Image-based face recognition has attained wide applications during the past decades in commerce and law enforcement areas, such as mug shot database matching, identity authentication, and access control. Existing face recognition techniques (e.g., Eigenface, Fisherface, and Elastic Bunch Graph Matching, etc.), however, do not perform well when the following case inevitably exists. The case is that, due to some variations in imaging conditions, e.g., pose and illumination changes, face images of the same person often have different appearances. These variations make face recognition techniques much challenging. With this concern in mind, the objective of my research is to develop robust face recognition techniques against variations. This thesis addresses two main variation problems in face recognition, i.e., pose and illumination variations. To improve the performance of face recognition systems, the following methods are proposed: (1) a face feature extraction and representation method using non-uniformly selected Gabor convolution features, (2) an illumination normalization method using adaptive region-based image enhancement for face recognition under variable illumination conditions, (3) an eye detection method in gray-scale face images under various illumination conditions, and (4) a virtual pose generation method for pose-invariant face recognition. The details of these proposed methods are explained in this thesis. In addition, we conduct a comprehensive survey of the existing face recognition methods. Future research directions are pointed out.

No description available.

No description available.

It has often been argued that the processing of faces is 'special' relative to the processing of other objects and there is much evidence in support of this notion. One source of evidence is the inversion effect, which occurs when faces presented upright are recognised significantly better than faces presented upside down. This effect of stimulus inversion has been shown to impair face recognition to a greater extent than for any other object class. It is this disproportionate effect that has been given as one source of evidence that face processing is special. However, other research has argued that effects of inversion can be found for non-face stimuli providing that there is sufficient development of expertise with them and that these stimuli can be defined by a common prototype. This thesis further explores this idea. Inversion effects were investigated for both prototypically and non-prototypically defined, abstract, chequerboard stimuli and compared with those for faces. When subjects learned to categorise chequerboard stimuli that were defined by a common prototype equal size inversion effects were found to those observed for faces. However, inversion effects were not observed for category training with multiple exemplars of chequerboard stimuli that were not defined by a common prototype. Together the findings are consistent with the idea that inversion effects are a general phenomenon resulting from the acquisition of category expertise with any prototype defined stimulus category. They undermine the inversion effect as a source of evidence for the specialness of face processing. Further, using a new Moving Windows technique, additional experiments investigated the underlying mechanisms responsible for the effects of inversion found for faces and chequerboards. These showed that the diagnostic image regions searched differ across the two stimulus classes. However, on the basis of the results, it is argued that the inversion effects found for both could result from impaired processing of second-order configural information.

Image-based face recognition has attained wide applications during the past decades in commerce and law enforcement areas, such as mug shot database matching, identity authentication, and access control. Existing face recognition techniques (e.g., Eigenface, Fisherface, and Elastic Bunch Graph Matching, etc.), however, do not perform well when the following case inevitably exists. The case is that, due to some variations in imaging conditions, e.g., pose and illumination changes, face images of the same person often have different appearances. These variations make face recognition techniques much challenging. With this concern in mind, the objective of my research is to develop robust face recognition techniques against variations. This thesis addresses two main variation problems in face recognition, i.e., pose and illumination variations. To improve the performance of face recognition systems, the following methods are proposed: (1) a face feature extraction and representation method using non-uniformly selected Gabor convolution features, (2) an illumination normalization method using adaptive region-based image enhancement for face recognition under variable illumination conditions, (3) an eye detection method in gray-scale face images under various illumination conditions, and (4) a virtual pose generation method for pose-invariant face recognition. The details of these proposed methods are explained in this thesis. In addition, we conduct a comprehensive survey of the existing face recognition methods. Future research directions are pointed out. / Applied Science, Faculty of / Electrical and Computer Engineering, Department of / Graduate

No description available.

No description available.

No description available.

Tang Ho-Man. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2003. / Includes bibliographical references (leaves 85-92). / Abstracts in English and Chinese. / Abstract --- p.i / Acknowledgement --- p.iv / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Background --- p.1 / Chapter 1.2 --- Face Recognition --- p.2 / Chapter 1.3 --- Contributions --- p.4 / Chapter 1.4 --- Organization of this Thesis --- p.6 / Chapter 2 --- Literature Review --- p.8 / Chapter 2.1 --- Committee Machine --- p.8 / Chapter 2.1.1 --- Static Structure --- p.9 / Chapter 2.1.2 --- Dynamic Structure --- p.10 / Chapter 2.2 --- Face Recognition Algorithms Overview --- p.11 / Chapter 2.2.1 --- Eigenface --- p.12 / Chapter 2.2.2 --- Fisherface --- p.17 / Chapter 2.2.3 --- Elastic Graph Matching --- p.19 / Chapter 2.2.4 --- Support Vector Machines --- p.23 / Chapter 2.2.5 --- Neural Networks --- p.25 / Chapter 2.3 --- Commercial System and Applications --- p.27 / Chapter 2.3.1 --- FaceIT --- p.28 / Chapter 2.3.2 --- ZN-Face --- p.28 / Chapter 2.3.3 --- TrueFace --- p.29 / Chapter 2.3.4 --- Viisage --- p.30 / Chapter 3 --- Static Structure --- p.31 / Chapter 3.1 --- Introduction --- p.31 / Chapter 3.2 --- Architecture --- p.32 / Chapter 3.3 --- Result and Confidence --- p.33 / Chapter 3.3.1 --- "Eigenface, Fisherface, EGM" --- p.34 / Chapter 3.3.2 --- SVM --- p.35 / Chapter 3.3.3 --- Neural Networks --- p.36 / Chapter 3.4 --- Weight --- p.37 / Chapter 3.5 --- Voting Machine --- p.38 / Chapter 4 --- Dynamic Structure --- p.40 / Chapter 4.1 --- Introduction --- p.40 / Chapter 4.2 --- Architecture --- p.41 / Chapter 4.3 --- Gating Network --- p.42 / Chapter 4.4 --- Feedback Mechanism --- p.44 / Chapter 5 --- Face Recognition System --- p.46 / Chapter 5.1 --- Introduction --- p.46 / Chapter 5.2 --- System Architecture --- p.47 / Chapter 5.2.1 --- Face Detection Module --- p.48 / Chapter 5.2.2 --- Face Recognition Module --- p.49 / Chapter 5.3 --- Face Recognition Process --- p.50 / Chapter 5.3.1 --- Enrollment --- p.51 / Chapter 5.3.2 --- Recognition --- p.52 / Chapter 5.4 --- Distributed System --- p.54 / Chapter 5.4.1 --- Problems --- p.55 / Chapter 5.4.2 --- Distributed Architecture --- p.56 / Chapter 5.5 --- Conclusion --- p.59 / Chapter 6 --- Experimental Results --- p.60 / Chapter 6.1 --- Introduction --- p.60 / Chapter 6.2 --- Database --- p.61 / Chapter 6.2.1 --- ORL Face Database --- p.61 / Chapter 6.2.2 --- Yale Face Database --- p.62 / Chapter 6.2.3 --- AR Face Database --- p.62 / Chapter 6.2.4 --- HRL Face Database --- p.63 / Chapter 6.3 --- Experimental Details --- p.64 / Chapter 6.3.1 --- Pre-processing --- p.64 / Chapter 6.3.2 --- Cross Validation --- p.67 / Chapter 6.3.3 --- System details --- p.68 / Chapter 6.4 --- Result --- p.69 / Chapter 6.4.1 --- ORL Result --- p.69 / Chapter 6.4.2 --- Yale Result --- p.72 / Chapter 6.4.3 --- AR Result --- p.73 / Chapter 6.4.4 --- HRL Result --- p.75 / Chapter 6.4.5 --- Average Running Time --- p.76 / Chapter 6.5 --- Discussion --- p.77 / Chapter 6.5.1 --- Advantages --- p.78 / Chapter 6.5.2 --- Disadvantages --- p.79 / Chapter 6.6 --- Conclusion --- p.80 / Chapter 7 --- Conclusion --- p.82 / Bibliography --- p.92

Fung Cheuk Luk. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2005. / Includes bibliographical references (leaves 104-107). / Abstracts in English and Chinese. / Abstract --- p.ii / 論文摘要 --- p.v / Acknowledgement --- p.v / Notations --- p.vi / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Summary of our approach --- p.2 / Chapter 1.2 --- Summary of our contributions --- p.2 / Chapter 1.3 --- Report Outline --- p.3 / Chapter 2 --- Related Work --- p.4 / Chapter 2.1 --- Face Detection and Tracking literature --- p.4 / Chapter 2.1.1 --- Face Detection approaches --- p.5 / Chapter 2.1.2 --- Face Tracking approaches --- p.6 / Chapter 2.2 --- Overview of Face Detection Procedure --- p.7 / Chapter 2.3 --- Haar-like Feature Cascade Upright Face Detector --- p.12 / Chapter 2.3.1 --- Face Detector Design --- p.12 / Chapter 2.3.2 --- Rectangular Edge Feature f(.) --- p.15 / Chapter 2.3.3 --- Fast Feature Computation Structure: Integral Image --- p.22 / Chapter 2.3.4 --- Feature Selection and parameter estimation --- p.25 / Chapter 2.4 --- Other Related Work --- p.29 / Chapter 2.4.1 --- Rotated Summed Area Table --- p.29 / Chapter 2.4.2 --- Condensation Framework --- p.33 / Chapter 3 --- Rotated Face Detector and Interleaved Face Tracker --- p.35 / Chapter 3.1 --- Rotated Detector Overview --- p.36 / Chapter 3.1.1 --- Parameter Transform and Rotated Face Detection --- p.38 / Chapter 3.1.2 --- Sample Transformation of Detector Parameters --- p.44 / Chapter 3.1.3 --- Post-processing of the detector responses --- p.48 / Chapter 3.2 --- Face Tracking Modeling --- p.51 / Chapter 3.2.1 --- Interleaved Detection --- p.51 / Chapter 3.2.2 --- CONDENSATION filter modeling --- p.53 / Chapter 4 --- Experiments --- p.57 / Chapter 4.1 --- Experiments on Rotated Face Detector --- p.57 / Chapter 4.1.1 --- Rotated Image Face Detector --- p.58 / Chapter 4.1.2 --- Face Image Rotation Test --- p.58 / Chapter 4.1.3 --- Real-life Image Experiment --- p.70 / Chapter 4.1.4 --- CMU Rotated Face Image Test --- p.74 / Chapter 4.2 --- Experiments on Interleaved Face Tracker --- p.82 / Chapter 4.2.1 --- Experiment Parameter Settings --- p.82 / Chapter 4.2.2 --- Moving Face Video Experiment --- p.84 / Chapter 4.2.3 --- Scale Varying Face Video Experiment --- p.90 / Chapter 4.2.4 --- Rotating Face Video Experiment --- p.94 / Chapter 5 --- Conclusion and Discussion --- p.98 / Chapter A --- Feature Selection and Parameter Estimation --- p.101 / Bibliography --- p.104