Thesis (M. Phil.)--University of Hong Kong, 2006. / Title proper from title frame. Also available in printed format.

<p>Face detection is a preprocessing step for face recognition algorithms. It is the localization of face/faces in an image or image sequence. Once the face(s) are localized, other computer vision algorithms such as face recognition, image compression, camera auto focusing etc are </p><p>applied. Because of the multiple usage areas, there are many research efforts in face processing. Face detection is a challenging computer vision problem because of lighting conditions, a high degree of variability in size, shape, background, color, etc. To build fully </p><p>automated systems, robust and efficient face detection algorithms are required. </p><p> </p><p>Numerous techniques have been developed to detect faces in a single image; in this project we have used a classification-based face detection method using Gabor filter features. We have designed five frequencies corresponding to eight orientations channels for extracting facial features from local images. The feature vector based on Gabor filter is used as the input of the face/non-face classifier, which is a Support Vector Machine (SVM) on a reduced feature </p><p>subspace extracted by using principal component analysis (PCA). </p><p> </p><p>Experimental results show promising performance especially on single face images where 78% accuracy is achieved with 0 false acceptances.</p>

Existing computer programs designed to perform visual recognition of objects suffer from a basic weakness: the inability to spotlight regions in the image that potentially correspond to objects of interest. The brain's mechanisms of visual attention, elucidated by psychophysicists and neurophysiologists, may suggest a solution to the computer's problem of object recognition.

The face inversion effect has been widely documented as an effect of the uniqueness of face processing. Using a computational model, we show that the face inversion effect is a byproduct of expertise with respect to the face object class. In simulations using HMAX, a hierarchical, shape based model, we show that the magnitude of the inversion effect is a function of the specificity of the representation. Using many, sharply tuned units, an ``expert'' has a large inversion effect. On the other hand, if fewer, broadly tuned units are used, the expertise is lost, and this ``novice'' has a small inversion effect. As the size of the inversion effect is a product of the representation, not the object class, given the right training we can create experts and novices in any object class. Using the same representations as with faces, we create experts and novices for cars. We also measure the feasibility of a view-based model for recognition of rotated objects using HMAX. Using faces, we show that transfer of learning to novel views is possible. Given only one training view, the view-based model can recognize a face at a new orientation via interpolation from the views to which it had been tuned. Although the model can generalize well to upright faces, inverted faces yield poor performance because the features change differently under rotation.

One of the key challenges in face perception lies in determining the contribution of different cues to face identification. In this study, we focus on the role of color cues. Although color appears to be a salient attribute of faces, past research has suggested that it confers little recognition advantage for identifying people. Here we report experimental results suggesting that color cues do play a role in face recognition and their contribution becomes evident when shape cues are degraded. Under such conditions, recognition performance with color images is significantly better than that with grayscale images. Our experimental results also indicate that the contribution of color may lie not so much in providing diagnostic cues to identity as in aiding low-level image-analysis processes such as segmentation.

Over the last century, images of faces have become integral components of many institutional identification systems. A driver’s licence, a passport and often even a health care card, all usually feature prominently images representing the face of their bearer as part of the mechanism for linking real-world bodies to institutional records. Increasingly the production, distribution and inspection of these documents is becoming computer-mediated. As photo ID documents become ‘enhanced’ by computerization, the design challenges and compromises become increasingly coded in the hierarchy of gazes aimed at individual faces and their technologically mediated surrogates. In Western visual culture, representations of faces have been incorporated into identity documents since the 15th century when Renaissance portraits were first used to visually and legally establish the social and institutional positions of particular individuals. However, it was not until the 20th century that official identity documents and infrastructures began to include photographic representations of individual faces. This work explores photo ID documents within the context of “the face,”—a theoretical model for understanding relationships of power coded using representations of particular human faces as tokens of identity. “The face” is a product of mythology for linking ideas of stable identity with images of particular human beings. This thesis extends the panoptic model of the body and contributes to the understanding of changes posed by computerization to the norms of constructing institutional identity and interaction based on surrogates of faces. The exploration is guided by four key research questions: What is “the face”? How does it work? What are its origins (or mythologies)? And how is “the face” being transformed through digitization? To address these questions this thesis weaves ideas from theorists including Foucault, Deleuze and Lyon to explore the rise of “the face” as a strategy for governing, sorting, and classifying members of constituent populations. The work re-examines the techno-political value of captured faces as identity data and by tracing the cultural and techno-political genealogies tying faces to ideas of stable institutional identities this thesis demonstrates face-based identity practices are being improvised and reconfigured by computerization and why these practices are significant for understanding the changing norms of interaction between individuals and institutions.

The accomplishment of the γ-alkylation reaction from β-keto esters of tropinone and the enantioselective aziridine formation from nortropinone is first reported. This opened two new paths to develop tropinone enolate chemistry. One is indirect α-alkylation of tropinone, another is the nucleophilic attack from α-C enolate to the nitrogen atom.<p>Seven interesting chiral amines have been synthesized and applied into the enolate chemistry of two interesting precursors of synthesis of natural products: 1,4- cyclohexanedione monoethylene ketal and tropinone.<p>The aldol reaction between the lithium enolate of 1,4-cyclohexanedione monoethylene ketal and benzaldehyde demonstrated the high diastereoselectivity (up to 98% de) and the moderate to high enantioselectivity (up to 75% ee) induced by those chiral lithium amides. On the other hand, high diastereoselectivity (up to 100% de) and the low enantioselectivity were obtained from the aldol reaction of tropinone enolate with benzaldehyde differentiated by chiral lithium amides with extra electron donor atoms.<p>An analysis method to determine enantioselectivity from racemic α-hydroxytropinone was developed. That will, no doubt, benefit the further enantioselective α-hydroxylation reaction of tropinone.

Over the last century, images of faces have become integral components of many institutional identification systems. A driver’s licence, a passport and often even a health care card, all usually feature prominently images representing the face of their bearer as part of the mechanism for linking real-world bodies to institutional records. Increasingly the production, distribution and inspection of these documents is becoming computer-mediated. As photo ID documents become ‘enhanced’ by computerization, the design challenges and compromises become increasingly coded in the hierarchy of gazes aimed at individual faces and their technologically mediated surrogates. In Western visual culture, representations of faces have been incorporated into identity documents since the 15th century when Renaissance portraits were first used to visually and legally establish the social and institutional positions of particular individuals. However, it was not until the 20th century that official identity documents and infrastructures began to include photographic representations of individual faces. This work explores photo ID documents within the context of “the face,”—a theoretical model for understanding relationships of power coded using representations of particular human faces as tokens of identity. “The face” is a product of mythology for linking ideas of stable identity with images of particular human beings. This thesis extends the panoptic model of the body and contributes to the understanding of changes posed by computerization to the norms of constructing institutional identity and interaction based on surrogates of faces. The exploration is guided by four key research questions: What is “the face”? How does it work? What are its origins (or mythologies)? And how is “the face” being transformed through digitization? To address these questions this thesis weaves ideas from theorists including Foucault, Deleuze and Lyon to explore the rise of “the face” as a strategy for governing, sorting, and classifying members of constituent populations. The work re-examines the techno-political value of captured faces as identity data and by tracing the cultural and techno-political genealogies tying faces to ideas of stable institutional identities this thesis demonstrates face-based identity practices are being improvised and reconfigured by computerization and why these practices are significant for understanding the changing norms of interaction between individuals and institutions.

The present thesis investigates whether the emotional background (context) in which a neutral face is viewed changes one’s memory for that face. In Experiment 1, neutral faces were overlaid centrally onto emotional (positive or negative) or neutral background scenes, and recognition memory for faces was assessed. Memory for faces initially encoded in negative contexts was boosted relative to memory for faces initially encoded in neutral contexts. Further investigation was necessary to reveal the mechanism behind the influence that emotional context had on memory for faces. In Experiments 2 and 3 the spotlight theory of attention was tested to examine whether visual attention was mediating the memory effect. The spotlight theory of attention postulates that positive affective states broaden one’s scope of attention, while negative affective states narrow one’s scope of attention (Easterbrook, 1959; Derryberry & Tucker, 1994). According to this theory, the negative contexts may have narrowed attentional scope and therefore led to a richer processing of the face which happened to be presented centrally in Experiment 1, leading to boosted recognition of these faces. To test whether the varying emotional contexts did indeed shift attentional scope, Experiment 2 was designed in which neutral faces were presented once again in positive, negative or neutral contexts, however location of face presentation was peripheral rather than central. Results revealed a loss of the memory boost, for faces paired with negative contexts, reported in Experiment 1. Experiment 3 was designed to test the spotlight theory of attention using an intermixed design in which faces were presented either centrally or peripherally, randomly across trials, in emotional and neutral background scenes. In this experiment, faces were better remembered when they were viewed peripherally in positive, relative to neutral, contexts at time of study. Experiment 4 was designed to assess the validity of the spotlight theory in accounting for how emotional scenes change visual attention, by examining how performance on a flanker task differed when emotionally positive or negative scenes were presented centrally. Results suggest that positive scenes broaden the spotlight of attention, relative to negatives ones. In summary, emotional contexts lead to a boost in memory for faces paired with negative information, and this effect may be due to shifts in attention varied by the valence of the context.

Face detection is a preprocessing step for face recognition algorithms. It is the localization of face/faces in an image or image sequence. Once the face(s) are localized, other computer vision algorithms such as face recognition, image compression, camera auto focusing etc are applied. Because of the multiple usage areas, there are many research efforts in face processing. Face detection is a challenging computer vision problem because of lighting conditions, a high degree of variability in size, shape, background, color, etc. To build fully automated systems, robust and efficient face detection algorithms are required. Numerous techniques have been developed to detect faces in a single image; in this project we have used a classification-based face detection method using Gabor filter features. We have designed five frequencies corresponding to eight orientations channels for extracting facial features from local images. The feature vector based on Gabor filter is used as the input of the face/non-face classifier, which is a Support Vector Machine (SVM) on a reduced feature subspace extracted by using principal component analysis (PCA). Experimental results show promising performance especially on single face images where 78% accuracy is achieved with 0 false acceptances.