of all the plankton classes are used for a single classifier training. It may be difficult to select one feature space to optimally represent and classify all the patterns. Therefore, the overall accuracy rate may be low. In this work, we propose a pairwise classification framework, in which the complex multi-class plankton recognition problem is transformed into a set of two-class problems. Such a problem decomposition leads to a number of simpler classification problems to be solved, and it provides an approach for independent feature selection for each pair of classes. This is the first time for such a framework introduced in plankton recognition. We achieve nearly perfect classification accuracy on every pairwise classifier with less number of selected features, since it is easier to select an optimal feature vector to discriminate the two-class patterns. The ensemble of these pairwise classifiers will increase the overall performance. A high accuracy rate of 94.49% is obtained from a collection of more than 3000 plankton images, making it comparable with what a trained biologist can achieve by using conventional manual techniques. / Plankton including phytoplankton and zooplankton form the base of the food chain in the ocean and are a fundamental component of marine ecosystem dynamics. The rapid mapping of plankton abundance together with taxonomic and size composition can help the oceanographic researchers understand how climate change and human activities affect marine ecosystems. / Recently the University of South Florida developed the Shadowed Image Particle Profiling and Evaluation Recorder (SIPPER), an underwater video system which can continuously capture the magnified plankton images in the ocean. The SIPPER images differ from those used for most previous research in four aspects: (i) the images are much noisier, (ii) the objects are deformable and often partially occluded, (iii) the images are projection variant, i.e., the images are video records of three-dimensional objects in arbitrary positions and orientations, and (iv) the images are binary thus are lack of texture information. To deal with these difficulties, we implement three most valuable general features (i.e., moment invariants, Fourier descriptors, and granulometries) and propose a set of specific features such as circular projections, boundary smoothness, and object density to form a more complete description of the binary plankton patterns. These features are translation, scale, and rotation invariant. Moreover, they are less sensitive to noise. High-quality features will surely benefit the overall performance of the plankton recognition system. / Since all the features are extracted from the same plankton pattern, they may contain much redundant information and noise as well. Different types of features are incompatible in length and scale and the combined feature vector has a higher dimensionality. To make the best of these features for the binary SIPPER plankton image classification, we propose a two-stage PCA based scheme for feature selection, combination, and normalization. The first-stage PCA is used to compact every long feature vector by removing the redundant information and reduce noise as well, and the second-stage PCA is employed to compact the combined feature vector by eliminating the correlative information among different types of features. In addition, we normalize every component in the combined feature vector to the same scale according to its mean value and variance. In doing so, we reduce the computation time for the later recognition stage, and improve the classification accuracy. / Zhao Feng. / "May 2006." / Adviser: Xiaoou Tang. / Source: Dissertation Abstracts International, Volume: 67-11, Section: B, page: 6666. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2006. / Includes bibliographical references (p. 121-136). / 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.