Dissertations / Theses on the topic 'Iterative enhancement'

Contemporary algorithms employed for reconstruction of 3D volumes from helical cone beam projections are so called non-exact algorithms. This means that the reconstructed volumes will contain artifacts irrespective of the detector resolution and number of projections angles employed in the process.

It has been proposed that these artifacts can be suppressed using an iterative scheme which comprises computation of projections from the already reconstructed volume as well as the non-exact reconstruction itself.

The purpose of the present work is to examine if the iterative scheme can be applied to the non-exact reconstruction method PI-original in order to improve the reconstruction result. An important part in this implementation is a careful design of the projection operator, as a poorly designed projection operator may result in aliasing and/or other artifacts in the reconstruction result. Since the projection data is truncated, special care must be taken along the boundaries of the detector. Three different ways of handling this interpolation problem is proposed and examined.

The results show that artifacts caused by the PI-original method can indeed be reduced by the iterative scheme. However, each iteration requires at least three times more processing time than the initial reconstruction, which may call for certain compromises, smartness and/or parallelization in the innermost loops. Furthermore, at higher cone angles certain types of artifacts seem to grow by each iteration instead of being suppressed.

Thesis (M.S. in electrical engineering)--Washington State University, August 2010.
Title from PDF title page (viewed on July 28, 2010). "School of Electrical Engineering and Computer Science." Includes bibliographical references (p. 51).

Kwok, Ka Wai.
Thesis (M.Phil.)--Chinese University of Hong Kong, 2005.
Includes bibliographical references (leaves 113-116).
Abstracts in English and Chinese.
ABSTRACT --- p.i
Chapter 1. --- INTRODUCTION --- p.1
Chapter 1.1 --- Artistic robot in western art --- p.1
Chapter 1.2 --- Chinese calligraphy robot --- p.2
Chapter 1.3 --- Our robot drawing system --- p.3
Chapter 1.4 --- Thesis outline --- p.3
Chapter 2. --- ROBOT DRAWING SYSTEM --- p.5
Chapter 2.1 --- Robot drawing manipulation --- p.5
Chapter 2.2 --- Input modes --- p.6
Chapter 2.3 --- Visual-feedback system --- p.8
Chapter 2.4 --- Footprint study setup --- p.8
Chapter 2.5 --- Chapter summary --- p.10
Chapter 3. --- LINE STROKE EXTRACTION AND ORDER ASSIGNMENT --- p.11
Chapter 3.1 --- Skeleton-based line trajectory generation --- p.12
Chapter 3.2 --- Line stroke vectorization --- p.15
Chapter 3.3 --- Skeleton tangential slope evaluation using MIC --- p.16
Chapter 3.4 --- Skeleton-based vectorization using Bezier curve interpolation --- p.21
Chapter 3.5 --- Line stroke extraction --- p.25
Chapter 3.6 --- Line stroke order assignment --- p.30
Chapter 3.7 --- Chapter summary --- p.33
Chapter 4. --- PROJECTIVE RECTIFICATION AND VISION-BASED CORRECTION --- p.34
Chapter 4.1 --- Projective rectification --- p.34
Chapter 4.2 --- Homography transformation by selected correspondences --- p.35
Chapter 4.3 --- Homography transformation using GA --- p.39
Chapter 4.4 --- Visual-based iterative correction example --- p.45
Chapter 4.5 --- Chapter summary --- p.49
Chapter 5. --- ITERATIVE ENHANCEMENT ON OFFSET EFFECT AND BRUSH THICKNESS --- p.52
Chapter 5.1 --- Offset painting effect by Chinese brush pen --- p.52
Chapter 5.2 --- Iterative robot drawing process --- p.53
Chapter 5.3 --- Iterative line drawing experimental results --- p.56
Chapter 5.4 --- Chapter summary --- p.67
Chapter 6. --- GA-BASED BRUSH STROKE GENERATION --- p.68
Chapter 6.1 --- Brush trajectory representation --- p.69
Chapter 6.2 --- Brush stroke modeling --- p.70
Chapter 6.3 --- Stroke simulation using GA --- p.72
Chapter 6.4 --- Evolutionary computing results --- p.77
Chapter 6.5 --- Chapter summary --- p.95
Chapter 7. --- BRUSH STROKE FOOTPRINT CHARACTERIZATION --- p.96
Chapter 7.1 --- Footprint video capturing --- p.97
Chapter 7.2 --- Footprint image property --- p.98
Chapter 7.3 --- Experimental results --- p.102
Chapter 7.4 --- Chapter summary --- p.109
Chapter 8. --- CONCLUSIONS AND FUTURE WORKS --- p.111
BIBLIOGRAPHY --- p.113

碩士
國立臺北科技大學
電機工程系研究所
100
Speech signals are tend to decrease the speech quality when corrupted by background noises. The aim of speech enhancement is to reduce the background noise from a noisy speech signal while keeping the speech distortion as low as possible. And this Speech technique is usually used in speech transmission and speech recognition that recovers the clean speech from noisy speech by using a noise tracking algorithm. There are three categories for speech enhancement including filtering techniques, spectral restoration techniques, and speech model techniques. In this thesis three speech enhancement methods based on linear predictive model are investigated that includes Kalman filter (KF), modified Kalman filter (MKF), and iterative Wiener filter (IWF). Two other famous methods the Wiener filter (WF) method and maximum-likelihood spectral amplitude (MLSA) method are also included for comparison. In the experiments, each enhancement method incorporates with three well-known noise tracking algorithms, including minimum statistics (MS), minima controlled recursive averaging (MCRA), and improved minima controlled recursive averaging (IMCRA) for recovering clean speech. The experimental results show that compared with the Wiener filter and maximum-likelihood spectral amplitude, the proposed iterative Wiener filter in the linear predictive model provides superior performance. Among all combinations, the latter with MCRA noise tracking can achieves the most excellent results.

碩士
亞洲大學
資訊傳播學系碩士班
99
Most speech enhancement systems suffer from annoying residual noise. Many studies attempt to suppress the magnitude of residual noise, enabling the speech quality to be kept at an acceptable level. An auditory masking-property adapted speech enhancement is the best method among the algorithms. It is due to the fact that this method preserves inaudible residual noise by preserving more amounts of noisy speech components, yielding the reduction of speech distortion. However, the residual noise is still annoying to the human ear in the enhanced speech. The major reason is the inaccurate estimation on the magnitude of background noise. In this study, we aim at improving the performance of a perceptual speech enhancement system by reducing more amount of residual noise. This can be obtained by iteratively modifying the noise masking threshold in the perceptual gain factor. Experimental results show that the proposed method can significantly improve the performance of the perceptual gain in noise reduction.

碩士
國立中興大學
電機工程學系所
98
In this paper, we propose an adaptive image enlargement scheme based on iterative back-projection. Initial estimates of each enlarged image can be individually created from the spatial and temporal domains by using sub-pixel interpolation and sub-pixel motion estimation. Then, based on the initial estimates and image content, reconstructed images are derived by using a modified iterative back-projection technique and fused into a enlarged image. Finally, a low-pass filter as a post-processing is applied to reduce the blocking artifacts in the reconstructed high-resolution images. Our experiment results demonstrate that, in terms of PSNR and NQM, the proposed scheme is superior to existing methods.

碩士
國立中正大學
前瞻製造系統碩士學位學程
106
Helical interpolation machining is a widely used technique in manufacturing and arms industries to create helical contours or take advantages of this technique in machining processes. In general, reducing helical contour errors is a difficult task because the helix is a 3 -dimensional curve, making it extremely hard to estimate the actual contour error through individual axis-tracking errors. In recent years, iterative learning control (ILC) has been successfully applied to precision motion controllers specializing in repeated tasks. Conventionally, most of them have focused on reducing axis-tracking errors, however, that cannot guarantee to obtain smaller actual contour errors in general. To overcome these difficulties, a new concept “equivalent contour error” model is taken as our control objective instead of using the complex actual contour model. In this study, an online ILC framework will be introduced to gradually enhance helical contours via learning processes by adjusting input commands. The online terminology meaning that after data coming, we simultaneously update the input command for the next learning iteration at each time step. This online technique avoids the batch learning process of so-called offline ILC, which consumes a huge amount of memory to save collected data along with high computation time. In particular, our proposed control law is able not only to iteratively reduce the control objective but also to deal with initial state errors problem resulting from different initial states at each learning iteration. Furthermore, in our learning control framework, we employ a fuzzy decision support system to adaptively select local learning convergence rates for speeding up the learning process and preventing the noise amplification phenomenon. Practically, we employ a PC-based controller board connected to a real CNC machine to control feed-drive systems. Our algorithm is implemented in C-programming language, which is one of the fastest computing languages, and optimally organized in terms of data structures. Finally, experimental results validate our proposed online iterative learning control framework and verify the feasibility of integrating the advanced control function into real precision motion controllers.

碩士
國立臺北科技大學
電腦與通訊研究所
102
In this thesis, we mainly study the performance improvement of channel estimation in MIMO-OFDMA systems. In order to improve the quality of the channel estimation of the comb type arrangement, we firstly use the least square (LS) algorithm to estimate the channel frequency response of pilots. Secondly, we use the iterative interpolation method (IIM) to interpolate the channel frequency response of data signals. Then, the channel impulse response is obtained by using the inverse fast Fourier transform (IFFT). To find the frequency response of the actual channel, we utilize the improved path detection (PD) method to estimate the accurate channel path. One threshold was set to separate the channel impulse response and noise in this PD method. If the magnitude of the channel impulse response is greater than the threshold, the corresponding channel can be regarded as main channel path. The others are regarded as noise and are set to zero to improve the accuracy of estimation, enabling the estimated results to close to the actual channel frequency response. Experimental results show that the proposed scheme obtains the best performance in terms of the bit error rate (BER) among the IIM scheme combined with the different methods, such as, the traditional LS, FFT-based, Hayashi, PD, and the least mean square (LMS) estimation methods.

For time-varying signals such as speech and audio, short-time analysis becomes necessary to compute specific signal attributes and to keep track of their evolution. The standard technique is the short-time Fourier transform (STFT), using which one decomposes a signal in terms of windowed Fourier bases. An advancement over STFT is the wavelet analysis in which a function is represented in terms of shifted and dilated versions of a localized function called the wavelet. A specific modeling approach particularly in the context of speech is based on short-time linear prediction or short-time Wiener filtering of noisy speech. In most nonstationary signal processing formalisms, the key idea is to analyze the properties of the signal locally, either by first truncating the signal and then performing a basis expansion (as in the case of STFT), or by choosing compactly-supported basis functions (as in the case of wavelets). We retain the same motivation as these approaches, but use polynomials to model the signal on a short-time basis (“short-time polynomial representation”). To emphasize the local nature of the modeling aspect, we refer to it as “local polynomial modeling (LPM).” We pursue two main threads of research in this thesis: (i) Short-time approaches for speech enhancement; and (ii) LPM for enhancing smooth signals, with applications to ECG, noisy nonuniformly-sampled signals, and voiced/unvoiced segmentation in noisy speech. Improved iterative Wiener filtering for speech enhancement A constrained iterative Wiener filter solution for speech enhancement was proposed by Hansen and Clements. Sreenivas and Kirnapure improved the performance of the technique by imposing codebook-based constraints in the process of parameter estimation. The key advantage is that the optimal parameter search space is confined to the codebook. The Nonstationary signal enhancement solutions assume stationary noise. However, in practical applications, noise is not stationary and hence updating the noise statistics becomes necessary. We present a new approach to perform reliable noise estimation based on spectral subtraction. We first estimate the signal spectrum and perform signal subtraction to estimate the noise power spectral density. We further smooth the estimated noise spectrum to ensure reliability. The key contributions are: (i) Adaptation of the technique for non-stationary noises; (ii) A new initialization procedure for faster convergence and higher accuracy; (iii) Experimental determination of the optimal LP-parameter space; and (iv) Objective criteria and speech recognition tests for performance comparison. Optimal local polynomial modeling and applications We next address the problem of fitting a piecewise-polynomial model to a smooth signal corrupted by additive noise. Since the signal is smooth, it can be represented using low-order polynomial functions provided that they are locally adapted to the signal. We choose the mean-square error as the criterion of optimality. Since the model is local, it preserves the temporal structure of the signal and can also handle nonstationary noise. We show that there is a trade-off between the adaptability of the model to local signal variations and robustness to noise (bias-variance trade-off), which we solve using a stochastic optimization technique known as the intersection of confidence intervals (ICI) technique. The key trade-off parameter is the duration of the window over which the optimum LPM is computed. Within the LPM framework, we address three problems: (i) Signal reconstruction from noisy uniform samples; (ii) Signal reconstruction from noisy nonuniform samples; and (iii) Classification of speech signals into voiced and unvoiced segments. The generic signal model is x(tn)=s(tn)+d(tn),0 ≤ n ≤ N - 1. In problems (i) and (iii) above, tn=nT(uniform sampling); in (ii) the samples are taken at nonuniform instants. The signal s(t)is assumed to be smooth; i.e., it should admit a local polynomial representation. The problem in (i) and (ii) is to estimate s(t)from x(tn); i.e., we are interested in optimal signal reconstruction on a continuous domain starting from uniform or nonuniform samples. We show that, in both cases, the bias and variance take the general form: The mean square error (MSE) is given by where L is the length of the window over which the polynomial fitting is performed, f is a function of s(t), which typically comprises the higher-order derivatives of s(t), the order itself dependent on the order of the polynomial, and g is a function of the noise variance. It is clear that the bias and variance have complementary characteristics with respect to L. Directly optimizing for the MSE would give a value of L, which involves the functions f and g. The function g may be estimated, but f is not known since s(t)is unknown. Hence, it is not practical to compute the minimum MSE (MMSE) solution. Therefore, we obtain an approximate result by solving the bias-variance trade-off in a probabilistic sense using the ICI technique. We also propose a new approach to optimally select the ICI technique parameters, based on a new cost function that is the sum of the probability of false alarm and the area covered over the confidence interval. In addition, we address issues related to optimal model-order selection, search space for window lengths, accuracy of noise estimation, etc. The next issue addressed is that of voiced/unvoiced segmentation of speech signal. Speech segments show different spectral and temporal characteristics based on whether the segment is voiced or unvoiced. Most speech processing techniques process the two segments differently. The challenge lies in making detection techniques offer robust performance in the presence of noise. We propose a new technique for voiced/unvoiced clas-sification by taking into account the fact that voiced segments have a certain degree of regularity, and that the unvoiced segments do not possess any smoothness. In order to capture the regularity in voiced regions, we employ the LPM. The key idea is that regions where the LPM is inaccurate are more likely to be unvoiced than voiced. Within this frame-work, we formulate a hypothesis testing problem based on the accuracy of the LPM fit and devise a test statistic for performing V/UV classification. Since the technique is based on LPM, it is capable of adapting to nonstationary noises. We present Monte Carlo results to demonstrate the accuracy of the proposed technique.