Data collection techniques

The objective of this chapter is to provide an overview of traffic data collection that can and should be used for the calibration and validation of traffic simulation models. There are big differences in availability of data from different sources. Some types of data such as loop detector data are widely available and used. Some can be measured with additional effort, for example, travel time data from GPS probe vehicles. Some types such as trajectory data are available only in rare situations such as research projects.

Detection of dynamic primary user with cooperative spectrum sensing

In this paper we demonstrate that existing cooperative spectrum sensing formulated for static primary users cannot accurately detect dynamic primary users regardless of the information fusion method. Performance error occurs as the sensing parameters calculated by the conventional detector result in sensing performance that violates the sensing requirements. Furthermore, the error is accumulated and compounded by the number of cooperating nodes. To address this limitation, we design and implement the duty cycle detection model for the context of cooperative spectrum sensing to accurately calculate the sensing parameters that satisfy the sensing requirements. We show that longer sensing duration is required to compensate for dynamic primary user traffic.

An experimental approach to air gap optimisation for a correction-less small field diode

Aim A recent Monte Carlo based study has shown that it is possible to design a diode that measures small field output factors equivalent to that in water. This is accomplished by placing an appropriate sized air gap above the silicon chip (1) with experimental results subsequently confirming that a particular Monte Carlo design was accurate (2). The aim of this work was to test if a new correction-less diode could be designed using an entirely experimental methodology. Method: All measurements were performed on a Varian iX at a depth of 5 cm, SSD of 95 cm and field sizes of 5, 6, 8, 10, 20 and 30 mm. Firstly, the experimental transfer of kq,clin,kq,msr from a commonly used diode detector (IBA, stereotactic field diode (SFD)) to another diode detector (Sun Nuclear, unshielded diode, (EDGEe)) was tested. These results were compared to Monte Carlo calculated values of the EDGEe. Secondly, the air gap above the EDGEe silicon chip was optimised empirically. Nine different air gap “tops” were placed above the EDGEe (air depth = 0.3, 0.6, 0.9 mm; air width = 3.06, 4.59, 6.13 mm). The sensitivity of the EDGEe was plotted as a function of air gap thickness for the field sizes measured. Results: The transfer of kq,clin,kq,msr from the SFD to the EDGEe was correct to within the simulation and measurement uncertainties. The EDGEe detector can be made “correction-less” for field sizes of 5 and 6 mm, but was ∼2% from being “correction-less” at field sizes of 8 and 10 mm. Conclusion Different materials will perturb small fields in different ways. A detector is only “correction-less” if all these perturbations happen to cancel out. Designing a “correction-less” diode is a complicated process, thus it is reasonable to expect that Monte Carlo simulations should play an important role.

Scene signatures : localised and point-less features for localisation

This paper is about localising across extreme lighting and weather conditions. We depart from the traditional point-feature-based approach as matching under dramatic appearance changes is a brittle and hard thing. Point feature detectors are fixed and rigid procedures which pass over an image examining small, low-level structure such as corners or blobs. They apply the same criteria applied all images of all places. This paper takes a contrary view and asks what is possible if instead we learn a bespoke detector for every place. Our localisation task then turns into curating a large bank of spatially indexed detectors and we show that this yields vastly superior performance in terms of robustness in exchange for a reduced but tolerable metric precision. We present an unsupervised system that produces broad-region detectors for distinctive visual elements, called scene signatures, which can be associated across almost all appearance changes. We show, using 21km of data collected over a period of 3 months, that our system is capable of producing metric localisation estimates from night-to-day or summer-to-winter conditions.

Determination of urinary thymidine glycol using affinity chromatography, HPLC and post-column reaction detection : a biomarker of oxidative DNA damage upon kidney transplantation

Reactive oxygen species are generated during ischaemia-reperfusion of tissue. Oxidation of thymidine by hydroxyl radicals (HO) leads to the formation of 5,6-dihydroxy-5,6-dihydrothymidine (thymidine glycol). Thymidine glycol is excreted in urine and can be used as biomarker of oxidative DNA damage. Time dependent changes in urinary excretion rates of thymidine glycol were determined in six patients after kidney transplantation and in six healthy controls. A new analytical method was developed involving affinity chromatography and subsequent reverse-phase high-performance liquid chromatography (RP-HPLC) with a post-column chemical reaction detector and endpoint fluorescence detection. The detection limit of this fluorimetric assay was 1.6 ng thymidine glycol per ml urine, which corresponds to about half of the physiological excretion level in healthy control persons. After kidney transplantation the urinary excretion rate of thymidine glycol increased gradually reaching a maximum around 48 h. The excretion rate remained elevated until the end of the observation period of 10 days. Severe proteinuria with an excretion rate of up to 7.2 g of total protein per mmol creatinine was also observed immediately after transplantation and declined within the first 24 h of allograft function (0.35 + 0.26 g/mmol creatinine). The protein excretion pattern, based on separation of urinary proteins on sodium dodecyl sulphate-polyacrylamide gel electrophorosis (SDS-PAGE), as well as excretion of individual biomarker proteins, indicated nonselective glomerular and tubular damage. The increased excretion of thymidine glycol after kidney transplantation may be explained by ischaemia-reperfusion induced oxidative DNA damage of the transplanted kidney.

Design and experimental testing of air slab caps which convert commercial electron diodes into dual purpose, correction-free diodes for small field dosimetry

Purpose Two diodes which do not require correction factors for small field relative output measurements are designed and validated using experimental methodology. This was achieved by adding an air layer above the active volume of the diode detectors, which canceled out the increase in response of the diodes in small fields relative to standard field sizes. Methods Due to the increased density of silicon and other components within a diode, additional electrons are created. In very small fields, a very small air gap acts as an effective filter of electrons with a high angle of incidence. The aim was to design a diode that balanced these perturbations to give a response similar to a water-only geometry. Three thicknesses of air were placed at the proximal end of a PTW 60017 electron diode (PTWe) using an adjustable “air cap”. A set of output ratios (ORfclin Det ) for square field sizes of side length down to 5 mm was measured using each air thickness and compared to ORfclin Det measured using an IBA stereotactic field diode (SFD). k fclin, f msr Qclin,Qmsr was transferred from the SFD to the PTWe diode and plotted as a function of air gap thickness for each field size. This enabled the optimal air gap thickness to be obtained by observing which thickness of air was required such that k fclin, f msr Qclin,Qmsr was equal to 1.00 at all field sizes. A similar procedure was used to find the optimal air thickness required to make a modified Sun Nuclear EDGE detector (EDGEe) which s “correction-free” in small field relative dosimetry. In addition, the feasibility of experimentally transferring k fclin, f msr Qclin,Qmsr values from the SFD to unknown diodes was tested by comparing the experimentally transferred k fclin, f msr Qclin,Qmsr values for unmodified PTWe and EDGEe diodes to Monte Carlo simulated values. Results 1.0 mm of air was required to make the PTWe diode correction-free. This modified diode (PTWeair) produced output factors equivalent to those in water at all field sizes (5–50 mm). The optimal air thickness required for the EDGEe diode was found to be 0.6 mm. The modified diode (EDGEeair) produced output factors equivalent to those in water, except at field sizes of 8 and 10 mm where it measured approximately 2% greater than the relative dose to water. The experimentally calculated k fclin, f msr Qclin,Qmsr for both the PTWe and the EDGEe diodes (without air) matched Monte Carlo simulated results, thus proving that it is feasible to transfer k fclin, f msr Qclin,Qmsr from one commercially available detector to another using experimental methods and the recommended experimental setup. Conclusions It is possible to create a diode which does not require corrections for small field output factor measurements. This has been performed and verified experimentally. The ability of a detector to be “correction-free” depends strongly on its design and composition. A nonwater-equivalent detector can only be “correction-free” if competing perturbations of the beam cancel out at all field sizes. This should not be confused with true water equivalency of a detector.

Text recognition approaches for indoor robotics : a comparison

This paper evaluates the performance of different text recognition techniques for a mobile robot in an indoor (university campus) environment. We compared four different methods: our own approach using existing text detection methods (Minimally Stable Extremal Regions detector and Stroke Width Transform) combined with a convolutional neural network, two modes of the open source program Tesseract, and the experimental mobile app Google Goggles. The results show that a convolutional neural network combined with the Stroke Width Transform gives the best performance in correctly matched text on images with single characters whereas Google Goggles gives the best performance on images with multiple words. The dataset used for this work is released as well.

Very small field dosimetry

This thesis investigated in detail the physics of small X-ray fields used in radiotherapy treatments. Because of this work, the ability to accurately measure dose from these very small X-ray fields has been improved in several ways. These include scientifically quantifying when highly accurate measurements are required by introducing the concept of a very small field, and by the invention of a new detector that responds the same in very small fields as in normal fields.

Identification of New Zealand bats in flight from analysis of echolocation calls by artificial neural networks.

Time-expanded and heterodyned echolocation calls of the New Zealand long-tailed Chalinolobus tuberculatus and lesser short-tailed bat Mystacina tuberculata were recorded and digitally analysed. Temporal and spectral parameters were measured from time-expanded calls and power spectra generated for both time-expanded and heterodyned calls. Artificial neural networks were trained to classify the calls of both species using temporal and spectral parameters and power spectra as input data. Networks were then tested using data not previously seen. Calls could be unambiguously identified using parameters and power spectra from time-expanded calls. A neural network, trained and tested using power spectra of calls from both species recorded using a heterodyne detector set to 40 kHz (the frequency with the most energy of the fundamental of C. tuberculatus call), could identify 99% and 84% of calls of C. tuberculatus and M. tuberculata, respectively. A second network, trained and tested using power spectra of calls from both species recorded using a heterodyne detector set to 27 kHz (the frequency with the most energy of the fundamental of M. tuberculata call), could identify 34% and 100% of calls of C. tuberculatus and M. tuberculata, respectively. This study represents the first use of neural networks for the identification of bats from their echolocation calls. It is also the first study to use power spectra of time-expanded and heterodyned calls for identification of chiropteran species. The ability of neural networks to identify bats from their echolocation calls is discussed, as is the ecology of both species in relation to the design of their echolocation calls.

A comparison of the performance of a brand of broad-band and several brands of narrow-band bat detectors in two different habitat types.

The use of bat detectors to monitor bat activity is common. Although several papers have compared the performance of different brands, none have dealt with the effect of different habitats nor have they compared narrow- and broad-band detectors. In this study the performance of four brands of ultrasonic bat detector, including three narrowband and one broad-band model, were compared for their ability to detect a 40 kHz continuous sound of variable amplitude along 100 metre transects. Transects were laid out in two contrasting bat habitat types: grassland and forest. Results showed that the different brands of detector differed in their ability to detect the source in terms of maximum and minimum detectable distance of the source. The rate of sound degradation with distance as measured by each brand was also different. Significant differences were also found in the performance of different brands in open grassland versus deep forest. No significant differences were found within any brand of detector. Though not as sensitive as narrow-band detectors, broad-band models hold an advantage in their ability to identify species where several species are found sympatrically.

Representing team behaviours from noisy data using player role

Due to their unobtrusive nature, vision-based approaches to tracking sports players have been preferred over wearable sensors as they do not require the players to be instrumented for each match. Unfortunately however, due to the heavy occlusion between players, variation in resolution and pose, in addition to fluctuating illumination conditions, tracking players continuously is still an unsolved vision problem. For tasks like clustering and retrieval, having noisy data (i.e. missing and false player detections) is problematic as it generates discontinuities in the input data stream. One method of circumventing this issue is to use an occupancy map, where the field is discretised into a series of zones and a count of player detections in each zone is obtained. A series of frames can then be concatenated to represent a set-play or example of team behaviour. A problem with this approach though is that the compressibility is low (i.e. the variability in the feature space is incredibly high). In this paper, we propose the use of a bilinear spatiotemporal basis model using a role representation to clean-up the noisy detections which operates in a low-dimensional space. To evaluate our approach, we used a fully instrumented field-hockey pitch with 8 fixed high-definition (HD) cameras and evaluated our approach on approximately 200,000 frames of data from a state-of-the-art real-time player detector and compare it to manually labeled data.

Corner detection in images under different noise levels

Corner detection has shown its great importance in many computer vision tasks. However, in real-world applications, noise in the image strongly affects the performance of corner detectors. Few corner detectors have been designed to be robust to heavy noise by now, partly because the noise could be reduced by a denoising procedure. In this paper, we present a corner detector that could find discriminative corners in images contaminated by noise of different levels, without any denoising procedure. Candidate corners (i.e., features) are firstly detected by a modified SUSAN approach, and then false corners in noise are rejected based on their local characteristics. Features in flat regions are removed based on their intensity centroid, and features on edge structures are removed using the Harris response. The detector is self-adaptive to noise since the image signal-to-noise ratio (SNR) is automatically estimated to choose an appropriate threshold for refining features. Experimental results show that our detector has better performance at locating discriminative corners in images with strong noise than other widely used corner or keypoint detectors.

Scale adaptive tracking using mean shift and efficient feature matching

The mean shift tracker has achieved great success in visual object tracking due to its efficiency being nonparametric. However, it is still difficult for the tracker to handle scale changes of the object. In this paper, we associate a scale adaptive approach with the mean shift tracker. Firstly, the target in the current frame is located by the mean shift tracker. Then, a feature point matching procedure is employed to get the matched pairs of the feature point between target regions in the current frame and the previous frame. We employ FAST-9 corner detector and HOG descriptor for the feature matching. Finally, with the acquired matched pairs of the feature point, the affine transformation between target regions in the two frames is solved to obtain the current scale of the target. Experimental results show that the proposed tracker gives satisfying results when the scale of the target changes, with a good performance of efficiency.

Improving robot vision models for object detection through interaction

We propose a method for learning specific object representations that can be applied (and reused) in visual detection and identification tasks. A machine learning technique called Cartesian Genetic Programming (CGP) is used to create these models based on a series of images. Our research investigates how manipulation actions might allow for the development of better visual models and therefore better robot vision. This paper describes how visual object representations can be learned and improved by performing object manipulation actions, such as, poke, push and pick-up with a humanoid robot. The improvement can be measured and allows for the robot to select and perform the `right' action, i.e. the action with the best possible improvement of the detector.

Safety envelope for security

We present an approach for detecting sensor spoofing attacks on a cyber-physical system. Our approach consists of two steps. In the first step, we construct a safety envelope of the system. Under nominal conditions (that is, when there are no attacks), the system always stays inside its safety envelope. In the second step, we build an attack detector: a monitor that executes synchronously with the system and raises an alarm whenever the system state falls outside the safety envelope. We synthesize safety envelopes using a modified machine learning procedure applied on data collected from the system when it is not under attack. We present experimental results that show effectiveness of our approach, and also validate the several novel features that we introduced in our learning procedure.