The purpose of this thesis is to automate the online detection of weaving defects by a computerized system based on image processing software. Obviously, fabric inspection has an importance to prevent risk of delivering inferior quality product. Until recently, the visual defect detection is still under taken offline and manually by humans with many drawbacks such as tiredness, boredom, and, inattentiveness. Fortunately, the continuous development in computer technology introduces the online automated fabric inspection as an effective alternative. Because the defect-free fabric has a periodic regular structure, the occurrence of a defect in the fabric breaks the regular structure. Therefore, the fabric defects can be detected by monitoring fabric structure. In our work, Fast Fourier Transform and Cross-correlation techniques, i.e. linear operations, are first implemented to examine the structure regularity features of the fabric image in the frequency domain. To improve the efficiency of the technique and overcome the problem of detection errors, further thresholding operation is implemented using a level selection filter. Through this filter, the technique is able to detect only the actual or real defects and highlight their exact dimensions. A software package such as Matlab or Scilab is used for this procedure. It is implemented firstly on a simulated plain fabric to determine the most important parameters during the process of defect detection and then to optimize each of them even considering noise. To verify the success of the technique, it is implemented on real plain fabric samples with different colours containing various defects. Finally, a vision-based fabric inspection prototype that could be accomplished on-loom to inspect the fabric under construction with 100% coverage is proposed.

The S shaped log log drawdown time curve typical of pumping tests in unconfined aquifers is reinvestigated via numerical experiments. We examine the temporal and spatial evolution of the rate of change in storage in an unconfined aquifer during pumping. This evolution is related to the transition of water release mechanisms from the expansion of water and compaction of the porous medium to the drainage of water from the unsaturated zone above the initial water table and initially saturated pores as the water table falls during the pumping of the aquifer. We conclude that the transition of the water release mechanisms and vertical flow are the cause of the S shaped drawdown time. Cross-correlation analysis is then employed to examine the relationship between hydraulic properties of an unconfined aquifer and pressure observations. The analysis reveals that head observed in the saturated zone at late times along a streamline is positively correlated with the conductivity (K(s)) of the region upstream of the observation location, and negatively correlated with the K(s) of the region downstream of the observation location along the same streamline. Besides, head observations in the saturated zone at the early time are positively correlated with specific storage (S(s)) in a narrow region between the observation and pumping locations. At intermediate and late times, the head positively correlates with the heterogeneity of α (pore-size distribution parameter) in a thin disk-shaped unsaturated region above the pumping and observation locations. Saturated water content θ(s) in the vadose zone directly above the pumping and monitoring locations is found positively correlated with the head observations during the intermediate times and late times.In the end, a stochastic inverse estimation is conducted to jointly interpret a sequential pumping test in a three dimensional unconfined aquifer. K(s), S(s), θ(s) and α are estimated at the same time. The estimated results capture the pattern of the heterogeneous parameters as well as the details with a smooth distribution. The estimated heterogeneous parameter fields produce better head predictions than the traditional homogeneous method.

People use orientation of their faces to convey rich, inter-personal information. For example, a person will direct his face to indicate who the intended target of the conversation is. Similarly in a conversation, face orientation is a non-verbal cue to listener when to switch role and start speaking, and a nod indicates that a person has understands, or agrees with, what is being said. Further more, face pose estimation plays an important role in human-computer interaction, virtual reality applications, human behaviour analysis, pose-independent face recognition, driver s vigilance assessment, gaze estimation, etc. Robust face recognition has been a focus of research in computer vision community for more than two decades. Although substantial research has been done and numerous methods have been proposed for face recognition, there remain challenges in this field. One of these is face recognition under varying poses and that is why face pose estimation is still an important research area. In computer vision, face pose estimation is the process of inferring the face orientation from digital imagery. It requires a serious of image processing steps to transform a pixel-based representation of a human face into a high-level concept of direction. An ideal face pose estimator should be invariant to a variety of image-changing factors such as camera distortion, lighting condition, skin colour, projective geometry, facial hairs, facial expressions, presence of accessories like glasses and hats, etc. Face pose estimation has been a focus of research for about two decades and numerous research contributions have been presented in this field. Face pose estimation techniques in literature have still some shortcomings and limitations in terms of accuracy, applicability to monocular images, being autonomous, identity and lighting variations, image resolution variations, range of face motion, computational expense, presence of facial hairs, presence of accessories like glasses and hats, etc. These shortcomings of existing face pose estimation techniques motivated the research work presented in this thesis. The main focus of this research is to design and develop novel face pose estimation algorithms that improve automatic face pose estimation in terms of processing time, computational expense, and invariance to different conditions.

The objective of this study was to investigate the spatial relationship between the most likely distribution of saturated hydraulic conductivity (K-s) and the observed pressure head (P) distribution within a hillslope. The cross-correlation analysis method was used to investigate the effects of the variance of lnK(s), spatial structure anisotropy of lnK(s), and vertical infiltration flux (q) on P at some selected locations within the hillslope. The cross-correlation analysis shows that, in the unsaturated region with a uniform flux boundary, the dominant correlation between P and Ks is negative and mainly occurs around the observation location of P. A relatively high P value is located in a relatively low Ks zone, while a relatively low P value is located in a relatively high Ks zone. Generally speaking, P is positively correlated with q/Ks at the same location in the unsaturated region. In the saturated region, the spatial distribution of K-s can significantly affect the position and shape of the phreatic surface. We therefore conclude that heterogeneity can cause some parts of the hillslope to be sensitive to external hydraulic stimuli (e.g., rainfall and reservoir level change), and other parts of the hillslope to be insensitive. This is crucial to explaining why slopes with similar geometries would show different responses to the same hydraulic stimuli, which is significant to hillslope stability analysis. (C) 2016 Hohai University. Production and hosting by Elsevier B.V.

Doctor of Philosophy / Department of Biological and Agricultural Engineering / Naiqian Zhang / An optical sensor for determining water velocity in natural open channels like creeks and rivers has been designed and tested. The sensor consists of a plastic body which is shaped so that water flows through a U-shaped channel into which are mounted LEDs and matching phototransistors at various angles. A small amount of dye is injected into the water just upstream of two sets of LEDs and phototransistors which are spaced 4 cm apart. The time delay between the dye’s effects on these signals depends on water velocity and is determined using a biased cross correlation calculation. In addition to providing velocity, the LEDs and phototransistors can also be used to estimate soil sediment concentration. A previous version of the sensor was tested in enclosed flow to confirm that the general design of the sensor, including LEDs, phototransistors, dye and electronics, would indeed work to detect the velocity of water flowing through the sensor. Although the conditions for the test were unlike those experienced in natural open channels, the ability to catch all the fluid flowing through the sensor provided a simple confirmation of the velocity estimate that was not available in field settings. Further testing in the field then confirmed that the sensor worked in the field but also identified several areas needing improvement. Computational fluid dynamics was used to improve the sensor body. The electronics and program running the sensor were also redesigned. After making these improvements, a new version of the sensor was produced. The testing of the new version of the sensor confirmed its ability to accurately detect velocity in natural open channels. The velocity measurements from this sensor were compared to the commercially available Flowtracker velocity sensor. A regression analysis on the measurements from the two sensors found that the velocity measurements from each sensor were nearly identical across a range of velocities. Other tests established that the electronics and programming running the sensor performed as designed. The development and testing of this sensor has resulted in a system which works in natural open channels like creeks and rivers.

Weak gravitational lensing has developed to be one of the most powerful tools for studying the (dark) matter distribution in the Universe. Most weak lensing studies thus far were con- ducted in the optical and near infrared. Measuring weak lensing in the radio though, provided it is feasible, can be very advantageous. One can exploit the well known and deterministic beam pattern of a radio telescope and the polarisation information in radio data to reduce shape biases and intrinsic alignment effects respectively. Combining the information from an optical and a radio survey can also help remove systematics from both datasets. This has motivated this study that uses archival radio and optical data to treat telescope systematics and measure an unbiased weak lensing signal using shape information derived from radio observations. Using simulations I have shown that an unbiased convergence cross power spectrum can be measured in the presence of the large scale (θ > 1◦) systematics detected in FIRST and SDSS. The method however amplifies the uncertainties by a factor ∼2.5 compared to the errors due to cosmic variance and noise due to galaxy intrinsic shape alone. Using the shape information from the two surveys I measure a Ckappakappa spectrum signal that is inconsistent with zero at the 2.7sigma. The placed constraints are consistent with the expected signal in the concordance cosmological model assuming recent estimates of the cosmological parameters from the Planck satellite and literature values for the median redshifts of SDSS and FIRST.Through simulations I also show that I can successfully remove position based small scale systematics (θ5). Using the deconvolved information for the resolved sources I calculate a FWHM median size and flux density of 0.5'' and 300μJy respectively. Comparing the source number density and RMS noise of the study with those of FIRST, I extrapolate to predict that the number density of sources at > 5sigma will be ∼5arcmin-2, assuming the target noise threshold for the survey is reached.

This dissertation is composed of two parts, an integrative essay and a set of published papers. The essay and the collection of papers are placed in the context of development and application of time series econometric models in a temporal-axis from 1970s through 2005, with particular focus in the Marketing discipline. The main aim of the integrative essay is on modelling the effects of marketing actions on performance variables, such as sales and market share in competitive markets. Such research required the estimation of two kinds of time series econometric models: multivariate and multiple time series models. I use Autoregressive Integrated Moving Average (ARIMA) intervention models and the Pierce and Haugh statistical test to model the impact of a single marketing instrument, mainly price promotions, to measure own and cross-short term sales effects, and to study asymmetric marketing competition. I develop and apply Vector AutoRegressive (VAR) and Bayesian Vector AutoRegressive (BVAR) models to estimate dynamic relationships in the market and to forecast market share. Especially, BVAR models are advantageous because they contain all relevant dynamic and interactive effects. They accommodate not only classical competitive reaction effects, but also own and cross-market share brand feedback effects and internal decision rules and provided substantively useful insights into the dynamics of demand. The integrative essay is structured in four main parts. The introduction sets the basic ideas behind the published papers, with particular focus on the motivation of the essay, the types of competitive reaction effects analysed, an overview of the time series econometric models in marketing, a short discussion of the basic methodology used in the research and a brief description of the inter-relationships across the published papers and structure of the essay. The discussion is centred on how to model the effects of marketing actions at the selective demand or brand level and at the primary demand or product level. At the brand level I discuss the research contribution of my work on (i) modelling promotional short-term effects of price and non-price actions on sales and market share for consumer packaged goods, with no competition, (ii) how to measure own and cross short-term sales effects of advertising and price, in particular, cross-lead and lag effects, asymmetric sales behaviour and competition without retaliatory actions, in an automobile market, (iii) how to model the marketing-mix effectiveness at the short and long-term on market shares in a car market, (iv) what is the best method to forecast market share, and (v) the study of causal linkages at different time horizons between sales and marketing activity for a particular brand. At the product or commodity level, I propose a way to model the flows of tourists that come from different origins (countries) to the same country-destination as market segments defining the primary demand of a commodity - the product

<p>an Internal Navigational System and a Global Navigational Satellite System (GNSS). In navigational warfare the GNSS can be jammed, therefore are a third navigational system is needed. The system that has been tried in this thesis is camera based navigation. Through a video camera and a sensor reference the position is determined. This thesis will process the matching between the sensor reference and the video image.</p><p>Two methods have been implemented: normalized cross correlation and position determination through a homography. Normalized cross correlation creates a correlation matrix. The other method uses point correspondences between the images to determine a homography between the images. And through the homography obtain a position. The more point correspondences the better the position determination will be.</p><p>The results have been quite good. The methods have got the right position when the Euler angles of the UAV have been known. Normalized cross correlation has been the best method of the tested methods.</p>

Theoretical and experimental studies of real-time interrogation of fiber Bragg grating (FBG) sensors based on chirped pulse compression with increased interrogation resolution and signal-to-noise ratio are presented. Two interrogation systems are proposed in this thesis. In the first interrogation system, a linearly chirped FBG (LCFBG) is employed as the sensing element. By incorporating the LCFBG in an optical interferometer as the sensor encoding system, employing wavelength-to-time mapping and chirped pulse compression technique, the correlation of output microwave waveform with a chirped reference waveform would provide an interrogation result with high speed and high resolution. The proposed system can provide an interrogation resolution as high as 0.25 μ at a speed of 48.6 MHz. The second interrogation system is designed to achieve simultaneous measurement of strain and temperature. In this system, a high-birefringence LCFBG (Hi-Bi LCFBG) is employed as a sensing element.

Theoretical and experimental studies of real-time interrogation of fiber Bragg grating (FBG) sensors based on chirped pulse compression with increased interrogation resolution and signal-to-noise ratio are presented. Two interrogation systems are proposed in this thesis. In the first interrogation system, a linearly chirped FBG (LCFBG) is employed as the sensing element. By incorporating the LCFBG in an optical interferometer as the sensor encoding system, employing wavelength-to-time mapping and chirped pulse compression technique, the correlation of output microwave waveform with a chirped reference waveform would provide an interrogation result with high speed and high resolution. The proposed system can provide an interrogation resolution as high as 0.25 μ at a speed of 48.6 MHz. The second interrogation system is designed to achieve simultaneous measurement of strain and temperature. In this system, a high-birefringence LCFBG (Hi-Bi LCFBG) is employed as a sensing element.