We continually encounter different ground terrain such as slippery, compliant, uneven, rocky, and irregular terrain when walking, yet we know very little about how individuals safely negotiate this type of complex environment. Furthermore, we know little about how aging affects stability in these situations despite the increased risk of falls and fall-related injuries among older adults. Paramount to our comprehension of how individuals safely traverse challenging ground terrain is to understand how visual information is utilized as vision is the first line of defense for preparing for and/or avoiding potentially hazardous terrain or obstacles. Thus, the objective of this thesis was to provide a better understanding towards how individuals negotiate different ground terrain in the environment to maintain dynamic stability and prevent the occurrence of a fall. In particular, the role of vision and the effects of aging were investigated. Three studies focused on the role of vision while negotiating varying ground terrain while two studies examined stability across these surfaces. Two main conclusions can be drawn from the results of the three studies on the role of vision. First, regardless of age individuals fixate on highly task-relevant areas (i.e. surfaces eventually stepped on) in an on-line manner and by fixating approximately two steps ahead. Second, visual information from the lower visual field is important for negotiating varying ground terrain. This latter finding has implications for older adults who wear multi-focal glasses and suggests that these individuals should be cautious when wearing these glasses in complex environments. In terms of stability, the results suggest that young and older adults demonstrate greater instability when walking across varying unstable ground terrain compared to solid level ground. Older adults are particularly more unstable in the medial-lateral direction when negotiating the challenging terrain, which may explain the frequency of laterally directed falls and increased hip-fracture risk with advancing age. Interestingly, older adults appear more stable in the anterior-posterior direction; although, this can largely be explained by the cautious gait strategy (i.e. slower walking speed and shorter steps) adopted by these individuals. The results of the studies of my thesis provide valuable insight into how individuals safely negotiate different types of challenging ground terrain when walking. Importantly, this knowledge can serve as an initial step in attempting to reduce falling among those at risk.

No description available.

No description available.

Visual processing is not fixed, but changes dynamically depending on the spatiotemporal context of the presented stimulus, and the behavioural task being performed. In this thesis, I describe theoretical and experimental work that was conducted to investigate how and why visual perception and neural responses are altered by the behavioural and statistical context of presented stimuli. The process by which stimulus expectations are acquired and then shape our sensory experiences is not well understood. To investigate this, I conducted a psychophysics experiment where participants were asked to estimate the direction of motion of presented stimuli, with some directions presented more frequently than others. I found that participants quickly developed expectations for the most frequently presented directions and that this altered their perception of new stimuli, inducing biases in the perceived motion direction as well as visual hallucinations in the absence of a stimulus. These biases were well explained by a model that accounted for their behaviour using a Bayesian strategy, combining a learned prior of the stimulus statistics with their sensory evidence using Bayes’ rule. Altering the behavioural context of presented stimuli results in diverse changes to visual neuron responses, including alterations in receptive field structure and firing rates. While these changes are often thought to reflect optimization towards the behavioural task, what exactly is being optimized and why different tasks produce such varying effects is unknown. To account for the effects of a behavioural task on visual neuron responses, I extend previous Bayesian models of visual processing, hypothesizing that the brain learns an internal model that predicts how both the sensory input and the reward received for performing different actions are determined by a common set of explanatory causes. Short-term changes in visual neural responses would thus reflect optimization of this internal model to deal with changes in the sensory environment (stimulus statistics) and behavioural demands (reward statistics), respectively. This framework is used to predict a range of experimentally observed effects of goal-orientated attention on visual neuron responses. Together, these studies provide new insight into how and why sensory processing adapts in response to changes in the environment. The experimental results support the idea of a very plastic visual system, in which prior knowledge is rapidly acquired and used to shape perception. The theoretical work extends previous Bayesian models of sensory processing, to understand how visual neural responses are altered by the behavioural context of presetned stimuli. Finally, these studies provide a unified description of ‘expectations’ and ‘goal-orientated attention’, as corresponding to continuous adaptation of an internal generative model of the world to account for newly received contextual information.

The cerebral cortex of mammals comprises a series of topographic maps, forming sensory and motor areas such as those in the visual, auditory, and somatosensory systems. Understanding the rules that govern the development of these maps and how this topographic organization relates to information processing is critical for the understanding of cortical processing and whole brain function. Previous computational models have shown that topographic maps can develop through a process of self-organization, if spatially localized patches of cortical neurons are activated by particular stimuli. This thesis presents a series of computational models, based on this principle of self-organization, that focus on the development of the map of orientation preference in primary visual cortex (V1). This map is the prototypical example of topographic map development in the brain, and is the most widely studied, however the same self-organizing principles can also apply to maps of many other visual features and maps in many other sensory areas. Experimental evidence indicates that orientation preference maps in V1 develop in a stable way, with the initial layout determined before eye opening. This constraint is at odds with previous self-organizing models, which have used biologically unfounded ad-hoc methods to obtain robust and reliable development. Such mechanisms inherently lead to instability, by causing massive reorganization over time. The first model presented in this thesis (ALISSOM) shows how ad-hoc methods can be replaced with biologically realistic homeostatic mechanisms that lead to development that is both robust and stable. This model shows for the first time how orientation maps can remain stable despite the massive circuit reconstruction and change in visual inputs occurring during development. This model also highlights the requirements for homeostasis in the developing visual circuit. A second model shows how this development can occur using circuitry that is consistent with the known wiring in V1, unlike previous models. This new model, LESI, contains Long-range Excitatory and Short-range Inhibitory connections between model neurons. Instead of direct long-range inhibition, it uses di-synaptic inhibition to ensure that when visual stimuli are at high contrast, long-range excitatory connections have an overall inhibitory influence. The results match previous models in the special case of the high-contrast inputs that drive development most strongly, but show how the behavior relates to the underlying circuitry, and also make it possible to explore effects at a wide range of contrasts. The final part of this thesis explores the differences between rodents and higher mammals that lead to the lack of topographic organization in rodent species. A lack of organization for orientation also implies local disorder in retinotopy, and analysis of retinotopy data from two-photon calcium imaging in mouse (provided by Tom Mrsic- Flogel, University College London) confirms this hypothesis. A self-organizing model is used to investigate how this disorder can arise via variation in either feed-forward connections to V1 or lateral connections within V1, and how the effects of disorder may vary between species. These results suggest that species with and without topographic maps implement similar visual algorithms differing only in the values of some key parameters, rather than having fundamental differences in architecture. Together, these results help us understand how and why neurons develop preferences for visual features such as orientation, and how maps of these neurons are formed. The resulting models represent a synthesis of a large body of experimental evidence about V1 anatomy and function, and offer a platform for developing a more complete explanation of cortical function in future work.

One of the fundamental problems in the field of computer vision is the task of classifying objects, which are present in an image or sequence of images, based on their appearance. This task is commonly referred to as the object recognition problem. A system designed to perform this task must be able to learn visual cues such as shape, colour and texture from examples of objects presented to it. These cues are then later used to identify examples of the known objects in previously unseen scenes. The work presented in this thesis is based on a statistical representation of shape known as a pairwise geometric histogram which has been demonstrated by other researchers in 2-dimensional object recognition tasks. An analysis of the performance of recognition based on this representation has been conducted and a number of contributions to the original recognition algorithm have been made. An important property of an object recognition system is its scalability. This is the. ability of the system to continue performing as the number of known objects is increased. The analysis of the recognition algorithm presented here considers this issue by relating the classification error to the number of stored model objects. An estimate is also made of the number of objects which can be represented uniquely using geometric histograms. One of the main criticisms of the original recognition algorithm based on geometric histograms was the inability to recognise objects at different scales. An algorithm is presented here that is able to recognise objects over a range of scale using the geometric histogram representation. Finally, a novel pairwise geometric histogram representation for arbitrary surfaces has been proposed. This inherits many of the advantages of the 2-dimensional shape descriptor but enables recognition of 3-dimensional object from arbitrary viewpoints.

Much effort has been spent over the last decade in producing so called "Machine Vision" systems for use in robotics, automated inspection, assembly and numerous other fields. Because of the large amount of data involved in an image (typically ¼ MByte) and the complexity of many algorithms used, the processing times required have been far in excess of real time on a VAX-class serial processor. We review a number of image understanding algorithms that compute a globally defined "state", and show that they may be computed using simple local operations that are suited to parallel implementation. In recent years, many massively parallel machines have been designed to apply local operations rapidly across an image. We review several vision machines. We develop an algebraic analysis of the performance of a vision machine and show that, contrary to the commonly-held belief, the time taken to relay images between serial streams can exceed by far the time spent processing. We proceed to investigate the roles that a variety of pipelining techniques might play. We then present three pipelined designs for vision, one of which has been built. This is a parallel pipelined bit slice convolution processor, capable of operating at video rates. This design is examined in detail, and its performance analysed in relation to the theoretical framework of the preceeding chapters. The construction and debugging of the device, which is now operational in its hardware is detailed.

We consider the problem of localizing objects in weakly labeled images/videos. An image/video (e.g., Flickr image and YouTube video) is weakly labeled if it is associated with a tag describing the main object present in the image/video. It is weakly labeled because the tag only indicates the presence/absence of the object, but does not provide the detailed spatial location of the object. Given an image/video with an object tag, our goal is to localize the object in it. In this thesis, we propose two novel techniques to handle this challenging problem. First, we build a video-specific object appearance model and then incorporate temporal consistency information to localize the object. Second, we make use of existing detectors of some other object classes (which we call "familiar objects") to build the appearance model of the unseen object class (i.e., the object of interest). Experimental results show the effectiveness of the proposed methods. / October 2016

No description available.

This thesis addresses the problem of automatic detection and identification of blemishes in digital images of potatoes. Potatoes are an important food crop, with clear unblemished skin being the main factor affecting consumer preference. Potatoes with defects, diseases and blemishes caused by otherwise benign (to human) skin infections, are strongly avoided by consumers. Most potatoes are sorted into dfferent grades by hand, with inevitable mistakes and losses. The currently deployed computer vision systems for sorting potatoes require manual training and have limited accuracy and high unit costs. A further limitation of typical machine vision systems is that the set of image features for pattern recognition has to be designed by the system engineer to work with a specific configuration of produce, imaging system and operating conditions. Such systems typically do not generalise well to other configurations, where the required image features may well differ from those used to design the original system. The objective of the research presented in this thesis is to introduce an automatic method for detecting and identifying blemishes in digital images of potatoes, where the presented solution involves classifying individual pixels. A human expert is required to mark up areas of blemishes and non-blemishes in a set of training images. For blemish detection, each pixel is classified as either blemish or non-blemish. For blemish identification, each pixel is classified according to a number of pre-determined blemish categories. After training, the system should be able to classify individual pixels in new images of previously unseen potatoes with high accuracy. After segmenting the potato from the image background, a very large set of candidate features, based on statistical information relating to the colour and texture of the region surrounding a given pixel, is first extracted. The features include statistical summaries of the whole potato and local regions centred on each pixel as well as the data of the pixel itself. Then an adaptive boosting algorithm (AdaBoost) is used to automatically select the best features for discriminating between blemishes and non-blemishes. The AdaBoost algorithm (Freund and Schapire, 1999) is used to build a classifier, which combines results from so-called "weak" classifiers, each constructed using one of the candidate features, into one "strong" classifier that performs better than any of the weak classifiers alone. With this approach, different features can be selected for different potato varieties, while also handling the natural variation in fresh produce due to different seasons, lighting conditions, etc. For blemish detection, the classifier was trained using a subset of pixels which had been marked as blemish or non-blemish. Tests were done with the full set of features, "lesion experiments" were carried out to explore the impact of removing specific feature types, and experiments were also carried out on methods of speeding up classification both by restricting the number of weak classifiers and restricting the numbers of unique candidate features which can be used to produce weak classifiers. The results were highly accurate with visible examples of disagreement between classifier output and markup being caused by human inaccuracies in the markup rather than classifier inaccuracy. For blemish identification, a set of classifiers were trained on subsets of pixels marked as each blemish class against a subset of pixels drawn from all other classes. For classification, each pixel was tested with all classifiers and assigned to the classifier which returned the highest confidence of a positive result. Experiments were again performed with methods of speeding up classification as well as lesion experiments. Finally, to demonstrate how the system would work in an industrial context, the classification results were summarised for each potato, providing a high overall accuracy in detecting the presence or absence of significant blemish coverage for each blemish type.