Jeremy F. Burn

Robust texture features for blurred images using Undecimated Dual-Tree Complex Wavelets

This paper presents a new descriptor for texture classification. The descriptor is rotationally invariant and blur insensitive, which provides great benefits for various applications that suffer from out-of-focus content or involve fast moving or shaking cameras. We employ an Undecimated Dual-Tree Complex Wavelet Transform (UDT-CWT) [1] to extract texture features. As the UDT-CWT fully provides local spatial relationship between scales and subband orientations, we can straightforwardly create bit-planes of the images representing local phases of wavelet coefficients. We also discard some of the finest decomposition levels where are most affected by the blur. A histogram of the resulting code words is created and used as features in texture classification. Experimental results show that our approach outperforms existing methods by up to 40% for synthetic blurs and up to 30% for natural video content due to camera motion when walking.

Episodic Memory and Appetite Regulation in Humans

Psychological and neurobiological evidence implicates hippocampal-dependent memory processes in the control of hunger and food intake. In humans, these have been revealed in the hyperphagia that is associated with amnesia. However, it remains unclear whether ‘memory for recent eating’ plays a significant role in neurologically intact humans. In this study we isolated the extent to which memory for a recently consumed meal influences hunger and fullness over a three-hour period. Before lunch, half of our volunteers were shown 300 ml of soup and half were shown 500 ml. Orthogonal to this, half consumed 300 ml and half consumed 500 ml. This process yielded four separate groups (25 volunteers in each). Independent manipulation of the ‘actual’ and ‘perceived’ soup portion was achieved using a computer-controlled peristaltic pump. This was designed to either refill or draw soup from a soup bowl in a covert manner. Immediately after lunch, self-reported hunger was influenced by the actual and not the perceived amount of soup consumed. However, two and three hours after meal termination this pattern was reversed - hunger was predicted by the perceived amount and not the actual amount. Participants who thought they had consumed the larger 500-ml portion reported significantly less hunger. This was also associated with an increase in the ‘expected satiation’ of the soup 24-hours later. For the first time, this manipulation exposes the independent and important contribution of memory processes to satiety. Opportunities exist to capitalise on this finding to reduce energy intake in humans.

Biomechanically inspired modeling of pedestrian-induced vertical self-excited forces

AbstractAlthough many models of pedestrian dynamic loading have been proposed, possible bidirectional interactions between the walker and the excited structure are generally ignored, particularly for vertical vibrations. This shortcoming has arisen from scarcity of data on gait-adaptation strategies used in the presence of structural motion and, as a consequence, the absence of a credible fundamental pedestrian model capable of capturing the underlying relations between the two dynamic systems. To address this inadequacy of current approaches, a biomechanically inspired inverted-pendulum pedestrian model has been applied to the human-structure interaction problem. The behavior of the model is studied when subjected to vertical motion of the supporting structure, in particular, in relation to potential self-excited forces that can be generated. A mechanism has been identified by which the timing of pedestrian footsteps can be altered subtly, giving a net damping effect on the structure, without necessarily...

Hoof landing velocity is related to track surface properties in trotting horses

The resistance of a track surface to deformation is known to be positively related to the magnitude of foot impact experienced during locomotion. Although passive mechanics suggests that this might be entirely due to the action of the track surface material decelerating the foot, it is also possible that the dynamics of locomotion are altered in a way that changes the landing velocity of the foot. The observed relationship between track properties and foot impact would then be due to a combination of the direct effect of the surface material and altered foot kinematics at impact. In this study we measured hoof landing velocity, stance time and limb landing angle in horses trotting over surfaces that differed significantly in their deformability. In comparison with a surface that underwent negligible deformation during stance phase, a surface that deformed 25 mm led to significantly increased stance time, significantly greater leg landing angle and significantly greater hoof landing velocity. Although the increased hoof landing velocity would act to counteract the increased shock absorption on the softer surface, we suggest that this effect is relatively small.

Scaling of elastic energy storage in mammalian limb tendons: do small mammals really lose out?

It is widely believed that elastic energy storage is more important in the locomotion of larger mammals. This is based on: (a) comparison of kangaroos with the smaller kangaroo rat; and (b) calculations that predict that the capacity for elastic energy storage relative to body mass increases with size. Here we argue that: (i) data from kangaroos and kangaroo rats cannot be generalized to other mammals; (ii) the elastic energy storage capacity relative to body mass is not indicative of the importance of elastic energy to an animal; and (iii) the contribution of elastic energy to the mechanical work of locomotion will not increase as rapidly with size as the mass-specific energy storage capacity, because larger mammals must do relatively more mechanical work per stride. We predict how the ratio of elastic energy storage to mechanical work will change with size in quadrupedal mammals by combining empirical scaling relationships from the literature. The results suggest that the percentage contribution of elastic energy to the mechanical work of locomotion decreases with size, so that elastic energy is more important in the locomotion of smaller mammals. This now needs to be tested experimentally.

Distorting limb design for dynamically similar locomotion

Terrestrial mammals of different sizes tend to move in a dynamically similar manner when travelling at speeds corresponding to equal values of the Froude number. This means that certain dimensionless locomotor parameters, including peak vertical ground reaction force relative to body weight, stride length relative to leg length and duty factor, are independent of animal size. The Froude number is consequently used to define equivalent speeds for mammals of different sizes. However, most musculoskeletal–tissue properties, including tendon elastic modulus, do not scale in a dynamically similar manner. Therefore, mammals could not be completely dynamically similar, even if perfectly geometrically similar. We argue that, for mammals to move in a dynamically similar manner, they must exhibit systematic ‘distortions’ of limb structure with size that compensate for the size independence of the tendon elastic modulus. An implication of this is that comparing mammals at equal Froude numbers cannot remove all size–dependent effects. We show that the previously published allometry of limb moment arms is sufficient to compensate for size–independent tendon properties. This suggests that it is an important factor in allowing mammals of different sizes to move in a dynamically similar manner.

Experimental identification of the behaviour of and lateral forces from freely-walking pedestrians on laterally oscillating structures in a virtual reality environment

Highlights • A novel setup for investigating pedestrian–structure interaction is presented.• Foot-placement is the main balance control mechanism on laterally vibrating ground.• All components of pedestrian force are uncovered, including self-excited forces.• Inverted pendulum pedestrian model qualitatively captures the nature of forces.• The ground and visual conditions cause significant changes in pedestrian loading.

Projective image restoration using sparsity regularization

This paper presents a method of image restoration for projective ground images which lie on a projection orthogonal to the camera axis. The ground images are initially transformed using homography, and then the proposed image restoration is applied. The process is performed in the dual-tree complex wavelet transform domain in conjunction with L0 reweighting and L2 minimisation (L0RL2) employed to solve this ill-posed problem. We also propose instant estimation of a blur kernel arising from the projective transform and the subsequent interpolation of sparse data. Subjective results show significant improvement of image quality. Furthermore, classification of surface type at various distances (evaluated using a support vector machine classifier) is also improved for the images restored using our proposed algorithm.

Robust texture features based on undecimated dual-tree complex wavelets and local magnitude binary patterns

Image degradation due to illumination change, blur and noise can have a significant influence on classification performance, and yet no descriptors that perform well under these conditions exist. We propose a novel method for obtaining texture features, robust to these distortions, based on an undecimated dual-tree complex wavelet transform (UDT-CWT)1. As the UDT-CWT provides a local spatial relationship between scales, we can straightforwardly create bit-planes of the images representing local phases of wavelet coefficients. Magnitudes of the UDT-CWT are captured via a local binary pattern (LBP), after discarding some of the finest scales that are most affected by the blur and noise. A histogram of the resulting binary code words then forms the features used in texture classification. Results show that our approach outperforms existing methods, that claim to be invariant to feature degradations.

Using Inertial Data to Enhance Image Segmentation - Knowing Camera Orientation Can Improve Segmentation of Outdoor Scenes

In the context of semantic image segmentation, we show that knowledge of world-centric camera orientation (from an inertial sensor) can be used to improve classification accuracy. This works because certain structural classes (such as the ground) tend to appear in certain positions relative to the viewer. We show that orientation information is useful in conjunction with typical image-based features, and that fusing the two results in substantially better classification accuracy than either alone – we observed an increase from 61% to 71% classification accuracy, over the six classes in our test set, when orientation information was added. The method is applied to segmentation using both points and lines, and we also show that combining points with lines further improves accuracy. This work is done towards our intended goal of visually guided locomotion for either an autonomous robot or human.