Sensorised environments offer opportunities in the support of our everyday lives, in particular, towards realising the concepts of 'Ageing in place'. Such environments are capable of allowing occupants to live independently by providing remote monitoring services and by supporting the completion of activities of daily living. This research focuses on augmenting sensorised environments and promoting improved health- care services with video based solutions. The aim was to demonstrate that video based solutions are feasible and have wide usability and potential in health care, elderly care and generally within sensorised environments. This aim was addressed by considering a number of research objectives, which have been investigated and presented as a series of studies within this thesis. Specifically, the first study targeted multiple occupancy within sensorised environments where a solution based on tracking persons through the use of video was proposed. The results show that multiple occupancy can be handled using video and that users can be successfully tracked within an environment. The second study used video to investigate repetitive behaviour patterns in persons with dementia. The experiment showed that the repetitive behaviour can be extracted and successfully analysed using a single camera. Thirdly, a target group of Parkinson's disease patients are considered with whom video analysis is used to build an automated diary describing their changing status over the day. Results showed that the changes in the patient's movement abilities can be revealed from a video. The final study investigated a specific type of movement disorder known as a tremor. A method involving frequency analysis of tremor from video data was validated in a clinical study involving 31 participants. Furthermore, this study resulted in the development of an open-source software application for routine tremor assessment. This thesis offers a contribution to knowledge by demonstrating that video can be used to further augment sensorised environments to support non-invasive remote monitoring and assessment.

Shape-from-shading is a classical problem in computer vision. The aim is to recover 3D surface shape from a single image of an object, based on a photometric analysis of the pattern of shading. Since the amount of light reflected by a surface is a function of the direction of the incident light and the viewer, relative to the local surface normal, image intensity conveys information about surface orientation and hence, indirectly, surface height. This thesis aims to relax some of the assumptions typically made in the shape-from-shading literature. Our aim is to move shape-fromshading away from the overly simplistic assumptions which have limited its applicability to images captured in lab conditions. In particular, we focus on assumptions of surface smoothness, point source illumination and reconstruction in the surface normal domain. In Chapter 3, we relax the assumption of a smooth surface. We exploit a psychology-inspired heuristic that pixels need only have a similar surface orientation if they are both in close proximity and have a similar intensity. This leads to an adaptive smoothing process which is able to preserve fine surface structure. We adapt a geometric shape-from-shading framework to overcome the problem of normals “flipping” between solutions which alternately satisfy data-closeness and smoothness terms. Under the classical assumption of point source illumination, we show that our method significantly outperforms a number of previously reported methods. In Chapter 4, we relax the assumption of illumination being provided by a single point light source. Specifically, we consider environment illumination in which lighting is represented by a spherical function which describes the incident radiance from all directions in the scene. We use the well known result that Lambertian reflectance acts like a low pass filter and hence the convolution of environment lighting and surface reflectance can be efficiently represented using a low order spherical harmonic. With an order-1 approximation, we show how the image irradiance equation can be solved as a quadratically-constrained linear least squares optimisation. The global optimum is found using the method of Lagrange multipliers. The order-2 case is non-convex and prone to converge on local minima if solved using local optimisation. We reformulate the problem as a bilinear system of equations which leads to an efficient and robust solution method. In both cases, we incorporate a structure-preserving smoothness constraint based on ideas from Chapter 3 to regularise the problem. In Chapter 5, we continue with the relaxed illumination assumption (i.e. we model environment illumination), but we develop algorithms which operate in the domain of surface height rather than surface normals. This has the advantage of reducing the dimensionality of the problem at the expense of increased complexity. Moreover, integrability is implicitly enforced via the problem formulation. We describe two contributions. The first is a linear method for recovering surface height directly from images formed by taking ratios between colour channels. In this case, the nonlinear normalisation term is factored out. This allows us to form a linear system of equations relating image intensity and surface height via a finite difference approximation to the surface gradient. Finally, we relax the assumption that the object must be globally convex (i.e. contains no self occlusions). We show that self occluded intensity can be related to unoccluded intensity via a quadratic inequality constraint. This is too weak a constraint to be used for shape-from-shading on its own. However, we use it to develop an occlusion-sensitive surface integration algorithm. We show that the problem can be formulated as a convex optimisation and solved using semidefinite programming.

Human motion analysis is a very important task for computer vision with a spectrum of potential applications. This thesis presents a novel approach to the problem of human motion understanding. The main contribution of the thesis is that fuzzy qualitative description has been developed for studying human motion from image sequences.

No description available.

The main focus of this thesis has been considering the benefits of using evolutionary based methods for object recognition purposes. The work has tested the hypothesis that evolutionary algorithms can offer solutions to difficult machine vision object recognition problems defined as optimisation problems.

Magnetoencephalography (MEG) is a non-invasive brain imaging technique which allows instant tracking of changes in brain activity. However, it is affected by strong artefact signals generated by the heart or the eye blinking. The blind source separation problem is typically encountered in MEG studies when a set of unknown signals, originating from different sources inside or outside the brain, is mixed with an also unknown mixing matrix during their recording. Independent component analysis (ICA) is a recently developed technique which aims to estimate the original sources given only the observed mixtures. ICA can decompose the observed data into the original biological sources. However, ICA suffers from a major intrinsic ambiguity. In particular, it cannot determine the order of extraction of the source signals. Thus, if there are numerous source signals hidden in lengthy MEG recordings, the extraction of the biological signal of interest can be an extremely prolonged procedure. In this thesis, a modification of the ordinary ICA is introduced in order to cope with this ambiguity. In case there is prior knowledge concerning one of the original signals, this information is exploited by adding a penalty/constraint term to the standard ICA quality function in order to favour the extraction of that particular signal. Our approach requires no reference signal, but the knowledge of some statistical property of one of the original sources, namely its autocorrelation function. Our algorithm is validated with simulated data for which the mixing matrix is known, and is also applied to real MEG data to remove artefact signals. Finally, it is demonstrated how ICA can simplify the ill-posed problem of localising the sources/dipoles in the cortex (inverse problem). The advantage of ICA lies in using nonaveraged trials. In addition, there is no need to know in advance the number of dipoles.

No description available.

Computer Vision is a rapidly developing field in which machines process visual data to extract some meaningful information. Digitised images in their pixels and bits serve no purpose of their own. It is only by interpreting the data, and extracting higher level information that a scene can be understood. The algorithms that enable this process are often complex, and data-intensive, limiting the processing rate when implemented in software. Hardware-accelerated implementations provide a significant performance boost that can enable real-time processing. The Trace Transform is a newly proposed algorithm that has been proven effective in image categorisation and recognition tasks. It is flexibly defined allowing the mathematical details to be tailored to the target application. However, it is highly computationally intensive, which limits its applications. Modern heterogeneous FPGAs provide an ideal platform for accelerating the Trace transform for real-time performance, while also allowing an element of flexibility, which highly suits the generality of the Trace transform. This thesis details the implementation of an extensible Trace transform architecture for vision applications, before extending this architecture to a full flexible platform suited to the exploration of Trace transform applications. As part of the work presented,.a general set of architectures for largewindowed median and weighted median filters are presented as required for a number of Trace transform implementations. Finally an acceleration of Hidden Markov Model decoding for person detection is presented. Such a system can be used to extract frames of interest from a video sequence, to be subsequently processed by the Trace transform. All these architectures emphasise the need for considered, platform-driven design in achieving maximum performance through hardware acceleration.

Scene perception is the general task of constructing an interpretation of low-level sense data in terms of high-level objects, whose e number is not known in advance. This task is examined from a Bayesian perspective. Simple examples from musical Machine Listening are provided, but this thesis is intended as a general view of scene perception.

One of the central problems in computer vision is the interpretation of the content of a single image. A particularly interesting example of this is the extraction of the underlying 3D structure apparent in an image, which is especially challenging due to the ambiguity introduced by having no depth information. Nevertheless, knowledge of the regular and predictable nature of the 3D world imposes constraints upon images, which can be used to recover basic structural information. Our work is inspired by the human visual system, which appears to have little difficulty in interpreting complex scenes from only a single viewpoint. Humans are thought to• rely heavily on learned prior knowledge for this. As such we take a machine learning approach, to learn the relationship between appearance and scene structure from training examples. This thesis investigates this challenging area by focusing on the task of plane detection, which is important since planes are a ubiquitous feature of human-made environments. We develop a new plane detection method, which works by learning from labelled training data, and can find planes and estimate their orientation. This is done from a single image, without relying on explicit geometric information, nor requiring depth. This is achieved by first introducing a method to identify whether an individual image region is planar or not, and if so to estimate its orientation with respect to the camera. This is done by describing the image region using basic feature descriptors, and classifying against training data. This forms the core of our plane detector, since by applying it repeatedly to overlapping image regions we can estimate plane likelihood across the image, which is used to segment it into individual planar and non-planar regions. We evaluate both these algorithms against known ground truth, giving good results, and compare to prior work. We also demonstrate an application of this plane detection algorithm, showing how it is useful for visual odometry (localisation of a camera in an unknown environment). This is done by enhancing a planar visual odometry system to detect planes from one frame, thus being able to quickly initialise planes in appropriate locations, avoiding a search over the whole image. This enables rapid extraction of structured maps while exploring, and may increase accuracy over the baseline system.