Automatic object detection and reconstruction has been one of the main objectives of remote sensing applications. The primary goal of this research work is to develop techniques for the automated production of dense landscape object models by combining various optical imagery with 3D information. It seeks to focus object- identification ratio updating to an applicable level through multi source data fusion with a two fold application building detection and tree detection in urban areas, and impact crater detection on planetary surfaces. The system for urban-area object detection involves focusing by DTM construction, refinement of Rol by data fusion of 3D range data and multi-spectral signatures, and object identification by boundary generalisation and fitting. On the other hand, the impact crater detection system consists of three sub stages: region of interest definition by texture analysis and by edge direction analysis, optimal ellipse generation as a second processing step, and final verification and refinement by template matching. Both systems are applied to four different data sets, after which quality assessments are made by comparing the results with ground truth and show reasonable detection accuracies. In addition, several 3D range data extraction algorithms including efficient image matchers have been developed and tested in both applications.

Bibliography leaves [31] - [40] / viii, 135, [40] leaves : ill. (some col.) ; 30 cm. / Title page, contents and abstract only. The complete thesis in print form is available from the University Library. / Thesis (Ph.D.)--University of Adelaide, Dept. of Physics and Mathematical Physics, 1994

The number of remote sensing platforms and sensors rises almost every year, yet much work on the interpretation of land cover is still carried out using either single images or images from the same source taken at different dates. Two questions could be asked of this proliferation of images: can the information contained in different scenes be used to improve the classification accuracy and, what is the best way to combine the different imagery? Two of these multiple image sources are MODIS on the Terra platform and ETM+ on board Landsat7, which are suitably complementary. Daily MODIS images with 36 spectral bands in 250-1000 m spatial resolution and seven spectral bands of ETM+ with 30m and 16 days spatial and temporal resolution respectively are available. In the UK, cloud cover may mean that only a few ETM+ scenes may be available for any particular year and these may not be at the time of year of most interest. The MODIS data may provide information on land cover over the growing season, such as harvest dates, that is not present in the ETM+ data. Therefore, the primary objective of this work is to develop a methodology for the integration of medium spatial resolution Landsat ETM+ image, with multi-temporal, multi-spectral, low-resolution MODIS \Terra images, with the aim of improving the classification of agricultural land. Additionally other data may also be incorporated such as field boundaries from existing maps. When classifying agricultural land cover of the type seen in the UK, where crops are largely sown in homogenous fields with clear and often mapped boundaries, the classification is greatly improved using the mapped polygons and utilising the classification of the polygon as a whole as an apriori probability in classifying each individual pixel using a Bayesian approach. When dealing with multiple images from different platforms and dates it is highly unlikely that the pixels will be exactly co-registered and these pixels will contain a mixture of different real world land covers. Similarly the different atmospheric conditions prevailing during the different days will mean that the same emission from the ground will give rise to different sensor reception. Therefore, a method is presented with a model of the instantaneous field of view and atmospheric effects to enable different remote sensed data sources to be integrated.

This thesis describes the application of ultrasound to the interrogation of concrete for the retrieval of quantitative information. In particular the use of air-coupled ultrasound is applied for the first time with recent improvement in ultrasonic technology making this possible. Broadband capacitance transducers are used in tandem with pulse compression to deliver and receive ultrasonic signals with greatly improved SNR’s. Pulse compression involves the cross correlation of a chirp signal to record accurate ultrasonic time of flights. This metric is used to makes structural inferences about concrete and to compare contact and non-contact ultrasonic systems. This comparison reveals that concrete strength estimation from ultrasonic pulse velocity (UPV), alone is inaccurate. Other metrics such as aggregate content and humidity should also be considered. A study in to the effect of humidity on the UPV is presented and a correction factor obtained that normalises UPV around a humidity that could be considered normal to a temperate climate. Images of reinforcement bars embedded in concrete are presented using the pulse compression technique. Time-frequency (t-f) analysis is applied to ultrasonic chirp signals. Extensive simulation is carried out and a comparison between three different methods presented. This ensures accurate tracking of the ultrasonic chirp signals, which allows for frequency scattering to be examined. T-f analysis is then applied to real ultrasonic signals and it is shown how frequencies spectrums of received chirps can be de-noised using the Hough transform. Images of embedded defects are then presented. The Superheterodyne technique is then described and applied to concrete interrogation. Although not overly successful it is shown how energy distributions of received tone burst signals vary with time and the need for further work is discussed.

Some mounted ill. / Includes bibliography / xii, [468] leaves : ill. (some col.) ; 30 cm. / Title page, contents and abstract only. The complete thesis in print form is available from the University Library. / Thesis (Ph.D.)--Dept. of Engineering, University of Adelaide, 1984

Remote sensing of the Earth's atmosphere, typically performed in the infrared region of the spectrum, plays an important role in scientific research. In the past the instruments used to perform these observations have been large, massive devices and correspondingly have only been able to be placed on large satellites. There is currently a trend toward smaller Earth observing platforms, so-called micro-satellites, and there is therefore a need for smaller, less massive instruments. Typically these instruments utilise a semiconductor device that responds to incoming infrared radiation in a known way. Such devices are subject to a number of noise sources that reduce their performance. By cooling them to temperatures around 80K it is possible to significantly reduce the amplitude of this noise compared to the incoming radiation of interest, thus increasing the signal-to-noise ratio (SNR). Typically this cooling is performed by a mechanical cooler, but currently many of them are too massive and require too much power to be suitable for use on a small remote sensing satellite. By considering a typical application, a performance target for a miniature cooler was determined to be a heat lift of 200mW at a cold tip temperature of 80K. Hardware has been created to investigate the feasibility of achieving this aim with a hybrid cooler/radiator. The cooler is a miniature integral Stirling machine and uses flat spiral flexures with a newly designed linear motor to drive the compressor piston; the displacer is driven pneumatically. The prototype initially underwent characterisation without first being pre-cooled by the radiator. Although significant cooling was observed (to below 170K), initial characterisation highlighted a low thermal resistance between the warm end of the cooler and the cold tip. With pre-cooling the cold tip was able to reach a minimum no-load cold tip temperature of 92K, and with 200mW applied to the cold tip a temperature of 122.4 K was sustainable. Attempts were made to increase the thermal resistance between the warm end and cold tip by introducing thermal breaks into the regenerator, and whilst these did increase the thermal resistance, the overall performance of the cooler decreased. The concept of a hybrid miniature cooler/radiator has been shown to be feasible. To achieve the target performance of a heat lift of 200mW at 80K further work needs to be performed to characterise loss processes within the cooler and increase the thermal resistance between the warm end and cold tip.

Bibliography: leaves xvii-xxvi / 1 v. (various foliations) : ill ; 30 cm. / Title page, contents and abstract only. The complete thesis in print form is available from the University Library. / Thesis (Ph.D.)--University of Adelaide, Dept. of Electrical and Electronic Engineering, 1985

This thesis concerns the development of optimisation algorithms to determine the relative co-location, (localisation), of a number of freely-flying 'Smart Dust mote' sensor platform elements using a non-deterministic data-set derived from the duplex wireless transmissions between elements. Smart dust motes are miniaturised, microprocessor based, electronic sensor platforms, frequently used for a wide range of remote environmental monitoring applications; including specific climate synoptic observation research and more general meteorology. For the application proposed in this thesis a cluster of the notional smart dust motes are configured to imitate discrete 'Radio Drop Sonde' elements of the wireless enabled monitoring system in use by meteorological research organisations worldwide. This cluster is modelled in software in order to establish the relative positions during the 'flight' ; the normal mode of deployment for the Drop Sonde is by ejection from an aeroplane into an upper-air zone of interest, such as a storm cloud. Therefore the underlying research question is, how to track a number of these independent, duplex wireless linked, free-flying monitoring devices in 3-dimensions and time (to give the monitored data complete spatio-temporal validity). This represents a significant practical challenge, the solution applied in this thesis was to generate 3-dimensional geometries using the only 'real-time' data available; the Radio Signal Strength Indicator (RSSI) data is generated through the 'normal' duplex wireless communications between motes. Individual RSSI values can be considered as a 'representation of the distance magnitude' between wireless devices; when collated into a spatio-temporal data-set it 'encodes' the relative, co-locational, 3-dimensional geometry of all devices in the cluster. The reconstruction, (or decoding), of the 3-dimensional geometries encoded in the spatio-temporal data-set is a complex problem that is addressed through the application of various algorithms. These include, Random Search, and optimisation algorithms, such as the Stochastic Hill-climber, and various forms of Evolutionary Algorithm. It was found that the performance of the geometric reconstruction could be improved through identification of salient aspects of the modelled environment, the result was heuristic operators. In general these led to a decrease in the time taken to reach a convergent solution or a reduction in the number of candidate search space solutions that must be considered. The software model written for this thesis has been implemented to generalise the fundamental characteristics of an optimisation algorithm and to incorporate them into a generic software framework; this then provides the common code to all model algorithms used.