The central goal of the current thesis was to increase our understanding of how representations of individual faces are built from instances that vary. The prototype effect was used as a tool to probe the nature of our internal face representations. In face recognition, the prototype effect refers to the tendency to recognize, or find familiar, the average image of a face after having studied a series of similar face images. The experiments presented in this thesis investigated the modulating role of different variables on the prototype effect in face recognition. In the study phase, two or more different exemplars based on the same identity were presented. In the test phase, one of the seen exemplars, the unseen prototype, and an unseen exemplar of each studied identity were presented one at a time, and participants were asked to make a recognition judgement about the prior occurrence of either the exact image or the person’s face. Variants of each face identity were either unaltered images of real people’s faces, or they were created artificially by manipulating images of faces using several different techniques. All experiments using artificial variants produced strong prototype effects. The unseen prototype image was recognized more confidently than the actually studied images. This was true even when the variants were so similar that they were barely perceptually discriminable. Importantly, even when participants were given additional exposure to the studied exemplars, no weakening of the prototype effect was observed. Surprisingly, in the experiments using natural images of real people’s faces, no clear recognition advantage for the prototype image was observed. Results suggest that the prototype effect in face recognition might not be tapping an averaging mechanism that operates solely on variations within the same identity.

published_or_final_version / Computer Science and Information Systems / Master / Master of Philosophy

Traditional appearance based face recognition (FR) systems use gray scale images, however recently attention has been drawn to the use of color images. Color inputs have a larger dimensionality, which increases the computational cost, and makes the small sample size (SSS) problem in supervised FR systems more challenging. It is therefore important to determine the scenarios in which usage of color information helps the FR system. In this thesis, it was found that inclusion of chromatic information in FR systems is shown to be particularly advantageous in poor illumination conditions. In supervised systems, a color input of optimal dimensionality would improve the FR performance under SSS conditions. A fusion of decisions from individual spectral planes also helps in the SSS scenario. Finally, chromatic information is integrated into a supervised ensemble learner to address pose and illumination variations. This framework significantly boosts FR performance under a range of learning scenarios.

In eyewitness identification cases, suspect misidentification is the leading factor attributed to wrongful convictions (Scheck, Neufeld, & Dwyer, 2000), thus, it is of applied importance to identify factors that contribute to the false recollection of faces. One potential factor addressed in the current study was whether face memory and subsequent identification for other-race-faces is biased by the degree to which a target face posses facial features associated with ethnic identity. Individual differences in level of processing (global, local) and prejudice were tested as potential mechanisms contributing to biased judgments. In Experiment 1 a standard face recognition task revealed that prejudice, level of processing, and face-type interacted to predict recognition bias. In Experiment 2 results showed that positive misidentifications (i.e., choosing an incorrect foil) were more likely when a stereotypical versus non-stereotypical Black actor was witnessed committing the crime. Results are discussed in terms of theoretical and practical implications.

With more and more images being uploaded to social networks each day, the resources for identifying a large portion of the world are available. However the tools to harness and utilize this information are not sufficient. This thesis presents a system, called PhacePhinder, which can build a face database from a social network and have it accessible from mobile devices. Through combining existing technologies, this is made possible. It also makes use of a fusion probabilistic latent semantic analysis to determine strong connections between users and content. Using this information we can determine the most meaningful social connection to a recognized person, allowing us to inform the user of how they know the person being recognized. We conduct a series of offline and user tests to verify our results and compare them to existing algorithms. We show, that through combining a user’s friendship information as well as picture occurrence information, we can make stronger recommendations than based on friendship alone. We demonstrate a working prototype that can identify a face from a picture taken from a mobile phone, using a database derived from images gathered directly from a social network, and return a meaningful social connection to the recognized face.

Traditional appearance based face recognition (FR) systems use gray scale images, however recently attention has been drawn to the use of color images. Color inputs have a larger dimensionality, which increases the computational cost, and makes the small sample size (SSS) problem in supervised FR systems more challenging. It is therefore important to determine the scenarios in which usage of color information helps the FR system. In this thesis, it was found that inclusion of chromatic information in FR systems is shown to be particularly advantageous in poor illumination conditions. In supervised systems, a color input of optimal dimensionality would improve the FR performance under SSS conditions. A fusion of decisions from individual spectral planes also helps in the SSS scenario. Finally, chromatic information is integrated into a supervised ensemble learner to address pose and illumination variations. This framework significantly boosts FR performance under a range of learning scenarios.

Human face detection and recognition techniques play an important role in applica- tions like face recognition, video surveillance, human computer interface and face image databases. Using color information in images is one of the various possible techniques used for face detection. The novel technique used in this project was the combination of various techniques such as skin color detection, template matching, gradient face de- tection to achieve high accuracy of face detection in frontal faces. The objective in this work was to determine the best rotation angle to achieve optimal detection. Also eye and mouse template matching have been put to test for feature detection.

Feature extraction is an important step in the classification of high-dimensional data such as face images. Furthermore, linear feature extractors are more prevalent due to computational efficiency and preservation of the Gaussianity. This research proposes a simple and fast linear feature extractor approximating the sufficient statistic for Gaussian distributions. This method preserves the discriminatory information in both first and second moments of the data and yields the linear discriminant analysis as a special case. Additionally, an accurate upper bound on the error probability of a plug-in classifier can be used to approximate the number of features minimizing the error probability. Therefore, tighter error bounds are derived in this work based on the Bayes error or the classification error on the trained distributions. These bounds can also be used for performance guarantee and to determine the required number of training samples to guarantee approaching the Bayes classifier performance.

In recent times, there has been an increase in national security awareness with a focus on improving current practices relating to the identification and verification of individuals and the reduction of identity fraud. One tool that has been found to assist in these areas is biometrics. This thesis examines some biometric technologies that may be potentially suitable for surveillance and access control applications, and shows why facial recognition technology has been the focus of this study. / Thesis (PhDElectronicSystemsEngineering)--University of South Australia, 2005.

Thesis (MScIng)--University of Stellenbosch, 2007. / Bibliography. Also available via the Internet.