The delta-phase spectrum with application to voice activity detection and speaker recognition

For several reasons, the Fourier phase domain is less favored than the magnitude domain in signal processing and modeling of speech. To correctly analyze the phase, several factors must be considered and compensated, including the effect of the step size, windowing function and other processing parameters. Building on a review of these factors, this paper investigates a spectral representation based on the Instantaneous Frequency Deviation, but in which the step size between processing frames is used in calculating phase changes, rather than the traditional single sample interval. Reflecting these longer intervals, the term delta-phase spectrum is used to distinguish this from instantaneous derivatives. Experiments show that mel-frequency cepstral coefficients features derived from the delta-phase spectrum (termed Mel-Frequency delta-phase features) can produce broadly similar performance to equivalent magnitude domain features for both voice activity detection and speaker recognition tasks. Further, it is shown that the fusion of the magnitude and phase representations yields performance benefits over either in isolation.

Noise robust voice activity detection using normal probability testing and time-domain histogram analysis

This paper presents a method of voice activity detection (VAD) suitable for high noise scenarios, based on the fusion of two complementary systems. The first system uses a proposed non-Gaussianity score (NGS) feature based on normal probability testing. The second system employs a histogram distance score (HDS) feature that detects changes in the signal through conducting a template-based similarity measure between adjacent frames. The decision outputs by the two systems are then merged using an open-by-reconstruction fusion stage. Accuracy of the proposed method was compared to several baseline VAD methods on a database created using real recordings of a variety of high-noise environments.

Person-independent facial expression detection using constrained local models

In automatic facial expression detection, very accurate registration is desired which can be achieved via a deformable model approach where a dense mesh of 60-70 points on the face is used, such as an active appearance model (AAM). However, for applications where manually labeling frames is prohibitive, AAMs do not work well as they do not generalize well to unseen subjects. As such, a more coarse approach is taken for person-independent facial expression detection, where just a couple of key features (such as face and eyes) are tracked using a Viola-Jones type approach. The tracked image is normally post-processed to encode for shift and illumination invariance using a linear bank of filters. Recently, it was shown that this preprocessing step is of no benefit when close to ideal registration has been obtained. In this paper, we present a system based on the Constrained Local Model (CLM) which is a generic or person-independent face alignment algorithm which gains high accuracy. We show these results against the LBP feature extraction on the CK+ and GEMEP datasets.

Dropout from internet-based treatment for psychological disorders

Purpose: The purpose of this review was to present an in-depth analysis of literature identifying the extent of dropout from Internet-based treatment programmes for psychological disorders, and literature exploring the variables associated with dropout from such programmes. ----- ----- Methods: A comprehensive literature search was conducted on PSYCHINFO and PUBMED with the keywords: dropouts, drop out, dropout, dropping out, attrition, premature termination, termination, non-compliance, treatment, intervention, and program, each in combination with the key words Internet and web. A total of 19 studies published between 1990 and April 2009 and focusing on dropout from Internet-based treatment programmes involving minimal therapist contact were identified and included in the review. ----- ----- Results: Dropout ranged from 2 to 83% and a weighted average of 31% of the participants dropped out of treatment. A range of variables have been examined for their association with dropout from Internet-based treatment programmes for psychological disorders. Despite the numerous variables explored, evidence on any specific variables that may make an individual more likely to drop out of Internet-based treatment is currently limited. ----- ----- Conclusions: This review highlights the need for more rigorous and theoretically guided research exploring the variables associated with dropping out of Internet-based treatment for psychological disorders.

Learning articulated motion structures with Bayesian Networks

This paper presents a general methodology for learning articulated motions that, despite having non-linear correlations, are cyclical and have a defined pattern of behavior Using conventional algorithms to extract features from images, a Bayesian classifier is applied to cluster and classify features of the moving object. Clusters are then associated in different frames and structure learning algorithms for Bayesian networks are used to recover the structure of the motion. This framework is applied to the human gait analysis and tracking but applications include any coordinated movement such as multi-robots behavior analysis.

Risk factors for fatal crashes in rural Australia

This paper presents findings from the rural and remote road safety study, conducted in Queensland, Australia, from March 2004 till June 2007, and compares fatal crashes and non-fatal but serious crashes in respect of their environmental, vehicle and operator factors. During the study period there were 613 non-fatal crashes resulting in 684 hospitalised casualties and 119 fatal crashes resulting in 130 fatalities. Additional information from police sources was available on 103 fatal and 309 non-fatal serious crashes. Over three quarters of both fatal and hospitalised casualties were male and the median age in both groups was 34 years. Fatal crashes were more likely to involve speed, alcohol and violations of road rules and fatal crash victims were 2 and a 1/2 times more likely to be unrestrained inside the vehicle than non-fatal casualties, consistent with current international evidence. After controlling for human factors, vehicle and road conditions made a minimal contribution to the seriousness of the crash outcome. Targeted interventions to prevent fatalities on rural and remote roads should focus on reducing speed and drink driving and promoting seatbelt wearing.

Full Circle : a collection of prototypes

This exhibition engages with one of the key issues facing the fashion textiles industry in terms of future sustainability: that of the well being of fashion industry workers in Australia and New Zealand (people). This collection formed the basis of my honours dissertation (completed in New Zealand in 2008) which examines the contribution that design can make to sustainable manufacturing; particularly design for local production and consumption. An important aspect this work is the discussion of source, the work suggests that the made in China syndrome (in reference to the current state of over-consumerism in Australia and New Zealand) could be bought to a close through design to minimize waste and maximize opportunity for ‘people’: in this case both garment workers and the SMEs that employ them. The garments reflect the possibilities of focusing on a local approach that could be put into practice by a framework of SMEs that already exist. In addition the design process is highly transferrable and could be put into practice almost anywhere with minimal set up costs and a design ethos that progresses at the same pace as the skills of workers. This collection is a physical and conceptual embodiment of a source local/make local/sell local approach. The collection is an example of design that demonstrates that this is not an unrealistic ideal and is in fact possible through the development of a sustainable industry, in the sense of people, profit and planet, through adoption of a design process model that stops the waste at the source, by making better use of the raw materials and labour involved in making fashion garments. Although the focus of this research appears to centre on people and profit, this kind of source local/make local/sell local approach also has great benefits in terms of environmental sustainability.

MobileMums: A randomized controlled trial of an SMS-based physical activity intervention

Background Postnatal women (<12 months postpartum) are at increased risk of physical inactivity. Purpose To evaluate the efficacy and feasibility of a theory-based physical activity (PA) intervention delivered to postnatal women primarily via mobile telephone short message service (SMS). Methods Eighty-eight women were randomized to the intervention (n=45) or minimal contact control (n=43) condition. The 12-week intervention consisted of a face-to-face PA goal-setting consultation, a goal-setting magnet, three to five personally tailored SMS/week and a nominated support person who received two SMS per week. SMS content targeted constructs of social cognitive theory. Frequency (days/week) and duration (min/week) of PA participation and walking for exercise were assessed via self-report at baseline, 6 and 13 weeks. Results Intervention participants increased PA frequency by 1.82 days/week (SE±0.18) by 13 weeks (F(2,85)=4.46, p=0.038) and walking for exercise frequency by 1.08 days/ week (SE±0.24) by 13 weeks (F(2,85)=5.38, p=0.02). Positive trends were observed for duration (min/week) of PA and walking for exercise. Conclusions Intervention exposure resulted in increased frequency of PA and walking for exercise in postnatal women.

Listening in to others : in between noise and silence

Since the launch of the ‘Clean Delhi, Green Delhi’ campaign in 2003, slums have become a significant social and political issue in India’s capital city. Through this campaign, the state, in collaboration with Delhi’s middle class through the ‘Bhagidari system’ (literally translated as ‘participatory system’), aims to transform Delhi into a ‘world-class city’ that offers a sanitised, aesthetically appealing urban experience to its citizens and Western visitors. In 2007, Delhi won the bid to host the 2010 Commonwealth Games; since then, this agenda has acquired an urgent, almost violent, impetus to transform Delhi into an environmentally friendly, aesthetically appealing and ‘truly international city’. Slums and slum-dwellers, with their ‘filth, dirt, and noise’, have no place in this imagined city. The violence inflicted upon slum-dwellers, including the denial of their judicial rights, is justified on these accounts. In addition, the juridical discourse since 2000 has ‘re-problematised slums as ‘nuisance’. The rising antagonism of the middle-classes against the poor, supported by the state’s ambition to have a ‘world-class city’, has allowed a new rhetoric to situate the slums in the city. These representations articulate slums as homogenised spaces of experience and identity. The ‘illegal’ status of slum-dwellers, as encroachers upon public space, is stretched to involve ‘social, cultural, and moral’ decadence and depravity. This thesis is an ethnographic exploration of everyday life in a prominent slum settlement in Delhi. It sensually examines the social, cultural and political materiality of slums, and the relationship of slums with the middle class. In doing so, it highlights the politics of sensorial ordering of slums as ‘filthy, dirty, and noisy’ by the middle classes to calcify their position as ‘others’ in order to further segregate, exclude and discriminate the slums. The ethnographic experience in the slums, however, highlights a complex sensorial ordering and politics of its own. Not only are the interactions between diverse communities in slums highly restricted and sensually ordained, but the middle class is identified as a sensual ‘other’, and its sensual practices prohibited. This is significant in two ways. First, it highlights the multiplicity of social, cultural experience and engagement in the slums, thereby challenging its homogenised representation. Second, the ethnographic exploration allowed me to frame a distinct sense of self amongst the slums, which is denied in mainstream discourses, and allowed me to identify the slums’ own ’others’, middle class being one of them. This thesis highlights sound – its production, performances and articulations – as an act with social, cultural, and political implications and manifestations. ‘Noise’ can be understood as a political construct to identify ‘others’ – and both slum-dwellers and the middle classes identify different sonic practices as noise to situate the ‘other’ sonically. It is within this context that this thesis frames the position of Listener and Hearer, which corresponds to their social-political positions. These positions can be, and are, resisted and circumvented through sonic practices. For instance, amplification tactics in the Karimnagar slums, which are understood as ‘uncultured, callous activities to just create more noise’ by the slums’ middle-class neighbours, also serve definite purposes in shaping and navigating the space through the slums’ soundscapes, asserting a presence that is otherwise denied. Such tactics allow the residents to define their sonic territories and scope of sonic performances; they are significant in terms of exerting one’s position, territory and identity, and they are very important in subverting hierarchies. The residents of the Karimnagar slums have to negotiate many social, cultural, moral and political prejudices in their everyday lives. Their identity is constantly under scrutiny and threat. However, the sonic cultures and practices in the Karimnagar slums allow their residents to exert a definite sonic presence – which the middle class has to hear. The articulation of noise and silence is an act manifesting, referencing and resisting social, cultural, and political power and hierarchies.

Segmentation of scenes of mobile objects and demonstrable backgrounds

In this paper we present a real-time foreground–background segmentation algorithm that exploits the following observation (very often satisfied by a static camera positioned high in its environment). If a blob moves on a pixel p that had not changed its colour significantly for a few frames, then p was probably part of the background when its colour was static. With this information we are able to update differentially pixels believed to be background. This work is relevant to autonomous minirobots, as they often navigate in buildings where smart surveillance cameras could communicate wirelessly with them. A by-product of the proposed system is a mask of the image regions which are demonstrably background. Statistically significant tests show that the proposed method has a better precision and recall rates than the state of the art foreground/background segmentation algorithm of the OpenCV computer vision library.

Automatically detecting pain in video through facial action units

In a clinical setting, pain is reported either through patient self-report or via an observer. Such measures are problematic as they are: 1) subjective, and 2) give no specific timing information. Coding pain as a series of facial action units (AUs) can avoid these issues as it can be used to gain an objective measure of pain on a frame-by-frame basis. Using video data from patients with shoulder injuries, in this paper, we describe an active appearance model (AAM)-based system that can automatically detect the frames in video in which a patient is in pain. This pain data set highlights the many challenges associated with spontaneous emotion detection, particularly that of expression and head movement due to the patient's reaction to pain. In this paper, we show that the AAM can deal with these movements and can achieve significant improvements in both the AU and pain detection performance compared to the current-state-of-the-art approaches which utilize similarity-normalized appearance features only.

Robust mean super-resolution for less cooperative NIR iris recognition at a distance and on the move

Less cooperative iris identification systems at a distance and on the move often suffers from poor resolution. The lack of pixel resolution significantly degrades the iris recognition performance. Super-resolution has been considered to enhance resolution of iris images. This paper proposes a pixelwise super-resolution technique to reconstruct a high resolution iris image from a video sequence of an eye. A novel fusion approach is proposed to incorporate information details from multiple frames using robust mean. Experiments on the MBGC NIR portal database show the validity of the proposed approach in comparison with other resolution enhancement techniques.

AITPM national newsletter president's report April 2010

Well, it has been Clem 7 month here in Brisbane and my impression is “so far, so good!” For those of you who know Brisbane, the four lane twin Clem Jones Tunnel (M7) is approximately 4.5km long, and connects Ipswich Road (A7) at the Princess Alexandra Hospital on the south side with Bowen Bridge Road (A3) at the Royal Brisbane Hospital on the north side. There are also south access ramps to the Pacific Motorway and east access ramps to Shafston Avenue (headed to/from Wynnum). Brisbanites have been enjoying a three week no-toll taste test, and I paced through it one evening with minimal fuss. The tunnel seems to have eased the congestion at the Stanley Street on-ramp to the Pacific Motorway quite a bit, and Ipswich Road – Main Street through the ‘Gabba. One must watch the signage carefully, but once we get used to the infrastructure, this will not likely be problematic. It will be interesting to see how traffic behaves when the system settles after tolling, which has likely commenced by the time you’re reading. I believe a passenger car toll is about $4.20 one way but saves about 24 signalised intersection pass-throughs.

Frequency decomposition techniques for increased discriminative 3D facial information capture

Eigen-based techniques and other monolithic approaches to face recognition have long been a cornerstone in the face recognition community due to the high dimensionality of face images. Eigen-face techniques provide minimal reconstruction error and limit high-frequency content while linear discriminant-based techniques (fisher-faces) allow the construction of subspaces which preserve discriminatory information. This paper presents a frequency decomposition approach for improved face recognition performance utilising three well-known techniques: Wavelets; Gabor / Log-Gabor; and the Discrete Cosine Transform. Experimentation illustrates that frequency domain partitioning prior to dimensionality reduction increases the information available for classification and greatly increases face recognition performance for both eigen-face and fisher-face approaches.

Grounding action in visuo-haptic space using experience networks

Traditional approaches to the use of machine learning algorithms do not provide a method to learn multiple tasks in one-shot on an embodied robot. It is proposed that grounding actions within the sensory space leads to the development of action-state relationships which can be re-used despite a change in task. A novel approach called an Experience Network is developed and assessed on a real-world robot required to perform three separate tasks. After grounded representations were developed in the initial task, only minimal further learning was required to perform the second and third task.