A Neural Network Architecture for Figure-ground Separation of Connected Scenic Figures

A neural network model, called an FBF network, is proposed for automatic parallel separation of multiple image figures from each other and their backgrounds in noisy grayscale or multi-colored images. The figures can then be processed in parallel by an array of self-organizing Adaptive Resonance Theory (ART) neural networks for automatic target recognition. An FBF network can automatically separate the disconnected but interleaved spirals that Minsky and Papert introduced in their book Perceptrons. The network's design also clarifies why humans cannot rapidly separate interleaved spirals, yet can rapidly detect conjunctions of disparity and color, or of disparity and motion, that distinguish target figures from surrounding distractors. Figure-ground separation is accomplished by iterating operations of a Feature Contour System (FCS) and a Boundary Contour System (BCS) in the order FCS-BCS-FCS, hence the term FBF, that have been derived from an analysis of biological vision. The FCS operations include the use of nonlinear shunting networks to compensate for variable illumination and nonlinear diffusion networks to control filling-in. A key new feature of an FBF network is the use of filling-in for figure-ground separation. The BCS operations include oriented filters joined to competitive and cooperative interactions designed to detect, regularize, and complete boundaries in up to 50 percent noise, while suppressing the noise. A modified CORT-X filter is described which uses both on-cells and off-cells to generate a boundary segmentation from a noisy image.

Evanescent field sensing: optical nanofibres, whispering-gallery-mode microspherical and microbubble resonators

In this thesis, the evanescent field sensing techniques of tapered optical nanofibres and microspherical resonators are investigated. This includes evanescent field spectroscopy of a silica nanofibre in a rubidium vapour; thermo-optical tuning of Er:Yb co-doped phosphate glass microspheres; optomechanical properties of microspherical pendulums; and the fabrication and characterisation of borosilicate microbubble resonators. Doppler-broadened and sub-Doppler absorption spectroscopic techniques are performed around the D2 transition (780.24 nm) of rubidium using the evanescent field produced at the waist of a tapered nanofibre with input probe powers as low as 55 nW. Doppler-broadened Zeeman shifts and a preliminary dichroic atomic vapour laser lock (DAVLL) line shape are also observed via the nanofibre waist with an applied magnetic field of 60 G. This device has the potential for laser frequency stabilisation while also studying the effects of atom-surface interactions. A non-invasive thermo-optical tuning technique of Er:Yb co-doped microspheres to specific arbitrary wavelengths is demonstrated particularly to 1294 nm and the 5S1/2F=3 to 5P3/2Fʹ=4 laser cooling transition of 85Rb. Reversible tuning ranges of up to 474 GHz and on resonance cavity timescales on the order of 100 s are reported. This procedure has prospective applications for sensing a variety of atomic or molecular species in a cavity quantum electrodynamics (QED) experiments. The mechanical characteristics of a silica microsphere pendulum with a relatively low spring constant of 10-4 Nm-1 are explored. A novel method of frequency sweeping the motion of the pendulum to determine its natural resonance frequencies while overriding its sensitivity to environmental noise is proposed. An estimated force of 0.25 N is required to actuate the pendulum by a displacement of (1-2) μm. It is suggested that this is of sufficient magnitude to be experienced between two evanescently coupled microspheres (photonic molecule) and enable spatial trapping of the micropendulum. Finally, single-input borosilicate microbubble resonators with diameters <100 μm are fabricated using a CO2 laser. Optical whispering gallery mode spectra are observed via evanescent coupling with a tapered fibre. A red-shift of (4-22) GHz of the resonance modes is detected when the hollow cavity was filled with nano-filtered water. A polarisation conversion effect, with an efficiency of 10%, is observed when the diameter of the coupling tapered fibre waist is varied. This effect is also achieved by simply varying the polarisation of the input light in the tapered fibre where the efficiency is optimised to 92%. Thus, the microbubble device acts as a reversible band-pass to band-stop optical filter for cavity-QED, integrated solid-state and semiconductor circuit applications.

Interaction of additive and quantization noises in digital PLLs

Phase-locked loops (PLLs) are a crucial component in modern communications systems. Comprising of a phase-detector, linear filter, and controllable oscillator, they are widely used in radio receivers to retrieve the information content from remote signals. As such, they are capable of signal demodulation, phase and carrier recovery, frequency synthesis, and clock synchronization. Continuous-time PLLs are a mature area of study, and have been covered in the literature since the early classical work by Viterbi [1] in the 1950s. With the rise of computing in recent decades, discrete-time digital PLLs (DPLLs) are a more recent discipline; most of the literature published dates from the 1990s onwards. Gardner [2] is a pioneer in this area. It is our aim in this work to address the difficulties encountered by Gardner [3] in his investigation of the DPLL output phase-jitter where additive noise to the input signal is combined with frequency quantization in the local oscillator. The model we use in our novel analysis of the system is also applicable to another of the cases looked at by Gardner, that is the DPLL with a delay element integrated in the loop. This gives us the opportunity to look at this system in more detail, our analysis providing some unique insights into the variance `dip' seen by Gardner in [3]. We initially provide background on the probability theory and stochastic processes. These branches of mathematics are the basis for the study of noisy analogue and digital PLLs. We give an overview of the classical analogue PLL theory as well as the background on both the digital PLL and circle map, referencing the model proposed by Teplinsky et al. [4, 5]. For our novel work, the case of the combined frequency quantization and noisy input from [3] is investigated first numerically, and then analytically as a Markov chain via its Chapman-Kolmogorov equation. The resulting delay equation for the steady-state jitter distribution is treated using two separate asymptotic analyses to obtain approximate solutions. It is shown how the variance obtained in each case matches well to the numerical results. Other properties of the output jitter, such as the mean, are also investigated. In this way, we arrive at a more complete understanding of the interaction between quantization and input noise in the first order DPLL than is possible using simulation alone. We also do an asymptotic analysis of a particular case of the noisy first-order DPLL with delay, previously investigated by Gardner [3]. We show a unique feature of the simulation results, namely the variance `dip' seen for certain levels of input noise, is explained by this analysis. Finally, we look at the second-order DPLL with additive noise, using numerical simulations to see the effects of low levels of noise on the limit cycles. We show how these effects are similar to those seen in the noise-free loop with non-zero initial conditions.

Design of digital differentiators

A digital differentiator simply involves the derivation of an input signal. This work includes the presentation of first-degree and second-degree differentiators, which are designed as both infinite-impulse-response (IIR) filters and finite-impulse-response (FIR) filters. The proposed differentiators have low-pass magnitude response characteristics, thereby rejecting noise frequencies higher than the cut-off frequency. Both steady-state frequency-domain characteristics and Time-domain analyses are given for the proposed differentiators. It is shown that the proposed differentiators perform well when compared to previously proposed filters. When considering the time-domain characteristics of the differentiators, the processing of quantized signals proved especially enlightening, in terms of the filtering effects of the proposed differentiators. The coefficients of the proposed differentiators are obtained using an optimization algorithm, while the optimization objectives include magnitude and phase response. The low-pass characteristic of the proposed differentiators is achieved by minimizing the filter variance. The low-pass differentiators designed show the steep roll-off, as well as having highly accurate magnitude response in the pass-band. While having a history of over three hundred years, the design of fractional differentiator has become a ‘hot topic’ in recent decades. One challenging problem in this area is that there are many different definitions to describe the fractional model, such as the Riemann-Liouville and Caputo definitions. Through use of a feedback structure, based on the Riemann-Liouville definition. It is shown that the performance of the fractional differentiator can be improved in both the frequency-domain and time-domain. Two applications based on the proposed differentiators are described in the thesis. Specifically, the first of these involves the application of second degree differentiators in the estimation of the frequency components of a power system. The second example concerns for an image processing, edge detection application.

Algorithms for indoor localization based on IEEE 802.15.4-2011 UWB and inertial sensors

In this thesis, extensive experiments are firstly conducted to characterize the performance of using the emerging IEEE 802.15.4-2011 ultra wideband (UWB) for indoor localization, and the results demonstrate the accuracy and precision of using time of arrival measurements for ranging applications. A multipath propagation controlling technique is synthesized which considers the relationship between transmit power, transmission range and signal-to-noise ratio. The methodology includes a novel bilateral transmitter output power control algorithm which is demonstrated to be able to stabilize the multipath channel, and enable sub 5cm instant ranging accuracy in line of sight conditions. A fully-coupled architecture is proposed for the localization system using a combination of IEEE 802.15.4-2011 UWB and inertial sensors. This architecture not only implements the position estimation of the object by fusing the UWB and inertial measurements, but enables the nodes in the localization network to mutually share positional and other useful information via the UWB channel. The hybrid system has been demonstrated to be capable of simultaneous local-positioning and remote-tracking of the mobile object. Three fusion algorithms for relative position estimation are proposed, including internal navigation system (INS), INS with UWB ranging correction, and orientation plus ranging. Experimental results show that the INS with UWB correction algorithm achieves an average position accuracy of 0.1883m, and gets 83% and 62% improvements on the accuracy of the INS (1.0994m) and the existing extended Kalman filter tracking algorithm (0.5m), respectively.

Sequential anomaly detection in the presence of noise and limited feedback

This paper describes a methodology for detecting anomalies from sequentially observed and potentially noisy data. The proposed approach consists of two main elements: 1) filtering, or assigning a belief or likelihood to each successive measurement based upon our ability to predict it from previous noisy observations and 2) hedging, or flagging potential anomalies by comparing the current belief against a time-varying and data-adaptive threshold. The threshold is adjusted based on the available feedback from an end user. Our algorithms, which combine universal prediction with recent work on online convex programming, do not require computing posterior distributions given all current observations and involve simple primal-dual parameter updates. At the heart of the proposed approach lie exponential-family models which can be used in a wide variety of contexts and applications, and which yield methods that achieve sublinear per-round regret against both static and slowly varying product distributions with marginals drawn from the same exponential family. Moreover, the regret against static distributions coincides with the minimax value of the corresponding online strongly convex game. We also prove bounds on the number of mistakes made during the hedging step relative to the best offline choice of the threshold with access to all estimated beliefs and feedback signals. We validate the theory on synthetic data drawn from a time-varying distribution over binary vectors of high dimensionality, as well as on the Enron email dataset. © 1963-2012 IEEE.

High-fidelity, broadband stimulated-Brillouin-scattering-based slow light using fast noise modulation.

We demonstrate a 5-GHz-broadband tunable slow-light device based on stimulated Brillouin scattering in a standard highly-nonlinear optical fiber pumped by a noise-current-modulated laser beam. The noisemodulation waveform uses an optimized pseudo-random distribution of the laser drive voltage to obtain an optimal flat-topped gain profile, which minimizes the pulse distortion and maximizes pulse delay for a given pump power. In comparison with a previous slow-modulation method, eye-diagram and signal-to-noise ratio (SNR) analysis show that this broadband slow-light technique significantly increases the fidelity of a delayed data sequence, while maintaining the delay performance. A fractional delay of 0.81 with a SNR of 5.2 is achieved at the pump power of 350 mW using a 2-km-long highly nonlinear fiber with the fast noise-modulation method, demonstrating a 50% increase in eye-opening and a 36% increase in SNR in the comparison.

Invariant measure selection by noise. An example

We consider a deterministic system with two conserved quantities and infinity many invariant measures. However the systems possess a unique invariant measure when enough stochastic forcing and balancing dissipation are added. We then show that as the forcing and dissipation are removed a unique limit of the deterministic system is selected. The exact structure of the limiting measure depends on the specifics of the stochastic forcing.

Comparing the model-simulated global warming signal to observations using empirical estimates of unforced noise.

The comparison of observed global mean surface air temperature (GMT) change to the mean change simulated by climate models has received much public and scientific attention. For a given global warming signal produced by a climate model ensemble, there exists an envelope of GMT values representing the range of possible unforced states of the climate system (the Envelope of Unforced Noise; EUN). Typically, the EUN is derived from climate models themselves, but climate models might not accurately simulate the correct characteristics of unforced GMT variability. Here, we simulate a new, empirical, EUN that is based on instrumental and reconstructed surface temperature records. We compare the forced GMT signal produced by climate models to observations while noting the range of GMT values provided by the empirical EUN. We find that the empirical EUN is wide enough so that the interdecadal variability in the rate of global warming over the 20(th) century does not necessarily require corresponding variability in the rate-of-increase of the forced signal. The empirical EUN also indicates that the reduced GMT warming over the past decade or so is still consistent with a middle emission scenario's forced signal, but is likely inconsistent with the steepest emission scenario's forced signal.

Ambient Noise in the Kitimat Fjord System

Sound is an important medium for communication and marine organisms have evolved to capitalize on the efficiency with which sound energy travels through water. Anthropogenic and natural sound sources contribute to ocean ambient noise, which can interfere with the use of this sensory modality by marine animals. Anthropogenic noise sources have been increasing steadily over recent decades largely due to coastal population growth, increased global transportation, and offshore industrialization. Understanding the potential impacts of anthropogenic noise requires the establishment of ambient acoustic baselines from which to measure change. Establishing baselines, especially in quiet areas still largely unaffected by anthropogenic stressors, is particularly crucial in the face of the expansion of offshore industries, increasing coastal population and growing reliance on the ocean for global transportation. Global demand for liquid natural gas (LNG), catalyzed primarily by a growing Asian market, is expected to increase significantly in the next 20 years. The geographic position of British Columbia relative to these markets, a growing supply of LNG and new technology for extraction and shipping situate British Columbia as a strong competitor in the lucrative market. The LNG industry could have many adverse impacts on these territories and ecosystems. The Kitimat Fjord System is slated for the development of these LNG export facilities increasing shipping traffic for the port and thus increasing ambient noise in the fjord system. The purpose of this study is to 1) quantify the existing sound levels in the area surrounding Gil Island and 2) identify potential source mechanisms in order to provide a baseline study of the acoustic environment in the Kitimat Fjord system prior to potential increases from LNG shipping.

Regularized iterative and noniterative procedures for object restoration in the presence of noise :an error analysis

An analysis is carried out, using the prolate spheroidal wave functions, of certain regularized iterative and noniterative methods previously proposed for the achievement of object restoration (or, equivalently, spectral extrapolation) from noisy image data. The ill-posedness inherent in the problem is treated by means of a regularization parameter, and the analysis shows explicitly how the deleterious effects of the noise are then contained. The error in the object estimate is also assessed, and it is shown that the optimal choice for the regularization parameter depends on the signal-to-noise ratio. Numerical examples are used to demonstrate the performance of both unregularized and regularized procedures and also to show how, in the unregularized case, artefacts can be generated from pure noise. Finally, the relative error in the estimate is calculated as a function of the degree of superresolution demanded for reconstruction problems characterized by low space–bandwidth products.

Object Restoration Beyond the Rayleigh Limit in the Presence of Noise

info:eu-repo/semantics/published

Measurements and simulations of noise imposed on supercontinuum generated in microstructure fiber

We present experimental and theoretical investigations of the highly nonlinear and broadband noise that exists on supercontinuum spectra generated from launching femtosecond Ti:Sapphire pulses into microstructure fiber.

Amplitude noise on supercontinuum generated in microstructure fiber: Measurements and simulations

Supercontinua generated in microstructure fiber can exhibit significant excess amplitude noise. We present experimental and numerical studies of the origins of this excess noise and its dependence on the input laser pulse parameters.

Fundamental Amplitude Noise Limitations to Supercontinuum Spectra Generated in a Microstructured Fiber

Broadband supercontinuum spectra are generated in a microstructured fiber using femtosecond laser pulses. Noise properties of these spectra are studied through experiments and numerical simulations based on a generalized stochastic nonlinear Schrödinger equation. In particular, the relative intensity noise as a function of wavelength across the supercontinuum is measured over a wide range of input pulse parameters, and experimental results and simulations are shown to be in good quantitative agreement. For certain input pulse parameters, amplitude fluctuations as large as 50% are observed. The simulations clarify that the intensity noise on the supercontinuum arises from the amplification of two noise inputs during propagation - quantum-limited shot noise on the input pulse, and spontaneous Raman scattering in the fiber. The amplification factor is a sensitive function of the input pulse parameters. Short input pulses are critical for the generation of very broad supercontinua with low noise.