Dissertations / Theses on the topic 'Reflectance fields'

Human viewers are extremely sensitive to the appearanceof peoples faces, which makes the rendering of realistic human faces a challenging problem. Techniques for doing this have continuously been invented and evolved since more than thirty years.

This thesis makes use of recent methods within the area of image based rendering, namely the acquisition of reflectance fields from human faces. The reflectance fields are used to synthesize and realistically render models of human faces.

A shape from shading technique, assuming that human skin adheres to the Phong model, has been used to estimate surface normals. Belief propagation in graphs has then been used to enforce integrability before reconstructing the surfaces. Finally, the additivity of light has been used to realistically render the models.

The resulting models closely resemble the subjects from which they were created, and can realistically be rendered from novel directions in any illumination environment.

For producing realistic images, reflection is an important visual effect. Reflections of the environment are important not only for highly reflective objects, such as mirrors, but also for more common objects such as brushed metals and glossy plastics. Generating these reflections accurately at real-time rates for interactive applications, however, is a difficult problem. Previous works in this area have made assumptions that sacrifice accuracy in order to preserve interactivity. I will present an algorithm that tries to handle reflection accurately in the general case for real-time rendering. The algorithm uses a database of prerendered environment maps to render both the original object itself and an additional bidirectional reflection distribution function (BRDF). The algorithm performs image-based rendering in reflection space in order to achieve accurate results. It also uses graphics processing unit (GPU) features to accelerate rendering.

The potential of hyperspectral proximal sensing to quantify sward characteristics important in making critical decisions on the management of sheep and dairy pastures in New Zealand has been investigated. Hyperspectral data were acquired using an ASD FieldSpec® Pro FR spectroradiometer attached to the Canopy Pasture Probe (CAPP). The CAPP was developed to enable the collection of in situ reflectance data from New Zealand pasture canopies independent of ambient light conditions. A matt white ceramic tile was selected as a reflectance standard to be used with the CAPP, after testing a variety of materials. Pasture reflectance factor spectra between 350-2500 nm (with spectral resolutions of 3 nm between 350-1000 nm and 10 nm between 1000-2500 nm) and pasture samples were collected from six hill country and lowland areas, across all seasons (August 2006 to September 2007) in a number of regions in the North Island of New Zealand. After pre-processing (e.g. spectral averaging, de-stepping, elimination of noisy wavelengths, smoothing) the spectral data collected from sites were correlated against pasture botanical composition (expressed as proportions of grass, legume and weed) and pasture nutrients (nitrogen, phosphorus, potassium, calcium, magnesium, sodium and sulphur) expressed in percentage of dry matter (%) and amount (kg ha-1) using partial least squares regressions (PLSR). The accuracy and precision of the calibrations were tested using either the full cross-validation leave-one-out method or testing datasets. Regressions were carried out using the reflectance factor data per se and after mathematical transformation, including first derivative, absorbance and continuum-removed spectra. Overall best results were obtained using the first derivative data. The quality of predictions varied greatly with the pasture attribute, site and season. Some reasonable results were achieved for the prediction of pasture grass and legume proportions when analysing samples collected during autumn (grass: R2 > 0.81 and SD/RMSEP 2.3 and legume: R2 > 0.80 and SD/RMSEP 2.2), but predicting pasture weed content was poor for all sites and seasons (R2 = 0.44 and SD/RMSEP = 1.2). The inaccurate predictions might be explained by the fact that the diversity found in the field and observed in the pasture spectral data was not taken into account in the pasture botanical separation. The potential for using proximal sensing techniques to predict pasture nutrients in situ was confirmed, with the sensing of pasture N, P and K increased by the procedure of separating the data according to the season of the year. The full potential of the technology will only be realised if a substantial dataset representing all the variability found in the field is gathered. The importance of obtaining representative datasets that embrace all the biophysical factors (e.g. pasture type, canopy structure) likely to affect the relat ionship, when building prediction calibrations, was highlighted in this research by the variance in the predictions for the same nutrient using different datasets, and by the inconsistency in the number of common wavelengths when examining the wavelengths contributing to the relationship. The ability to use a single model to predict multiple nutrients, or indeed individual nutrients, will only come through a good understanding of the factors likely to influence any calibration function. It has been demonstrated in this research that reasonably accurate and precise pasture nutrient predictions (R2 > 0.74 and SD/RMSEP 2.0) can be made from fresh in situ canopy measurements. This still falls short of the quality of the predictions reported for near infrared reflectance spectroscopy (NIRS) for dried, ground samples analysed under controlled laboratory conditions

The Remote Sensing Group of the Optical Sciences Center at the University of Arizona performs absolute radiometric calibration of Earth-viewing sensors using vicarious methods. The reflectance and irradiance-based methods require the nadir-view reflectance of a calibration site at sensor overpass. Errors in these reflectance data contribute directly to errors in the retrieved at sensor radiance, and therefore errors in the calibration. This research addresses two areas of improvement for the reflectance retrieval. The discreet laboratory data of the reference panel is spectrally interpolated using the measured hemispherical reflectance rather than a polynomial fit. This interpolation better fits an absorption feature of the reference material near 2200 nm. The desired reflectance is due to the directly-transmitted solar irradiance, but field measurements also include irradiance due to diffuse light. Non-lambertian properties of the reference and surface cause the ratio of the reflected total radiances to differ from the ratio of the reflected solar radiances. This difference can be corrected using additional field measurements, shaded surface/shaded-reference, output from a radiative transfer code, RTC-only, or a combination of both, shaded-reference. For the shaded-reference and RTC-only methods the shape of the bi-directional reflectance factor of the surface must be known to better than 10% to maintain a 2% accuracy for the retrievals, while the shaded-surface/shaded-reference method does not use the surface BRF. All three methods were applied to measurements of calibrated reflectance tarpaulins, and to measurements made at White Sands Missile Range. These data demonstrate that the shaded-surface/shaded-reference and RTC-only methods improve the surface reflectance retrieval, while the shaded-reference method is too sensitive to variations between the actual and modeled diffuse sky irradiance to be useful. This research represents significant improvements in the calculation of surface reflectance for vicarious calibration. The hemispherical reflectance interpolation will reduce uncertainties in the short wave infrared by 1%, and the diffuse corrections will reduce the errors in blue by 2% in some cases.

The objectives of this research were to develop and evaluate a field method for in situ measurement of soil properties using visible near-infrared reflectance spectroscopy (Vis-NIRS). A probe with an independent light source for acquiring soil reflectance spectra from soil cores was developed around an existing portable field spectrometer (ASD FieldSpecPro, Boulder, CO, USA; 350-2500 nm). Initial experiments tested the ability of the acquired spectra to predict plant root density, an important property in soil carbon dynamics. Reflectance spectra were acquired from soil containing ryegrass roots (Lolium multiflorum) grown in Allophanic and Fluvial Recent soils in a glasshouse pot trial. Differences in root density were created by differential nitrogen and phosphorus fertilization. Partial least squares regression (PLSR) was used to calibrate spectral data (pre-processed by smoothing and transforming spectra to the first derivative) against laboratory-measured root density data (wet-sieve technique). The calibration model successfully predicted root densities (r2 = 0.85, RPD = 2.63, RMSECV = 0.47 mg cm-3) observed in the pots to a moderate level of accuracy. This soil reflectance probe was then tested using a soil coring system to acquire reflectance spectra from two soils under pasture (0-60 mm soil depths) that had contrasting root densities. The PLSR calibration models for predicting root density were more accurate when soil samples from the two soils were separated rather than grouped. A more accurate prediction was found in Allophanic soils (r2 = 0.83, RPD = 2.44, RMSECV = 1.96 mg g-1) than in Fluvial Recent soils (r2 = 0.75, RPD = 1.98, RMSECV = 5.11 mg g-1). The Vis-NIRS technique was then modified slightly to work on a soil corer that could be used to measure root contents from deeper soil profiles (15- 600 mm depth) in arable land (90-day-old maize crop grown in Fluvial Recent soils). PLSR calibration models were constructed to predict the full range of maize root densities (r2 = 0.83, RPD = 2.42, RMSECV = 1.21 mg cm-3) and also soil carbon (C) and nitrogen (N) concentrations that had been determined in the laboratory (LECO FP- 2000 CNS Analyser; Leco Corp., St Joseph, MI, USA). Further studies concentrated on improving the Vis-NIRS technique for prediction of total C and N concentrations in differing soil types within different soil orders in the field. The soil coring method used in the maize studies was evaluated in permanent and recent pastoral soils (Pumice, Allophanic and Tephric Recent in the Taupo-Rotorua Volcanic Zone, North Island) with a wide range of soil organic matter contents resulting from different times (1-5 years) since conversion from forest soils. Without any sample preparation, other than the soil surface left after coring, it was possible to predict soil C and N concentrations with moderate success (C prediction r2 = 0.75, RMSEP = 1.23%, RPD = 1.97; N prediction r2 = 0.80, RMSEP = 0.10%, RPD = 2.15) using a technique of acquiring soil reflectance spectra from the horizontal cross-section of a soil core (H method). The soil probe was then modified to acquire spectra from the curved vertical wall of a soil core (V method), allowing the spectrometer’s field of view to increase to record the reflectance features of the whole soil sample taken for laboratory analysis. Improved predictions of soil C and N concentrations were achieved with the V method of spectral acquisition (C prediction r2 = 0.97, RMSECV = 0.21%, RPD = 5.80; N prediction r2 = 0.96, RMSECV = 0.02%, RPD = 5.17) compared to the H method (C prediction r2 = 0.95, RMSECV = 0.27%, RPD = 4.45; N prediction r2 = 0.94, RMSECV = 0.03%, RPD = 4.25). The V method was tested for temporal robustness by assessing its ability to predict soil C and N concentrations of Fluvial Recent soils under permanent pasture in different seasons. When principal component analysis (PCA) was used to ensure that the spectral dimensions (which were responsive to water content) of the data set used for developing the PLSR calibration model embraced those of the “unknown” soil samples, it was possible to predict soil C and N concentrations in “unknown” samples of widely different water contents (in May and November), with a high level of accuracy (C prediction r2 = 0.97, RMSEP = 0.36%, RPD = 3.43; N prediction r2 = 0.95, RMSEP = 0.03%, RPD = 3.44). This study indicates that Vis-NIRS has considerable potential for rapid in situ assessment of soil C, N and root density. The results demonstrate that field root densities in pastoral and arable soil can be predicted independently from total soil C, which will allow researchers to predict C sequestration from root production. The recommended “V” technique can be used to assess spatial and temporal variability of soil carbon and nitrogen within soil profiles and across the landscape. It can also be used to assess the rate of C sequestration and organic matter synthesis via root density prediction. It reduces the time, labour and cost of conventional soil analysis and root density measurement.

Boreal peatlands store carbon sequestered from the atmosphere over millennia and the importance of this and the other ecosystem services these areas provide is now widely recognised. However, a changing climate will affect these environments and, consequently, the services they provide to the global population. The rate and direction of environmental change to peatlands is currently unclear and they have not yet been included in many climate models. This may in part be due to the ecological heterogeneity and spatial extent of these areas and the sparse sampling survey methods currently adopted. Hyperspectral remote sensing from satellite platforms may in future offer an approach to surveying and do so at the high spectral and spatial resolutions necessary to infer ecological change in these peatlands. However, work is required to develop methods of analysis to determine if hyperspectral data can be used to measure the overstorey vegetation of these areas. This will require an understanding of how annual and inter-annual cyclical changes affect the peatland plant canopy reflectances that would be recorded by hyperspectral sensors and how these reflectances can be related to state variable of interest to climate scientists, ecologists and peatland managers. There are significant areas of peatland within Scotland and, as it is towards the southern extreme of the boreal peatlands, these may be an early indicator of environment change to the wider boreal region. Calluna vulgaris, a hardy dwarf shrub, is the dominant overstorey species over much of these peatlands and could serve as a proxy for ecological, and consequently, environmental change. However, little has been done to understand how variations in leaf pigments or canopy structural parameters influence the spectral reflectance of Calluna through annual and inter-annual growth and senescence cycles. Nor has much work been done to develop methods of analysis to enable images acquired by hyperspectral remote sensing to be utilised to monitor change to these Calluna dominated peatlands over time. To advance understanding of the optical properties of Calluna leaves and canopies and develop methods to analyse hyperspectral images laboratory, field and modelling studies have been carried out in time series over a number of years. The leaf and canopy parameters significantly affecting reflectance have been identified and quantified. Differences between published Chlorophyll(a+b) in vivo absorption spectra and those determined were found. Carotenoids and Anthocyanins were also identified and quantified. The absorption spectra of these pigments were incorporated into a canopy reflectance model and this was coupled to a Calluna growth model. This combined model enabled the reflectance of Calluna canopies to be modelled in daily increments through annual and inter-annual growth and senescence cycles. Reasonable results were achieved in spectral regions where reflectance changed systematically but only for homogeneous Calluna stands. However, it was noted during this research that the area of support for the spectral measurements appeared to differ from that assumed from the specification provided by the spectroradiometer manufacturers. The directional response functions (DRFs) of two spectroradiometers were investigated and wavelength, or wavelength region, specific spatial dependences were noted. The effect that the DRFs of the spectroradiometers would have on reflectances recorded from Calluna canopies was investigated through a modelling study. Errors and inaccuracies in the spectra that would be recorded from these canopies, and commonly used biochemical indices derived from them, have been quantified.

Introgression of genes from crops into ruderal populations is a multi-step process requiring sympatry, synchronous flowering, chromosomal compatibility, successful pollination and development of the zygote, germination, establishment and reproduction of hybrid progeny. The goal of this thesis was to generate data on as many steps in this process as possible and integrate them into a predictive statistical model to estimate the likelihood of successful introgression under a range of scenarios. Rape (Brassica napus) and wild turnip (B. rapa var. oleifera) were used as a model system. A homozygous dominant mutation in the rape genome conferring herbicide resistance provided a convenient marker for the study of introgression. Potential differences between wild turnip populations from a wide range of geographic locations in New Zealand were examined. Hand pollination established the genetic compatibility of rape and wild turnip and a high potential for gene introgression from rape to wild turnip. Interspecific hybrids were easily generated using wild turnip as the maternal plant, with some minor differences between wild turnip populations. The frequency of successful hybridisation between the two species was higher on the lower raceme. However, the upper raceme produced more dormant interspecific hybrid seed. Field trials, designed to imitate rare rape crop escapes into the ruderal environment, examined the ability of rare rape plants to pollinate wild turnip plants over four summers. At a ratio of 1 rape plant for every 400 wild turnip plants, the incidence of interspecific hybridisation was consistently low (<0.1 to 2.1 % of total seed on wild turnip plants). There was a significant year effect with the first season producing significantly more seed and a greater frequency of interspecific hybrid progeny than the other years. The frequency of interspecific hybrid progeny increases when the ratio of rape: wild turnip plant numbers increases. The relative importance of anemophily and entomophily in the production of interspecific hybrids was examined. Wild turnip plants produced twice as many seeds with bee pollination relative to wind pollination. However, the frequency of interspecific hybrids under wind pollination was nearly twice that for bee pollination. Light reflectance patterns under UV light revealed a marked difference between wild turnip and rape flowers compared to near identical appearance under visible light. The data indicates that bees are able to distinguish between rape and wild turnip flowers and exhibit floral constancy when foraging among populations with these two species. Hybrid survival in the seed bank, germination and seedling establishment in the field are important components of fitness. Seed banks established in the soil after the field trials described above germinated in subsequent spring seasons. The predominantly brassica weed populations were screened for herbicide resistance and the numbers of interspecific hybrids germinating compared to the original frequency in the field trial results. Frequency of interspecific hybrids was reduced in the populations compared to the original seed deposit. Seed with a known frequency of interspecific hybrid seed was sown in a separate trial, and the frequency of interspecific hybrids compared at 0, 4, 6, and 8 weeks after sowing. Poor germination resulted limited competition between seedlings, however the frequency of interspecific hybrids declined over time indicating low plant fitness. There were no significant population effects on any parameters tested. Interspecific hybrids grown in a glasshouse were backcrossed to the parental species and selfed within the plant and within populations. Pollen from the interspecific hybrids was found to have markedly reduced fertility. Interspecific hybrid plants had low female fertility, with the majority (88%) of the pollinated flowers aborting the siliques. Of the remaining siliques, most (98%) had only one to three seeds per silique. Inheritance of the herbicide resistance gene was regular in backcrosses but highly skewed following self pollination with an excess of herbicide-sensitive progeny. Production of a stochastic predictive model integrated the information acquired over the practical work phase of this thesis and utilised the capabilities of @risk, a new application of a risk analysis tool. The three outputs examined were the number of flowering plants resulting from backcrosses to rape and wild turnip and self pollination of the interspecific hybrid progeny. Five scenarios were modelled and all demonstrated the high likelihood of introgression failure in this system. In all scenarios, >75% of simulations resulted in no interspecific hybrid progeny surviving to flowering in the third generation. In all scenarios, and for all three outputs, the seed set on the interspecific hybrids of the second generation was the major factor that limited the number interspecific hybrid progeny surviving to flowering in the third generation.

In this thesis, we propose a method for estimating the parameters of a parametric bidirectional reflectance distribution function (BRDF) for an object surface. The method uses a novel Markov Random Field (MRF) formulation on triplets of corner vertex nodes to model the probability of sets of reflectance parameters for arbitrary reflectance models, given probabilistic surface geometry, camera, illumination, and reflectance image information. In this way, the BRDF parameter estimation problem is cast as a MRF parameter estimation problem. We also present a novel method for estimating the MRF parameters, which uses Population Monte Carlo (PMC) sampling to yield a posterior distribution over the parameters of the BRDF. This PMC based method for estimating the posterior distribution on MRF parameters is compared, using synthetic data, to other parameter estimation methods based on Markov Chain Monte Carlo (MCMC) and Levenberg-Marquardt nonlinear minimization, where it is found to have better results for convergence to the known correct synthetic data parameter sets than the MCMC based methods, and similar convergence results to the LM method. The posterior distributions on the parametric BRDFs for real surfaces, which are represented as evolved sample sets calculated using a Population Monte Carlo algorithm, can be used as features in other high-level vision material or surface classification methods. A variety of probabilistic distances between these features, including the Kullback-Leibler divergence, the Bhattacharyya distance and the Patrick-Fisher distance is used to test the classifiability of the materials, using the PMC evolved sample sets as features. In our experiments on real data, which comprises 48 material surfaces belonging to 12 classes of material, classification errors are counted by comparing the 1-nearest-neighbour classification results to the known (manually specified) material classes. Other classification error statistics such as WNN (worst nearest neighbour) are also calculated. The symmetric Kullback-Leibler divergence, used as a distance measure between the PMC developed sample sets, is the distance measure which gives the best classification results on the real data, when using the 1-nearest neighbour classification method. It is also found that the sets of samples representing the posterior distributions over the MRF parameter spaces are better features for material surface classification than the optimal MRF parameters returned by multiple-seed Levenberg-Marquardt minimization algorithms, which are configured to find the same MRF parameters. The classifiability of the materials is also better when using the entire evolved sample sets (calculated by PMC) as classification features than it is when using only the maximum a-posteriori sample from the PMC evolved sample sets as the feature for each material. It is therefore possible to calculate usable parametric BRDF features for surface classification, using our method.

Thesis (Ph. D.)--Ohio State University, 2005.
Title from first page of PDF file. Document formatted into pages; contains xvi, 145 p.; also includes graphics (some col.) Includes bibliographical references (p. 137-145).

Tethered polymer layers at solid-fluid interfaces are used widely in a variety of surface science applications. Although many of these applications require exposure to dynamic flow conditions, flow field penetration into densely grafted polymer brushes, is still a question open to debate despite the fact that it is a fundamental process crucial to mass transport through these polymer films. Although most theoretical work has indicated flow field penetration into polymer films, with varying predicted penetration depths predicted, the limited experimental attempts to investigate this phenomenon have resulted in inconsistent conclusions due to lack of a proper analytical method. To help resolve this controversy, in this Dissertation, a new spectroscopic method, FRET-TIRFM, based on a combination of Förster resonance energy transfer (FRET) and total internal reflectance fluorescence microscopy (TIRFM), is developed to provide the first direct, quantitative measurements on flow field penetration by measuring linear diffusion coefficients of small molecules through densely grafted, thin poly(N-isopropylacryl-amide) (pNIPAM) films. Decay curves from FRET of the acceptor with a donor covalently attached at the substrate surface were fit to a combined Taylor-Aris-Fickian diffusion model to obtain apparent linear diffusion coefficients of the acceptor molecules for different flow rates. These values can then be used to obtain quantitative estimates of flow field penetration depths. For a pNIPAM surface of 110 nm dry thickness, with a 0.6 chain/nm² grafting density, apparent diffusion coefficients ranging from 1.9-9.1 × 10-12 cm²/s were observed for flow rates ranging from 100 to 3000 μL/min. This increase in apparent diffusion coefficient with applied fluid flow rate is indicative of flow field penetration of the polymer film. The depth of penetration of the flow field is estimated to range from ~6% of the polymer film thickness to ~57% of the film thickness in going from 100 to 3000 μL/min flow rate of a good solvent. Factors other than flow rate that may impact flow field penetration were also investigated using this new FRET-TIRFM method. Solvent quality and polymer brush grafting density are the two most important parameters due to the fact that they influence changes in tethered polymer chain conformation. This work demonstrates that polymer films are most penetrable in a good solvent and least penetrable in a poor solvent under identical flow conditions. These findings are consistent with polymer chain conformational changes going from extended brushes to compact globules. For flow rates ranging from 100 to 3000 μL/min, flow field penetration depth ranges from ~6% of the film thickness to ~57% of film thickness for a good solvent compared to ~4% to ~19% for a poor solvent. Thus, by simply changing solvent quality from good to poor, flow field penetration decreases by about 38%. Grafting density has a less pronounced effect than solvent quality on penetration depth, probably due to the small range of grafting densities chosen for study. However, a roughly 10-20% difference in penetration depth was observed between high density (0.60 chain/nm²) and low density (0.27 chain/nm²) pNIPAM surfaces in the same solvent. Changes in grafting density have a less significant impact in a good solvent compared to a poor solvent. This is most likely caused by the fact that grafting density impacts polymer chain conformation mainly through polymer-polymer repulsion, which becomes less significant in a solvent that better solvates the polymer. For the two extreme cases studied here at flow rates ranging from 100 to 3000 μL/min, the penetration depth is estimated to range from ~4-19% of the original solvent-swollen film thickness for high density pNIPAM films in a poor solvent and from ~7-67% for low density films in a good solvent. One important assumption that underlies all of this work is that the dominant mass transport mechanism for small molecules in dense polymer brushes is diffusion. This assumption was further validated through the use of two different small molecule quenchers, RhB and 2-nitrobenzylalcohol. These molecules are significantly different in size, charge, and structure, and operate by different quenching mechanisms. Despite these differences, the results for flow field penetration are statistically the same for both. These observations validate the assumption of diffusive mass transport in these films.

The Arctic tundra, covering approx. 5.5 % of the Earth’s land surface, is one of the last ecosystems remaining closest to its untouched condition. Remote sensing is able to provide information at regular time intervals and large spatial scales on the structure and function of Arctic ecosystems. But almost all natural surfaces reveal individual anisotropic reflectance behaviors, which can be described by the bidirectional reflectance distribution function (BRDF). This effect can cause significant changes in the measured surface reflectance depending on solar illumination and sensor viewing geometries. The aim of this thesis is the hyperspectral and spectro-directional reflectance characterization of important Arctic tundra vegetation communities at representative Siberian and Alaskan tundra sites as basis for the extraction of vegetation parameters, and the normalization of BRDF effects in off-nadir and multi-temporal remote sensing data. Moreover, in preparation for the upcoming German EnMAP (Environmental Mapping and Analysis Program) satellite mission, the understanding of BRDF effects in Arctic tundra is essential for the retrieval of high quality, consistent and therefore comparable datasets. The research in this doctoral thesis is based on field spectroscopic and field spectro-goniometric investigations of representative Siberian and Alaskan measurement grids. The first objective of this thesis was the development of a lightweight, transportable, and easily managed field spectro-goniometer system which nevertheless provides reliable spectro-directional data. I developed the Manual Transportable Instrument platform for ground-based Spectro-directional observations (ManTIS). The outcome of the field spectro-radiometrical measurements at the Low Arctic study sites along important environmental gradients (regional climate, soil pH, toposequence, and soil moisture) show that the different plant communities can be distinguished by their nadir-view reflectance spectra. The results especially reveal separation possibilities between the different tundra vegetation communities in the visible (VIS) blue and red wavelength regions. Additionally, the near-infrared (NIR) shoulder and NIR reflectance plateau, despite their relatively low values due to the low structure of tundra vegetation, are still valuable information sources and can separate communities according to their biomass and vegetation structure. In general, all different tundra plant communities show: (i) low maximum NIR reflectance; (ii) a weakly or nonexistent visible green reflectance peak in the VIS spectrum; (iii) a narrow “red-edge” region between the red and NIR wavelength regions; and (iv) no distinct NIR reflectance plateau. These common nadir-view reflectance characteristics are essential for the understanding of the variability of BRDF effects in Arctic tundra. None of the analyzed tundra communities showed an even closely isotropic reflectance behavior. In general, tundra vegetation communities: (i) usually show the highest BRDF effects in the solar principal plane; (ii) usually show the reflectance maximum in the backward viewing directions, and the reflectance minimum in the nadir to forward viewing directions; (iii) usually have a higher degree of reflectance anisotropy in the VIS wavelength region than in the NIR wavelength region; and (iv) show a more bowl-shaped reflectance distribution in longer wavelength bands (>700 nm). The results of the analysis of the influence of high sun zenith angles on the reflectance anisotropy show that with increasing sun zenith angles, the reflectance anisotropy changes to azimuthally symmetrical, bowl-shaped reflectance distributions with the lowest reflectance values in the nadir view position. The spectro-directional analyses also show that remote sensing products such as the NDVI or relative absorption depth products are strongly influenced by BRDF effects, and that the anisotropic characteristics of the remote sensing products can significantly differ from the observed BRDF effects in the original reflectance data. But the results further show that the NDVI can minimize view angle effects relative to the contrary spectro-directional effects in the red and NIR bands. For the researched tundra plant communities, the overall difference of the off-nadir NDVI values compared to the nadir value increases with increasing sensor viewing angles, but on average never exceeds 10 %. In conclusion, this study shows that changes in the illumination-target-viewing geometry directly lead to an altering of the reflectance spectra of Arctic tundra communities according to their object-specific BRDFs. Since the different tundra communities show only small, but nonetheless significant differences in the surface reflectance, it is important to include spectro-directional reflectance characteristics in the algorithm development for remote sensing products.
Die arktische Tundra ist mit circa 5,5 % der Landoberfläche eines der letzten großen verbliebenen fast unberührten Ökosysteme unserer Erde. Nur die Fernerkundung ist in der Lage, benötigte Informationen über Struktur und Zustand dieses Ökosystems großräumig und in regelmäßigen Zeitabständen zur Verfügung zu stellen. Aber fast alle natürlichen Oberflächen zeigen individuelle anisotrope Reflexionsverhaltensweisen, welche durch die bidirektionale Reflektanzverteilungsfunktion (englisch: BRDF) beschrieben werden können. Dieser Effekt kann zu erheblichen Veränderungen im gemessenen Reflexionsgrad der Oberfläche in Abhängigkeit von den solaren Beleuchtung- und Blickrichtungsgeometrien führen. Zielstellung dieser Arbeit ist die hyperspektrale und spektro-direktionale Charakterisierung der Oberflächenreflexion wichtiger und repräsentativer arktischer Pflanzengesellschaften in Sibirien und Alaska, als Grundlage für die Extraktion von Vegetationsparametern und die Normalisierung von BRDF-Effekten in Off-Nadir und multi-temporalen Fernerkundungsdaten. In Vorbereitung auf die bevorstehende nationale EnMAP Satellitenmission ist ein Grundverständnis der BRDF-Effekte in der arktischen Tundra von wesentlicher Bedeutung für die Erstellung von hochqualitativen, konsistenten und damit vergleichbaren Datensätzen. Die in dieser Arbeit genutzten Daten beruhen auf geländespektroskopische und geländespektro-goniometrische Untersuchungen von repräsentativen Messflächen in Sibirien und Alaska. Die Entwicklung eines leichten, transportablen und einfach anzuwendenden Geländespektro-Goniometers, welches dennoch zuverlässig Daten liefert, war die erste Aufgabe. Hierfür habe ich ein Gerät mit der Bezeichnung ManTIS („Manual Transportable Instrument platform for ground-based Spectro-directional observations“) entwickelt. Die Ergebnisse der geländespektro-radiometrischen Messungen entlang wichtiger ökologischer Gradienten (regionales Klima, pH-Wert des Bodens, Bodenfeuchte, Toposequenz) zeigen, dass die Pflanzengesellschaften sich anhand ihrer Nadir-Reflektanzen unterscheiden lassen. Insbesondere die Möglichkeit der Differenzierung im sichtbaren (VIS) blauen und roten Wellenlängenbereich. Die Nah-Infrarot (NIR) Schulter und das NIR-Reflektanzplateau sind trotz ihrer niedrigeren Reflektanzwerte eine wertvolle Informationsquelle, die genutzt werden kann um die Pflanzengesellschaften entsprechend ihrer Biomasse und der Vegetationsstruktur voneinander zu unterscheiden. Im Allgemeinen zeigen die verschiedenen Pflanzengesellschaften der Tundra: (i) eine niedrige maximale NIR-Reflektanz; (ii) ein schwaches oder nicht sichtbares lokales Reflektanzmaximum im grünen VIS-Spektrum; (iii) einen schmalen „red-edge“ Bereich zwischen dem roten und NIR-Wellenlängenbereich und (iv) kein deutliches NIR-Reflektanzplateau. Diese gemeinsamen Nadir-Reflektanzeigenschaften sind entscheidend für das Verständnis der Variabilität der BRDF-Effekte in der arktischen Tundra. Keine der untersuchten Pflanzengesellschaften wies isotrope Reflektanzeigenschaften auf. Im Allgemeinen zeigt Tundravegetation: (i) die höchsten BRDF-Effekte in der solaren Hauptebene; (ii) die maximalen Reflexionsgrade in den rückwärts gerichteten Blickrichtungen; (iii) höhere Grade an Anisotropie im VIS-Spektrum als im NIR-Spektrum und (iv) schüsselförmige Reflexionsgradverteilungen in den längeren Wellenlängenbereichen (>700 nm). Die Analyse des Einflusses von hohen Sonnenzenitwinkeln auf die Anisotropie der Rückstrahlung zeigt, dass sich mit zunehmenden Sonnenzenitwinkeln die Anisotropie-Eigenschaften in azimutal-symmetrische schüsselförmige Reflexionsgradverteilungen ändern. Auch ergeben die spektro-direktionalen Analysen, dass Fernerkundungsprodukte wie der NDVI oder die relative Absorptionstiefe stark von BRDF-Effekten beeinflusst werden. Die anisotropen Eigenschaften der Fernerkundungsprodukte können sich erheblich von den beobachteten BRDF-Effekten in den ursprünglichen Reflektanzdaten unterscheiden. Auch lässt sich aus den Ergebnissen ableiten, dass der NDVI relativ gesehen die blickrichtungsabhängigen BRDF-Effekte minimieren kann. Für die untersuchten Pflanzengesellschaften der Tundra weichen die Off-Nadir NDVI-Werte nie mehr als 10 % von den Nadir-NDVI-Werten ab. Im Resümee dieser Studie wird nachgewiesen, dass Änderungen in der Sonnen-Objekt-Sensor-Geometrie direkt zu Reflektanzveränderungen in den Fernerkundungsdaten von arktischen Pflanzengesellschaften der Tundra entsprechend ihrer objekt-spezifischen BRDF-Charakteristiken führen. Da die verschiedenen Arten der Tundravegetation nur kleine, aber signifikante Unterschiede in der Oberflächenreflektanz zeigen, ist es wichtig die spektro-direktionalen Reflexionseigenschaften bei der Entwicklung von Algorithmen für Fernerkundungsprodukte zu berücksichtigen.

This paper presents parameters of evaluation of acoustic quality of the space. It is divided into parts presenting physical principle of the origin and movement of the acoustic signal, principles of its processing with current technology and properties of the acoustic field. This is followed by an analysis of the musical part and notes on psychoacoustics. The document contains a description of relevant parameters of acoustic spaces and way in which we can reach desired results, including material analysis. The paper mainly focuses on description of relevant parameters of three acoustic spaces which were measured. The last part of the work is a program for elementary acoustical measurement, which can be provided by means of commonly accessible equipment such as a notebook or a personal computer.

Breast conserving surgery (BCS) is a common treatment option for breast cancer patients. The goal of BCS is to remove the entire tumor from the breast while preserving as much normal tissue as possible for a better cosmetic outcome after surgery. Specifically, the excised specimen must have at least 2 mm of normal tissue surrounding the diseased mass. Unfortunately, a staggering 20-70% of patients undergoing BCS require repeated surgeries due to the incomplete removal of the tumor diagnosed post-operatively. Due to these high re-excision rates as well as limited post-operative histopathological sampling of the tumor specimen, there is an unmet clinical need for margin assessment. Quantitative diffuse reflectance spectral imaging has previously been explored as a promising, method for providing real-time visual maps of tissue composition to help surgeons determine breast tumor margins to ensure the complete removal of the disease during breast conserving surgery. We have leveraged the underlying sources of contrast in breast tissue, specifically total hemoglobin content, beta-carotene content, and tissue scattering, and developed various fiber optics based spectral imaging systems for this clinical application. Combined with a fast inverse Monte Carlo model of reflectance, previous studies have shown that this technology may be able to decrease re-excision rates for BCS. However, these systems, which all consist of a broadband source, fiber optics probes, an imaging spectrograph and a CCD, have severe limitations in system footprint, tumor area coverage, and speed for acquisition and analysis. The fiber based spectral imaging systems are not scalable to smaller designs that cover a large surveillance area at a very fast speed, which ultimately makes them impractical for use in the clinical environment. The objective of this dissertation was to design, develop, test, and show clinical feasibility of a novel wide field spectral imaging system that utilizes the same scientific principles of previously developed fiber optics based imaging systems, but improves upon the technical issues, such as size, complexity, and speed,to meet the demands of the intra-operative setting.

First, our simple re-design of the system completely eliminated the need for an imaging spectrograph and CCD by replacing them with an array of custom annular photodiodes. The geometry of the photodiodes were designed with the goal of minimizing optical crosstalk, maximizing SNR, and achieving the appropriate tissue sensing depth of up to 2 mm for tumor margin assessment. Without the imaging spectrograph and CCD, the system requires discrete wavelengths of light to launch into the tissue sample. A wavelength selection method that combines an inverse Monte Carlo model and a genetic algorithm was developed in order to optimize the wavelength choices specifically for the underlying breast tissue optical contrast. The final system design consisted of a broadband source with an 8-slot filter wheel containing the optimized set of wavelength choices, an optical light guide and quartz light delivery tube to send the 8 wavelengths of light in free space through the back apertures of each annular photodiode in the imaging array, an 8-channel integrating transimpedance amplifier circuit with a switch box and data acquisition card to collect the reflectance signal, and a laptop computer that controls all the components and analyzes the data.

This newly designed wide field spectral imaging system was tested in tissue-mimicking liquid phantoms and achieved comparable performance to previous clinically-validated fiber optics based systems in its ability to extract optical properties with high accuracy. The system was also tested in various biological samples, including a murine tumor model, porcine tissue, and human breast tissue, for the direct comparison with its fiber optics based counterparts. The photodiode based imaging system achieved comparable or better SNR, comparable extractions of optical properties extractions for all tissue types, and feasible improvements in speed and coverage for future iterations. We show proof of concept in performing fast, wide field spectral imaging with a simple, inexpensive design. With a reduction in size, cost, number of wavelengths used, and overall complexity, the system described by this dissertation allows for a more seamless scaling to higher pixel number and density in future iterations of the technology, which will help make this a clinically translatable tool for breast tumor margin assessment.

Dissertation

Worldwide, head and neck squamous cell cancers (HNSCC) account for over 375,000 deaths annually. The majority of these cancers arise in the outermost squamous cells which progress through a series of precancerous changes before developing into invasive HNSCC. It is widely accepted that prognosis is strongly correlated to the stage of diagnosis, with early detection more than doubling the patient’s chance of survival. Currently, however, 60% of HNSCCs are diagnosed when they have already progressed to stage 3 or stage 4 disease. The current diagnostic method of visual examination often fails to recognize early indicators of HNSCC, thereby missing an important prevention window.

Determination of cancer from non-malignant tissues is dependent on pathological examination of lesion biopsies. Thus, all patients with any clinically suspicious lesions undergo surgical biopsies. Furthermore, these surgical biopsies carry risks. In addition to the risk of general anesthesia for patients undergoing panedoscopy, some patients have poor healing and develop ulcerations or infections as a result of surgical biopsy at any anatomical site. Additionally, studies have shown that approximately 50% of suspected biopsies are later pathologically confirmed normal. An enormous amount of labor, facility, and monetary resources are expended on non-malignant biopsies and patients who ultimately have no malignancy. It would be of immense overall benefit to clinicians and patients to have a non-invasive and portable technique that could rapidly identify those patients that would benefit from further surgical biopsy from those that only need follow-up clinical observations.

Once carcinoma is confirmed in a patient, treatment currently involves modalities of surgery, radiation, and chemotherapy. Radiotherapy plays a significant role, particularly in the management of localized HNSCC, because it is a non-invasive and function-preserving modality. However, the effectiveness of radiotherapy is limited by hypoxia. Previous studies showed that tumors reoxygenated during radiotherapy treatment may have a better prognosis. Despite decades of work, there is still no reliable, cost-effective way for measuring tumor hypoxia over multiple time points to estimate the prognosis.

To address these unmet clinical needs, three aims were proposed. The first aim was to improve early detection by identifying biomarkers of early pre-cancer as well as developing an objective algorithm to detect early disease. Neovasculature is an important biomarker for early cancer diagnosis. Even before the development of a clinically detectable lesion, the tumor vasculature undergoes structural and morphological changes in response to oncogenic signaling pathways [8]. Without receiving a sufficient supply of oxygen and nutrients to proliferate, early tumor growth is limited to only 1-2 mm. High-resolution optical imaging is well suited to characterize the earliest neovascularization changes that accompany neoplasia owing to its sensitivity to hemoglobin absorption and resolution to visualize capillary level architecture. Dark field microscopy is a low-cost and robust method to image the neovasculature. We imaged neovascularization in vivo in a spontaneous hamster oral mucosa carcinogen model using a label-free, reflected-light spectral dark field microscope. Hamsters’ cheek pouches were painted with 7, 12-Dimethylbenz[a]anthracene (DMBA) to induce precancerous to cancerous changes, or mineral oil as control. Spectral dark field images were obtained during carcinogenesis and in control oral mucosa, and quantitative vascular features were computed. Vascular tortuosity increased significantly in oral mucosa diagnosed as hyperplasia, dysplasia and squamous cell carcinoma (SCC) compared to normal. Vascular diameter and area fraction decreased significantly in dysplasia and SCC compared to normal. The areas under the receiver operative characteristic (ROC) curves (AUC) computed using a Support Vector Machine (SVM) were 0.95 and 0.84 for identifying SCC or dysplasia, respectively, vs. normal and hyperplasia oral mucosa combined. To improve AUCs for identifying dysplasia, quantitative vascular features were computed again after the vessels were split into large and small vessels based on diameter. The large vessels preserved the same significant trends, while small vessels demonstrated the opposite trends. Significant increases in diameter and decreases in area fraction were observed in SCC and dysplasia. The AUCs were improved to 0.99 and 0.92 for identifying SCC and dysplasia. These results suggest that dark field vascular imaging is a promising tool for pre-cancer detection.

Optical imaging can also be applied to quantifying other important characteristics of solid tumors in head and neck cancer (HNC), such as hypoxia, abnormal vascularity and cell proliferation. Diffuse reflectance spectroscopy is a simple and robust method to measure tissue oxygenation, vascularity and cell density. It is particularly suitable for applications in the operation room because of its compact design and portability. In addition, a fiber probe-based system is ideal for obtaining measurements at suspicious lesions in the head and neck area during surgery. Thus, my second aim was to reduce the number of unnecessary HNSCC biopsies by developing a robust tool and rapid analysis method appropriate for clinical settings. We propose the use of morphological optical biomarkers for rapid detection of human HNSCC by leveraging the underlying tissue characteristics in the aerodigestive tracts Prior to biopsy, diffuse reflectance spectra were obtained from malignant and contra-lateral non-malignant tissues of 57 patients undergoing panendoscopy. Oxygen saturation (SO2), total hemoglobin concentration ([THb]), and the reduced scattering coefficient were extracted using an inverse Monte Carlo (MC) method previously developed by former student in our lab. Differences in malignant and non-malignant tissues were examined based on two different groupings: by anatomical site and by morphological tissue type. Measurements were acquired from 252 sites, 51 of which were pathologically classified as SCC. Optical biomarkers exhibited statistical differences between malignant and non-malignant samples. Contrast was enhanced when parsing tissues by morphological classification rather than by anatomical subtype for unpaired comparisons. Corresponding linear discriminant models using multiple optical biomarkers showed improved predictive ability when accounting for morphological classification, particularly in node-positive lesions. The false-positive rate was retrospectively found to decrease by 34.2% in morphologically- vs. anatomically-derived predictive models. In glottic tissue, the surgeon exhibited a false-positive rate of 45.7% while the device showed a lower false-positive rate of only 12.4%. Additionally, comparisons of optical parameters were made to further understand the physiology of tumor staging and potential causes of high surgeon false-positive rates. Optical spectroscopy is a user-friendly, non-invasive tool capable of providing quantitative information to discriminate malignant from non-malignant head and neck tissues. Predictive models demonstrated promising results for diagnostics. Furthermore, the strategy described appears to be well suited to reduce the clinical false-positive rate.

To further improve the speed for extracting the tissue oxygenation and [THb] to reduce the time when patients were under anesthesia, the third aim was to develop a rapid heuristic ratiometric analysis for estimating tissue [THb] and SO2 from measured tissue diffuse reflectance spectra. The analysis was validated in tissue-mimicking phantoms and applied to clinical measurements in head and neck, cervical and breast tissues. The analysis works in two steps. First, a linear equation that translates the ratio of the diffuse reflectance spectra at 584 nm to 545 nm to estimate the tissue [THb] using a Monte carlo (MC)-based lookup table was developed. This equation is independent of tissue scattering and oxygen saturation. Second, SO2 was estimated using non-linear logistic equations that translate the ratio of the diffuse reflectance spectra at 539 nm to 545 nm into the tissue SO2. Correlations coefficients of 0.89 (0.86), 0.77 (0.71) and 0.69 (0.43) were obtained for the tissue hemoglobin concentration (oxygen saturation) values extracted using the full spectral MC and the ratiometric analysis, for clinical measurements in head and neck, breast and cervical tissues, respectively. The ratiometric analysis was more than 4000 times faster than the inverse MC analysis for estimating tissue [THb] and SO2 in simulated phantom experiments. In addition, the discriminatory power of the two analyses was similar. These results show the potential of such empirical tools to rapidly estimate tissue hemoglobin and oxygenation for real-time applications.

In addition to its use as a diagnostic marker for various cancers, tissue oxygenation is believed to play a role in the success of cancer therapies, particularly radiotherapy. However, since little effort has been made to develop tools to exploit this relationship, the fourth aim was to estimate patient prognosis by measuring tumor hypoxia over multiple time points so physicians are able to develop more informed and effective clinical treatment plan. To test if oxygenation kinetics correlates with the likelihood for local tumor control following fractionated radiotherapy, we again used diffuse reflectance spectroscopy to noninvasively measure tumor vascular oxygenation and [THb] associated with radiotherapy of 5 daily fractions (7.5, 9 or 13.5 Gy/day) in FaDu xenografts. Spectroscopy measurements were obtained immediately before each daily radiation fraction and during the week after radiotherapy. SO2 and [THb] were computed using an inverse MC model. Oxygenation kinetics during and after radiotherapy, but before a change in tumor volume, was associated with local tumor control. Locally controlled tumors exhibited significantly faster increases in oxygenation after radiotherapy (days 12-15) compared with tumors that recurred locally. (2) Within the group of tumors that recurred, faster increases in oxygenation during radiotherapy (days 3-5) were correlated with earlier recurrence times. An AUC of 0.74 was achieved when classifying the local control tumors from all irradiated tumors using the oxygen kinetics with a logistic regression model. (3) The rate of increase in oxygenation was radiation dose dependent. Radiation doses ≤9.5 Gy/day did not initiate an increase in oxygenation whereas 13.5 Gy/day triggered significant increases in oxygenation during and after radiotherapy. Additional confirmation is required in other tumor models, but these results suggest that monitoring tumor oxygenation kinetics could aid in the prediction of local tumor control after radiotherapy.

Angiogenesis is a highly regulated process to support tissue growth. Neovasculature is designed by nature to grow toward areas lacking nutrition and oxygen. Cancer cells proliferate too quickly to have their nutritional and oxygen needs completely satisfied, which results in an imbalanced state of angiogenesis leading to tortuous blood vessels, hypoxic tissues and radioresistance. We characterized the tumor-induced vascular features with simple, robust and low-cost dark field microscopy and spectroscopy to enable early cancer diagnosis, improvement of surgical biopsy accuracy and better predict the prognosis of radiotherapy for HNC. Our results demonstrated that these noninvasively measured, label-free vascular features are able to detect pre-cancer, reduce unnecessary surgical biopsies and predict prognosis of radiotherapy.

Dissertation

Unmanned aerial vehicles (UAV) have revolutionized data collection in large scale agronomic field trials (10+ ha). Vegetative index (VI) maps derived from UAV imagery are a potential tool to characterize temporal and spatial treatment effects in a more efficient and non-destructive way compared to traditional data collection methods that require manual sampling. The overall objective of this study was to characterize and quantify maize responses to experimental treatments in field-scale research using UAV imagery. The specific objectives were: 1) to assess the performance of several VI as predictors of grain yield and to evaluate their ability to distinguish between fertilizer treatments, and the effects of removing soil and shadow background, 2) to assess the performance of VI and canopy cover fraction (CCF) as predictors of maize biomass at vegetative and reproductive growth stages under field-scale conditions, and 3) to compare the performance of VI derived from consumer-grade and multispectral sensors for predicting grain yield and identifying treatment effects. For the first objective, the results suggest that most VI were good indicators of grain yield at late vegetative and early reproductive growth stages, and that removing soil background improved the characterization of maize responses to experimental treatments. For objective two, overall, CCF was the best to predict biomass at early vegetative growth stages, while VI at reproductive growth stages. Finally, for objective three, performance of consumer-grade and multispectral derived VI were similar for predicting grain yield and identifying treatment effects.

A Dissertation submitted to the Faculty of Science, University of the Witwatersrand, in fulfilment of the requirements for the degree of Master of Science School of Computer Science and Applied Mathematics, March 2017
Mineral exploration is expensive, logistically challenging and can be detrimental to the environment. In addition to the physical disturbed of geological sampling, artisanal miners, charcoal burners and poachers follow in the wake of geological exploration teams, resulting in severe environmental degradation. The remote sensing of geological features is used in conjunction with geophysics to help refine the amount of ground based sampling where the surface geology is exposed (e.g. deserts, barren surfaces and rocky outcrops). However, it is not feasible to use these geological remote sensing techniques the earth’s surface is covered with vegetation. Studies have shown that plants respond to mineral nutrients or conversely toxicities in their growing environment, including metal concentrations in the soil, either through the presence or absence of particular species, or by exhibiting physiological or phenological changes in response to depleted or elevated substrate metal concentrations. The use of plant species composition and foliar elemental contents (methods known collectively as phytogeochemical exploration) have been successfully used to detect ore-bodies. Visible changes in leaf structure and chemical composition as a result of deficiencies in elemental nutrition or toxicities have been well-researched from botanical and soil science aspects, and are widely used for agronomic applications, but have yet to be exploited for mineral exploration. This study assessed the feasibility of using remotely-sensed spectral reflectance signatures of tree foliage to detect changes in substrate elemental concentrations across three geologies on the Witwatersrand Basin. The study comprises of an outcropping metal-rich ore body, the Black Reef (quartzite), flanked by dolomite to the South East and Ventersdorp Lavas to the North West. The soils of these three parent geologies can be expected to exhibit differences in plant nutrient availability, as well as deficiencies or toxicities. Each geology on the study site was characterised and classified into landscape functional types to account for aspect, position on the catena and soils characteristics, all of which could mask, conflict or auto correlate with any observed changes in vegetation stress spectral signatures associated with the changing geology. Three tree species with continuous across the study site were selected: Searsia lancea (L.f.) Moffet (previously Rhus lancea), Euclea crispa (Thunb.) Guerke var crispa and Acacia karroo Hayne. The study determined how the foliar and substrate elemental concentrations and uptake ratios differed between the three tree species, the three geologies and the landscape functional types. The study then related plant spectral response of three tree species to geology, landscape function type and to the foliar and substrate elemental content. Soil elemental concentrations were analysed and it was found that the three parent geologies could be classified by their relative concentrations of Mn, Cr, Ti, Cu Cr, Pb, Ba, Fe, and Zr in the soils. The findings revealed that the plants showed changes in physiological status associated with geology which were detectable through the use of vegetation indices. The study made use of eight different vegetation indices (NDVI, NDWI, PSRI, Red-edge NDVI, red-edge position, red-edge inflection point, and the 725/702 ratio of the first and second derivative), derived from handheld hyperspectral data. The three species differed in their spectral response to the changes in geology and in their stress response to elevated metal content on the Black Reef (p < 0.05). Regression (linear and non-parametric) was used to identify which foliar and substrate elemental concentrations most affected spectral response. The A. karroo samples were found to be most affected by Mn, Ti, Fe and Sr. The S. lancea samples were found to be most affected by As, Cu, Pb and Sn and the E. crispa response was found to be most affected by Cu, Mn, Na, Ni, Rb, Zn, and Zr (p < 0.01). In order to identify the changes in geology, it was found to be necessary to first classify the spectral response of the three species, and then detect spectral variations within each species class, as the species-specific spectral responses to changes in geology were significantly different (p< 0.05). The study successfully classified the three tree species according to their spectral response through the combined use of the eight vegetation indices. However, it was found that a subset of the samples which had either much higher or much lower elemental concentrations in the leaves and soils than the remaining samples for that species, showed a plant stress response which affected the spectral response of the plants sufficiently to result in an incorrect species classification. In conclusion, the finding of this study showed that VIs can be used to detect differences in spectral response between trees growing on different geologies. It was found that the combination of vegetation indices can be used to determine a “typical” spectral response per species, but that where the growing conditions were particularly stressful, the stress response could alter the plant spectral response sufficiently to result in a misclassification of the sample by species. Further work is required to validate this observation, and to investigate how more sophisticated spectral analysis could be used to distinguish between taxonomic and substrate induced spectral variation, before it would be possible to scale this work up to a canopy-scale remote sensing tool.
XL2018

The measurement of plant water content is essential to assess stress and disturbance in forest plantations. Traditional techniques to assess plant water content are costly, time consuming and spatially restrictive. Remote sensing techniques offer the alternative of a non destructive and instantaneous method of assessing plant water content over large spatial scales where ground measurements would be impossible on a regular basis. The aim of this research was to assess the relationship between plant water content and reflectance data in Eucalyptus grandis forest stands in KwaZulu-Natal, South Africa. Field reflectance and first derivative reflectance data were correlated with plant water content. The first derivative reflectance performed better than the field reflectance data in estimating plant water content with high correlations in the visible and mid-infrared portions of the electromagnetic spectrum. Several reflectance indices were also tested to evaluate their effectiveness in estimating plant water content and were compared to the red edge position. The red edge position calculated from the first derivative reflectance and from the linear four-point interpolation method performed better than all the water indices tested. It was therefore concluded that the red edge position can be used in association with other water indices as a stable spectral parameter to estimate plant water content on hyperspectral data. The South African satellite SumbandilaSat is due for launch in the near future and it is essential to test the utility of this satellite in estimating plant water content, a study which has not been done before. The field reflectance data from this study was resampled to the SumbandilaSat band settings and was put into a neural network to test its potential in estimating plant water content. The integrated approach involving neural networks and the resampled field spectral data successfully predicted plant water content with a correlation coefficient of 0.74 and a root mean square error (RMSE) of 1.41 on an independent test dataset outperforming the traditional multiple regression method of estimation. The potential of the SumbandilaSat wavebands to estimate plant water content was tested using a sensitivity analysis. The results from the sensitivity analysis indicated that the xanthophyll, blue and near infrared wavebands are the three most important wavebands used by the neural network in estimating plant water content. It was therefore concluded that these three bands of the SumbandilaSat are essential for plant water estimation. In general this study showed the potential of up-scaling field spectral data to the SumbandilaSat, the second South African satellite scheduled for launch in the near future.
Thesis (M.Sc.) - University of KwaZulu-Natal, Pietermaritzburg, 2008.

Hyperspectral imaging has become one of the most popular technologies in plant phenotyping because it can efficiently and accurately predict numerous plant physiological features such as plant biomass, leaf moisture content, and chlorophyll content. Various hyperspectral imaging systems have been deployed in both greenhouse and field phenotyping activities. However, the hyperspectral imaging quality is severely affected by the continuously changing environmental conditions such as cloud cover, temperature and wind speed that induce noise in plant spectral data. Eliminating these environmental effects to improve imaging quality is critically important. In this thesis, two approaches were taken to address the imaging noise issue in greenhouse and field separately. First, a computational simulation model was built to simulate the greenhouse microclimate changes (such as the temperature and radiation distributions) through a 24-hour cycle in a research greenhouse. The simulated results were used to optimize the movement of an automated conveyor in the greenhouse: the plants were shuffled with the conveyor system with optimized frequency and distance to provide uniform growing conditions such as temperature and lighting intensity for each individual plant. The results showed the variance of the plants’ phenotyping feature measurements decreased significantly (i.e., by up to 83% in plant canopy size) in this conveyor greenhouse. Secondly, the environmental effects (i.e., sun radiation) on aerial hyperspectral images in field plant phenotyping were investigated and modeled. An artificial neural network (ANN) method was proposed to model the relationship between the image variation and environmental changes. Before the 2019 field test, a gantry system was designed and constructed to repeatedly collect time-series hyperspectral images with 2.5 minutes intervals of the corn plants under varying environmental conditions, which included sun radiation, solar zenith angle, diurnal time, humidity, temperature and wind speed. Over 8,000 hyperspectral images of corn (Zea mays L.) were collected with synchronized environmental data throughout the 2019 growing season. The models trained with the proposed ANN method were able to accurately predict the variations in imaging results (i.e., 82.3% for NDVI) caused by the changing environments. Thus, the ANN method can be used by remote sensing professionals to adjust or correct raw imaging data for changing environments to improve plant characterization.

Dominant vegetation species of two structurally and functionally different montane ecosystems were studied by means of laboratory and field spectroscopy and remote sensing image data: (1) a homogeneous human-influenced evergreen coniferous forest represented by a Norway spruce forest in the Krušné hory Mountains and (2) a heterogeneous natural ecosystem of a relict arctic-alpine tundra in the Krkonoše Mountains with predominance of grasses. The first part dealing with the Norway spruce forest is especially focused on the methods of laboratory spectroscopy. The assessment of Norway spruce stands on a regional and a global scales requires detailed knowledge of their spectral properties at the level of needles and shoots in the beginning, but ground research is very time-demanding. Open spectral libraries could help to get more ground-truth data for subsequent analysis of tree species in forests ecosystems. However, the problem may arise with the comparability of spectra taken by different devices. The present thesis focuses on a comparability of spectra measured by a field spectroradiometer coupled with plant contact probe and/or two integrating spheres (Paper 3) and proves the significant differences in spruce needle spectra measured by the contact probe and integrating sphere, spectra of...