Martin J. Turner

Selection of optimum wavelengths for holography recording

Holography is an imaging technique which accurately can record both the amplitude and the phase of the scattered light from an object. However, to obtain a hologram in which both the 3D shape and the color of the object are required to be accurately reproduced, the recording of the hologram has to be performed by using at least three laser wavelengths. A mathematical model has been generated in order to simulate the holographic color rendering process by assuming ideal laser recording and reconstruction conditions which ignores the influence caused by the recording material and the processing. Based on this mathematical model a computer program with appropriate graphical user interface was implemented. The required amount of laser wavelengths and their distribution within the visible electromagnetic spectrum has been investigated in order to obtain the best possible color rendering. Simulations using three to seven laser wavelengths have been performed to better understand the sampling nature of color holography and by performing multiple simulations for all possible laser selections the optimum wavelengths have been obtained. We have found that three wavelengths are only sufficient if chosen carefully, but for improved color rendering four to five wavelengths are recommended.

Analysing false positives and 3d structure to create intelligent thresholding and weighting functions for SIFT features

This paper outlines image processes for object detection and feature match weighting utilising stereoscopic image pairs, the Scale Invariant Feature Transform (SIFT) [13,4] and 3D reconstruction. The process is called FEWER; Feature Extraction and Weighting for Enhanced Recognition. The object detection technique is based on noise subtraction utilising the false positive matches from random features. The feature weighting process utilises a 3D spatial information generated from the stereoscopic pairs and 3D feature clusters. The features are divided into three different types, matched from the target to the scene and weighted based on their 3D data and spatial cluster properties. The weightings are computed by analysing a large number of false positive matches and this gives an estimation of the probability that a feature is matched correctly. The techniques described provide increased accuracy, reduces the occurrence of false positives and can create a reduced set of highly relevant features.

FAW for multi-exposure fusion features

This paper introduces a process where fusion features assist matching scale invariant feature transform (SIFT) image features from high contrast scenes. FAW defines the order for extracting features: features, alignment then weighting. The process uses three quality measures to select features from a series of differently exposed images and select a subset of the features in favour of those areas that are defined as well exposed from the different images. The results show an advantage in using these features over features extracted from the common alternative techniques of exposure fusion and tone mapping which extract the features as AWF; alignment, weighting then features. This paper also shows that the process allows for a more robust response when using misaligned or stereoscopic image sets.

Towards object recognition using HDR video, stereoscopic depth information and SIFT

In this paper we propose a framework that will recognise objects from a moving platform using scale invariant features, high dynamic range (HDR) video and stereoscopic depth information. The paper focuses on initial work involving feature extraction from HDR images using SIFT. Initial results show an increase in the number of features extracted from HDR images compared to conventional, low dynamic range (LDR), images.

A High Performance Method for Calculating the Minimum Distance between Two 2D and 3D NURBS Curves

We present a fast, accurate, and robust method to compute the minimum distance between two 2D and 3D NURBS curves. This is carried out by first decomposing both of the NURBS curves into their piecewise-Bézier forms. Candidate pairs, as a subset of all possible pairs, are extracted based on a two-level selection process. The first-level selection uses upper-lower bounds of Bézier subcurves to remove pairs. The second-level selection is based on the spatial relationship test between a pair of Bézier curves. An iterative multidimensional Newton-Raphson method is applied on all candidate pairs in order to calculate the approximate local minimum distances. Finally, by comparing all local minimum distances between a pair of Bézier subcurves, we are able to find the global minimum distance. The accuracy is improved by further use of the multidimensional Newton-Raphson method to give high accuracy. Source code is available online.

Collaborative computational projects: visualisation applications survey

This extended abstract presents initial outcomes from three visualisation user needs surveys, and includes an invitation for new communities to engage with follow-on surveys. Statistical and text cluster analysis have been used to assist specific computational groups; in order to select certain visualisation application packages for software development and to select which new algorithms to implement. This analysis is now also available for advising and creating recommendations to build a long term visualisation support service. The focus of these surveys and this work has been on looking at the use of software toolkits and application packages rather then surveying specific visualisation algorithm techniques.

Design, control, and performance of the ‘weed’ 6 wheel robot in the UK MOD grand challenge

A new locomotion method for unmanned (autonomous) ground vehicles (UGV) is proposed based around six independently driven wheels mounted on three separate modules. Each module is attached to the overall robot via a pivot point and capable of independently controlling its orientation and velocity. This configuration allows the UGV to perform maneuvers conventional vehicles cannot perform, and in particular to control the body orientation separately from the movement direction. The locomotion method is mathematically analyzed to develop appropriate control algorithms and to demonstrate the vehicle performance criteria. A vehicle was constructed according to the proposed configuration and experimentally tested in the UK Ministry Of Defense grand challenge. The performance of the developed locomotion schemes helped the robot make it to the finale of the competition. Graphical Abstract

Visualizing a Spherical Geological Discrete Element Model of Fault Evolution

Discrete Element Modelling (DEM) is a numerical technique that uses a system of interacting discrete bodies to simulate the movement of material being exposed to external forces. This technique is often used to simulate granular systems; however by adding further elements that inter-connect the bodies, it can be used to simulate the deformation of a large volume of material. This method has precedent for use in the Earth Sciences and recently, with the increase of available computing power, it has been put to good use simulating the evolution of extensional faults in large scale crustal experiments that involve over half a million individual spherical bodies. An interactive environment that provides high quality rendering is presented, showing that interactivity is key in allowing the intelligent application of visualization methods such as colour-mapping and visibility thresholds in order to extract fault information from a geological DEM. It is also shown that glyph representation alone is not sufficient to provide full insight into the complex three dimensional geometries of the faults found within the model. To overcome this, a novel use of the MetaBall method is described, which results in implicit surface representations of sphere sub-sets. The surfaces produced are shown to provide greater insight into the faults found within the data but also raise questions as to their meaning.

Discrete element modelling using a parallelised physics engine

Discrete Element Modelling (DEM) is a technique used widely throughout science and engineering. It offers a convenient method with which to numerically simulate a system prone to developing discontinuities within its structure. Often the technique gets overlooked as designing and implementing a model on a scale large enough to be worthwhile can be both time consuming and require specialist programming skills. Currently there are a few notable efforts to produce homogenised software to allow researchers to quickly design and run DEMs with in excess of 1 million elements. However, these applications, while open source, are still complex in nature and require significant input from their original publishers in order for them to include new features as a researcher needs them. Recently software libraries notably from the computer gaming and graphics industries, known as physics engines, have emerged. These are designed specifically to calculate the physical movement and interaction of a system of independent rigid bodies. They provide conceptual equivalents of real world constructions with which an approximation of a realistic scenario can be quickly built. This paper presents a method to utilise the most notable of these engines, NVIDIAs PhysX, to produce a parallelised geological DEM capable of supporting in excess of a million elements.

A Lemon is not a Monstar: visualization of singularities of symmetric second rank tensor fields in the plane

In the visualization of the topology of second rank symmetric tensor fields in the plane one can extract some key points (degenerate points), and curves (separatrices) that characterize the qualitative behaviour of the whole tensor field. This can provide a global structure of the whole tensor field, and effectively reduce the complexity of the original data. To construct this global structure it is important to classify those degenerate points accurately. However, in existing visualization techniques, a degenerate point is only classified into two types: trisector and wedge types. In this work, we will apply the theory from the analysis of binary differential equations and demonstrate that, topologically, a simple degenerate point should be classified into three types: star (trisector), lemon and monstar. The later two types were mistakenly regarded as a single type in the existing visualization techniques.