Sparse representation for face images.

This thesis address issues for face recognition with multi-view face images. Several effective methods are proposed and compared with current state of the art. A novel framework that generalises existing sparse representation-based methods in order to exploit the sharing information to against pose variations of face images is proposed.

Emotional reactions following exposure to idealized bodies predict unhealthy body change attitudes and behaviors in women and men

Objective: We explored the extent to which changes in emotional states following exposure to images of idealized bodies predict unhealthy body change attitudes and behaviors in women and men, and whether particular psychological traits mediate these effects. Method: One hundred thirty-three women and 93 men were assessed for unhealthy attitudes and behaviors related to body weight and muscles using the Eating Disorder Inventory-2 (EDI-2), the Obligatory Exercise Questionnaire, and the strategies to increase muscles subscale of the Body Change Inventory. Psychological traits assessed included body dissatisfaction (EDI-2), internalization of the thin/athletic ideal (Sociocultural Attitudes Towards Appearance Questionnaire-3), body comparison (Body Comparison Scale), self-esteem (Rosenberg Self-Esteem Inventory), depression (Beck Depression Inventory-II), and identity confusion (Self-Concept Clarity Scale). Participants were then exposed to photographs of thin female models and muscular male models, and visual analogue scales were used to measure changes in postexposure state body dissatisfaction, anger, anxiety, and depression.
Results: Postexposure increases in state anger, anxiety, depression, and body dissatisfaction correlated with drive for thinness and disordered eating symptomatology in women, while postexposure increases in state body dissatisfaction correlated with muscle development in men. Analyses revealed that internalization and body comparison mediated these relationships, with trait body dissatisfaction, trait depression, self-esteem, and self-concept/identity confusion serving as mediators for women only. Conclusion: These results are indicative of gender differences in: (a) reactions to idealized bodies; (b) psychological traits that predispose individuals to experience these reactions; and (c) types of body change behavior that are associated with these reactions.

The telescopic eye: aqua optica

The Telescopic Eye: Aqua Optica was an installation work which explores concepts and philosophies related to instrumental, lens based observations of heavens as experienced when seeing through a telescope. In particular it considers how the eye of the observer and the lens of the telescope must act in unison to extend both the perception and conception of that which is remote and beyond direct naked eye sensing - an act of transcendence. A series of projected images (the video) of celestial objects are observed by the viewer through a large spherical aquatic lens which is mounted on a large wooden tripod. These images slowly dissolve and morph into each other. The sequence in which these images are presented is based on approximate chronological order in which these celestial objects were historically observed through the telescope – such as the local bodies of our the solar system starting with the Moon, Jupiter and Saturn (as first observed by Galileo in 1610) and eventually extending out to distant nebula, star clusters and galaxies. For the Transduction exhibition the Telescopic Eye: Aqua Optica was installed outside in a small quadrangle next to the Tech Bar and as the night drew darker it was able to also embody the lighting and other visual elements from the surrounding buildings and streets. The Aqua Optica system often encompasses such local lighting and images as a added layer to the video projection work. In this case (at night and outside at Federation Square) the extent of the layering was very successful and increased dramatically as the night became darker. This multi layering of images invokes (for the viewer) a connection between the immediate environment of the Federation square and cityscape with the vastly distant images of the cosmos. For me such a multilayered imaging represents how scientific instruments of observation can generate a complex and augmented visualisation of reality, one which has historically come to re-define our conception of the cosmos and our place in it both spatially and temporally.

Development of the multi-scale analysis model to simulate strain localization occurring during material processing

Abstract A detailed description of possibilities given by the developed Cellular Automata—Finite Element (CAFE) multi scale model for prediction of the initiation and propagation of micro shear bands and shear bands in metallic materials subjected to plastic deformation is presented in the work. Particular emphasis in defining the criterion for initiation of micro shear and shear bands, as well as in defining the transition rules for the cellular automata, is put on accounting for the physical aspects of these phenomena occurring in two different scales in the material. The proposed approach led to the creation of the real multi scale model of strain localization phenomena. This model predicts material behavior in various thermo-mechanical processes. Selected examples of applications of the developed model to simulations of metal forming processes, which involve strain localization, are presented in the work. An approach based on the Smoothed Particle Hydrodynamic, which allows to overcome difficulties with remeshing in the traditional CAFE method, is a subject of this work as well. In the developed model remeshing becomes possible and difficulties limiting application of the CAFE method to simple deformation processes are solved. Obtained results of numerical simulaA detailed description of possibilities given by the developed Cellular Automata—Finite Element (CAFE) multi scale model for prediction of the initiation and propagation of micro shear bands and shear bands in metallic materials subjected to plastic deformation is presented in the work. Particular emphasis in defining the criterion for initiation of micro shear and shear bands, as well as in defining the transition rules for the cellular automata, is put on accounting for the physical aspects of these phenomena occurring in two different scales in the material. The proposed approach led to the creation of the real multi scale model of strain localization phenomena. This model predicts material behavior in various thermo-mechanical processes. Selected examples of applications of the developed model to simulations of metal forming processes, which involve strain localization, are presented in the work. An approach based on the Smoothed Particle Hydrodynamic, which allows to overcome difficulties with remeshing in the traditional CAFE method, is a subject of this work as well. In the developed model remeshing becomes possible and difficulties limiting application of the CAFE method to simple deformation processes are solved. Obtained results of numerical simulations are compared with the experimental results of cold rolling process to show good predicative capabilities of the developed model.tions are compared with the experimental results of cold rolling process to show good predicative capabilities of the developed model.

Post processing of the hot torsion test results using a multi-dimensional modelling approach

A model selection scheme was extended to a multi-dimensional representation of the hot torsion test torque, twist and twist rate data to calculate partial derivatives of the torque data with respect to twist and twist rate. These enabled calculation of the instantaneous strain and strain rate hardening indices in the Fields and Backofen method. The concept of an iso-parametric shape function has been borrowed from the finite element method for adding twist rate as a dependant variable to the torque-twist models identified by the model selection scheme. Expressions to calculate the hardening indices, when employing a rational model of torsion data, were derived and presented. Subsequently, the models were used for post processing the data and determining hot strength behaviour, taking into account variations of strain and strain rate hardening indices during the deformation. To substantiate the technique, the hot flow behaviour of API-X70 micro-alloyed steel was determined using a range of hot torsion test data for the material. The flow stress obtained using the instantaneous hardening indices were compared with that obtained by the orthodox technique. For the investigated cases, the onset of dynamic recrystallization (DRX) predicted by the presented technique deviated considerably from those obtained when the average indices were used.

Sparse representation with multi-manifold analysis for texture classification from few training images

Texture classification is one of the most important tasks in computer vision field and it has been extensively investigated in the last several decades. Previous texture classification methods mainly used the template matching based methods such as Support Vector Machine and k-Nearest-Neighbour for classification. Given enough training images the state-of-the-art texture classification methods could achieve very high classification accuracies on some benchmark databases. However, when the number of training images is limited, which usually happens in real-world applications because of the high cost of obtaining labelled data, the classification accuracies of those state-of-the-art methods would deteriorate due to the overfitting effect. In this paper we aim to develop a novel framework that could correctly classify textural images with only a small number of training images. By taking into account the repetition and sparsity property of textures we propose a sparse representation based multi-manifold analysis framework for texture classification from few training images. A set of new training samples are generated from each training image by a scale and spatial pyramid, and then the training samples belonging to each class are modelled by a manifold based on sparse representation. We learn a dictionary of sparse representation and a projection matrix for each class and classify the test images based on the projected reconstruction errors. The framework provides a more compact model than the template matching based texture classification methods, and mitigates the overfitting effect. Experimental results show that the proposed method could achieve reasonably high generalization capability even with as few as 3 training images, and significantly outperforms the state-of-the-art texture classification approaches on three benchmark datasets. © 2014 Elsevier B.V. All rights reserved.

Multi-view subspace clustering for face images

In many real-world computer vision applications, such as multi-camera surveillance, the objects of interest are captured by visual sensors concurrently, resulting in multi-view data. These views usually provide complementary information to each other. One recent and powerful computer vision method for clustering is sparse subspace clustering (SSC); however, it was not designed for multi-view data, which break down its linear separability assumption. To integrate complementary information between views, multi-view clustering algorithms are required to improve the clustering performance. In this paper, we propose a novel multi-view subspace clustering by searching for an unified latent structure as a global affinity matrix in subspace clustering. Due to the integration of affinity matrices for each view, this global affinity matrix can best represent the relationship between clusters. This could help us achieve better performance on face clustering. We derive a provably convergent algorithm based on the alternating direction method of multipliers (ADMM) framework, which is computationally efficient, to solve the formulation. We demonstrate that this formulation outperforms other alternatives based on state-of-The-Arts on challenging multi-view face datasets.

Dynamic parallel job scheduling in multi-cluster computing systems

Job scheduling is a complex problem, yet it is fundamental to sustaining and improving the performance of parallel processing systems. In this paper, we address an on-line parallel job scheduling problem in heterogeneous multi-cluster computing systems. We propose a new space-sharing scheduling policy and show that it performs substantially better than the conventional policies.

Effect of the thermomechanical processing parameters on the microstructure-property relationships in C-Mn-Si TRIP steel

In this work a wide range of roughing (deformation in the austenite recrystallised region) and finishing (deformation in the non-recrystallised region) strains and isothermal holding times were used to clarify the effect of processing parameters on the transformation kinetics and mechanical properties of 0.2C-1.55Mn-1.55Si (wt%) TRIP steel. The results have highlighted the complex relationships between multi-phase microstructure and mechanical properties of TRIP steel. The presence of the triclinic carbides, formed during isothermal holding, deteriorated the mechanical properties of steel studied.

Multi-scale analysis of connectivity for image retrieval

Previously, we proposed the concept of connectivity to obtain discriminating shape descriptors. In this paper, we use connectivity to obtain superior distance histograms for multi-scale images. Experiments are performed to evaluate the distance histograms, based on connectivity, for shape-based retrieval of multi-scale images. Item S8 within the MPEG-7 still images content set is used for performing experiments. Experimental results show that the proposed method enhances retrieval performance significantly.

Investigation of processing conditions of melded parts to determine process boundaries

Melding, a novel method for producing seamless joints in thermosetting composites utilising the Quickstep™ process, is explored in this paper. The effect of processing conditions on the quality of melded joins is examined and a set of processing boundaries defined so that the strength of melded joints is optimized. HexPly® 914c pre-preg material was exposed to a range of processing temperatures prior to joining via the melding process. Differential Scanning Calorimetry analysis was carried out to investigate the degree of cure of material prior to final joining, and it was found that minimal cure occurs at temperatures below 120°C. After consolidation and cure of the melded parts, short beam shear testing was conducted to evaluate the strength of the melded interface. Exposure temperatures between 65°C and 120°C were found to optimize short beam shear join strength. Mode I double cantilever beam and mode II end notched flexural tests showed no detrimental effect of elevated exposure temperatures prior to joining.

Fast scene change detection based histogram

This work presents a new approach to detecting the scene change in the successive capture of photographs of a place within equal time interval. This method is based on a gray level histogram of every image. In this method the histogram of an image is processed to modify it for matching with the processed histogram of a reference image. The coefficient of correlation is taken as the measure of matching. As the method does not do any heavy signal processing, and the images are taken successively with a multi-shot digital still camera, it can be applied for real-time processing of such pictures for detection of a scene change. A multi-camera in multi-position approach is also shown to evaluate the change in scene simultaneously from different angles. Both multi-camera and single-camera approaches are compared in detecting a scene change.

Performance analysis of heterogeneous multi-cluster systems

When building a cost-effective high-performance parallel processing system, a performance model is a useful tool for exploring the design space and examining various parameters. However, performance analysis in such systems has proven to be a challenging task that requires the innovative performance analysis tools and methods to keep up with the rapid evolution and ever increasing complexity of such systems. To this end, we propose an analytical model for heterogeneous multi-cluster systems. The model takes into account stochastic quantities as well as network heterogeneity in bandwidth and latency in each cluster. Also, blocking and non-blocking network architecture model is proposed and are used in performance analysis of the system. The message latency is used as the primary performance metric. The model is validated by constructing a set of simulators to simulate different types of clusters, and by comparing the modeled results with the simulated ones.