Human identification from video using advanced gait recognition techniques

The solutions proposed in this thesis contribute to improve gait recognition performance in practical scenarios that further enable the adoption of gait recognition into real world security and forensic applications that require identifying humans at a distance. Pioneering work has been conducted on frontal gait recognition using depth images to allow gait to be integrated with biometric walkthrough portals. The effects of gait challenging conditions including clothing, carrying goods, and viewpoint have been explored. Enhanced approaches are proposed on segmentation, feature extraction, feature optimisation and classification elements, and state-of-the-art recognition performance has been achieved. A frontal depth gait database has been developed and made available to the research community for further investigation. Solutions are explored in 2D and 3D domains using multiple images sources, and both domain-specific and independent modality gait features are proposed.

Macular retinal layer thickness in childhood

Purpose To examine macular retinal thickness and retinal layer thickness with spectral domain optical coherence tomography (OCT) in a population of children with normal ocular health and minimal refractive errors. Methods High resolution macular OCT scans from 196 children aged from 4 to 12 years (mean age 8 ± 2 years) were analysed to determine total retinal thickness and the thickness of 6 different retinal layers across the central 5 mm of the posterior pole. Automated segmentation with manual correction was used to derive retinal thickness values. Results The mean total retinal thickness in the central 1 mm foveal zone was 255 ± 16 μm, and this increased significantly with age (mean increase of 1.8 microns per year) in childhood (p<0.001). Age-related increases in thickness of some retinal layers were also observed, with changes of highest statistical significance found in the outer retinal layers in the central foveal region (p<0.01). Significant topographical variations in thickness of each of the retinal layers were also observed (p<0.001). Conclusions Small magnitude, statistically significant increases in total retinal thickness and retinal layer thickness occur from early childhood to adolescence. The most prominent changes appear to occur in the outer retinal layers of the central fovea.

SAIVT-ADMRG @ MediaEval 2014 social event detection

This paper outlines the approach taken by the Speech, Audio, Image and Video Technologies laboratory, and the Applied Data Mining Research Group (SAIVT-ADMRG) in the 2014 MediaEval Social Event Detection (SED) task. We participated in the event based clustering subtask (subtask 1), and focused on investigating the incorporation of image features as another source of data to aid clustering. In particular, we developed a descriptor based around the use of super-pixel segmentation, that allows a low dimensional feature that incorporates both colour and texture information to be extracted and used within the popular bag-of-visual-words (BoVW) approach.

An observation about circular shortest paths: Dealing with additional constraints using branch and bound

Circular shortest paths represent a powerful methodology for image segmentation. The circularity condition ensures that the contour found by the algorithm is closed, a natural requirement for regular objects. Several implementations have been proposed in the past that either promise closure with high probability or ensure closure strictly, but with a mild computational efficiency handicap. Circularity can be viewed as a priori information that helps recover the correct object contour. Our "observation" is that circularity is only one among many possible constraints that can be imposed on shortest paths to guide them to a desirable solution. In this contribution, we illustrate this opportunity under a volume constraint but the concept is generally applicable. We also describe several adornments to the circular shortest path algorithm that proved useful in applications. © 2011 IEEE.

Tubule detection in testis images using boundary weighting and circular shortest path

In studies of germ cell transplantation, measureing tubule diameters and counting cells from different populations using antibodies as markers are very important. Manual measurement of tubule sizes and cell counts is a tedious and sanity grinding work. In this paper, we propose a new boundary weighting based tubule detection method. We first enhance the linear features of the input image and detect the approximate centers of tubules. Next, a boundary weighting transform is applied to the polar transformed image of each tubule region and a circular shortest path is used for the boundary detection. Then, ellipse fitting is carried out for tubule selection and measurement. The algorithm has been tested on a dataset consisting of 20 images, each having about 20 tubules. Experiments show that the detection results of our algorithm are very close to the results obtained manually. © 2013 IEEE.

A thermal expulsion approach to homogeneous large-volume methacrylate monolith preparation; enabling large-scale rapid purification of biomolecules

Numerous efforts have been dedicated to the synthesis of large-volume methacrylate monoliths for large-scale biomolecules purification but most were obstructed by the enormous release of exotherms during preparation, thereby introducing structural heterogeneity in the monolith pore system. A significant radial temperature gradient develops along the monolith thickness, reaching a terminal temperature that supersedes the maximum temperature required for structurally homogenous monoliths preparation. The enormous heat build-up is perceived to encompass the heat associated with initiator decomposition and the heat released from free radical-monomer and monomer-monomer interactions. The heat resulting from the initiator decomposition was expelled along with some gaseous fumes before commencing polymerization in a gradual addition fashion. Characteristics of 80 mL monolith prepared using this technique was compared with that of a similar monolith synthesized in a bulk polymerization mode. An extra similarity in the radial temperature profiles was observed for the monolith synthesized via the heat expulsion technique. A maximum radial temperature gradient of only 4.3°C was recorded at the center and 2.1°C at the monolith peripheral for the combined heat expulsion and gradual addition technique. The comparable radial temperature distributions obtained birthed identical pore size distributions at different radial points along the monolith thickness.

Automated counting of the Northern Pacific Sea Star in the Derwent using shape recognition

A novel shape recognition algorithm was developed to autonomously classify the Northern Pacific Sea Star (Asterias amurenis) from benthic images that were collected by the Starbug AUV during 6km of transects in the Derwent estuary. Despite the effects of scattering, attenuation, soft focus and motion blur within the underwater images, an optimal joint classification rate of 77.5% and misclassification rate of 13.5% was achieved. The performance of algorithm was largely attributed to its ability to recognise locally deformed sea star shapes that were created during the segmentation of the distorted images.

Towards coordinated vision-based docking using an Autonomous Surface Vehicle

This paper describes the development and experimental evaluation of a novel vision-based Autonomous Surface Vehicle with the purpose of performing coordinated docking manoeuvres with a target, such as an Autonomous Underwater Vehicle, on the water’s surface. The system architecture integrates two small processor units; the first performs vehicle control and implements a virtual force obstacle avoidance and docking strategy, with the second performing vision-based target segmentation and tracking. Furthermore, the architecture utilises wireless sensor network technology allowing the vehicle to be observed by, and even integrated within an ad-hoc sensor network. The system performance is demonstrated through real-world experiments.

Segmenting Australian high school students : two-step cluster analysis preliminary results

A one size fits all approach dominates alcohol programs in school settings (Botvin et al., 2007), which may limit program effectiveness (Snyder et al., 2004). Programs tailored to the meet the needs and wants of adolescent groups may be more effective. Limited attention has been directed towards employing a full segmentation process. Where segmentation has been examined, the focus has remained on socio-demographic characteristics and more recently psychographic variables (Mathijssen et al., 2012). The current study aimed to identify whether the addition of behaviour could be used to identify segments. Variables included attitudes towards binge drinking (α = 0.86), behavioral intentions’ (α = 0.97), perceived behavioral control (PBC), injunctive norms (α = 0.94); descriptive norms (α = 0.87), knowledge and reported behaviour. Data was collected from five schools, n = 625 (32.96% girls). Two-Step cluster analysis produced a sample (n = 625) with a silhouette measure of cohesion and separation of 0.4. The intention measure and whether students reported previously consuming alcohol were the most distinguishing characteristics - predictor importance scores of (1.0). A four segment solution emerged. The first segment (“Male abstainers” – 37.2%) featured the highest knowledge score (M: 5.9) along with the lowest-risk drinking attitudes and intentions to drink excessively. Segment 2 (“At risk drinkers” - 11.2%) were characterised by their high-risk attitudes and high-risk drinking intentions. Injunctive (M: 4.1) and descriptive norms (M: 4.9) may indicate a social environment where drinking is the norm. Segment 3 (”Female abstainers” – 25.9%) represents young girls, who have the lowest-risk attitudes and low intentions to drink excessively. The fourth and final segment (boys = 67.4%) (“Moderate drinkers” – 25.7%) all report previously drinking alcohol yet their attitudes and intentions towards excessive alcohol consumption are lower than other segments. Segmentation focuses on identifying groups of individuals who feature similar characteristics. The current study illustrates the importance of including reported behaviour in addition to psychographic and demographic characteristics to identify unique groups to inform intervention planning and design. Key messages The principle of segmentation has received limited attention in the context of school-based alcohol education programs. This research identified four segments amongst 14-16 year high school students, each of which can be targeted with a unique, tailored program to meet the needs and wants of the target audience.

A theoretical approach to segmenting children’s walking behaviour

Childhood obesity is a leading public health concern globally. This study aimed to extend research applying the principle of market segmentation to gain insight into changing the physical activity behaviour of children, particularly their walk to/from school behaviour. It further examined the utility of employing theory, specifically the Theory of Planned Behaviour (TPB), for this purpose. The study demonstrates the usefulness of behavioural, geographic and psychographic variables, as measured by the TPB, in distinguishing segments, offering an important contrast to prior segmentation studies emphasising demographic variables. This result provides empirical evidence of the value of employing the four segmentation bases, extending beyond a demographic focus, and the importance of incorporating behavioural theory in market segmentation. In so doing, this research provides key insights into changing children’s walking behaviour.

An external field prior for the hidden Potts model with application to cone-beam computed tomography

In images with low contrast-to-noise ratio (CNR), the information gain from the observed pixel values can be insufficient to distinguish foreground objects. A Bayesian approach to this problem is to incorporate prior information about the objects into a statistical model. A method for representing spatial prior information as an external field in a hidden Potts model is introduced. This prior distribution over the latent pixel labels is a mixture of Gaussian fields, centred on the positions of the objects at a previous point in time. It is particularly applicable in longitudinal imaging studies, where the manual segmentation of one image can be used as a prior for automatic segmentation of subsequent images. The method is demonstrated by application to cone-beam computed tomography (CT), an imaging modality that exhibits distortions in pixel values due to X-ray scatter. The external field prior results in a substantial improvement in segmentation accuracy, reducing the mean pixel misclassification rate for an electron density phantom from 87% to 6%. The method is also applied to radiotherapy patient data, demonstrating how to derive the external field prior in a clinical context.

Multi-action recognition via stochastic modelling of optical flow and gradients

In this paper we propose a novel approach to multi-action recognition that performs joint segmentation and classification. This approach models each action using a Gaussian mixture using robust low-dimensional action features. Segmentation is achieved by performing classification on overlapping temporal windows, which are then merged to produce the final result. This approach is considerably less complicated than previous methods which use dynamic programming or computationally expensive hidden Markov models (HMMs). Initial experiments on a stitched version of the KTH dataset show that the proposed approach achieves an accuracy of 78.3%, outperforming a recent HMM-based approach which obtained 71.2%.

Using viewer’s facial expression and heart rate for sports video highlights detection

Viewer interests, evoked by video content, can potentially identify the highlights of the video. This paper explores the use of facial expressions (FE) and heart rate (HR) of viewers captured using camera and non-strapped sensor for identifying interesting video segments. The data from ten subjects with three videos showed that these signals are viewer dependent and not synchronized with the video contents. To address this issue, new algorithms are proposed to effectively combine FE and HR signals for identifying the time when viewer interest is potentially high. The results show that, compared with subjective annotation and match report highlights, ‘non-neutral’ FE and ‘relatively higher and faster’ HR is able to capture 60%-80% of goal, foul, and shot-on-goal soccer video events. FE is found to be more indicative than HR of viewer’s interests, but the fusion of these two modalities outperforms each of them.

Principles of experience design for airport terminals

In this thesis, the issue of airport terminal design is examined from a novel perspective: that of the passenger rather than the airport operator. A qualitative approach, based on interviews with 199 passengers at Brisbane International Terminal was adopted. The outcomes of this research make the following three key contributions to existing knowledge: (i) identification of a paradox in the Level of Service metrics, (ii) development of a conceptual model of passenger experience and six design principles and (iii) contribution towards advancing the theoretical knowledge about passengers and their experience in airport terminals.

Content-independent embedding scheme for multi-modal medical image watermarking

Background As the increasing adoption of information technology continues to offer better distant medical services, the distribution of, and remote access to digital medical images over public networks continues to grow significantly. Such use of medical images raises serious concerns for their continuous security protection, which digital watermarking has shown great potential to address. Methods We present a content-independent embedding scheme for medical image watermarking. We observe that the perceptual content of medical images varies widely with their modalities. Recent medical image watermarking schemes are image-content dependent and thus they may suffer from inconsistent embedding capacity and visual artefacts. To attain the image content-independent embedding property, we generalise RONI (region of non-interest, to the medical professionals) selection process and use it for embedding by utilising RONI’s least significant bit-planes. The proposed scheme thus avoids the need for RONI segmentation that incurs capacity and computational overheads. Results Our experimental results demonstrate that the proposed embedding scheme performs consistently over a dataset of 370 medical images including their 7 different modalities. Experimental results also verify how the state-of-the-art reversible schemes can have an inconsistent performance for different modalities of medical images. Our scheme has MSSIM (Mean Structural SIMilarity) larger than 0.999 with a deterministically adaptable embedding capacity. Conclusions Our proposed image-content independent embedding scheme is modality-wise consistent, and maintains a good image quality of RONI while keeping all other pixels in the image untouched. Thus, with an appropriate watermarking framework (i.e., with the considerations of watermark generation, embedding and detection functions), our proposed scheme can be viable for the multi-modality medical image applications and distant medical services such as teleradiology and eHealth.