No description available.

This thesis studies statistical image processing of high frequency ultrasound imaging, with application to in-vivo exploration of human skin and noninvasive lesion assessment. More precisely, Bayesian methods are considered in order to perform tissue segmentation in ultrasound images of skin. It is established that ultrasound signals backscattered from skin tissues converge to a complex Levy Flight random process with non-Gaussian _-stable statistics. The envelope signal follows a generalized (heavy-tailed) Rayleigh distribution. Based on these results, it is proposed to model the distribution of multiple-tissue ultrasound images as a spatially coherent finite mixture of heavy-tailed Rayleigh distributions. Spatial coherence inherent to biological tissues is modeled by a Potts Markov random field. An original Bayesian algorithm combined with a Markov chain Monte Carlo method is then proposed to jointly estimate the mixture parameters and a label-vector associating each voxel to a tissue. The proposed method is successfully applied to the segmentation of in-vivo skin tumors in high frequency 2D and 3D ultrasound images. This method is subsequently extended by including the estimation of the Potts regularization parameter B within the Markov chain Monte Carlo (MCMC) algorithm. Standard MCMC methods cannot be applied to this problem because the likelihood of B is intractable. This difficulty is addressed by using a likelihood-free Metropolis-Hastings algorithm based on the sufficient statistic of the Potts model. The resulting unsupervised segmentation method is successfully applied to tridimensional ultrasound images. Finally, the problem of computing the Cramer-Rao bound (CRB) of B is studied. The CRB depends on the derivatives of the intractable normalizing constant of the Potts model. This is resolved by proposing an original Monte Carlo algorithm, which is successfully applied to compute the CRB of the Ising and Potts models.

With the increased capacity of the Internet it is possible to use it for real-time multimedia applications, such as video conferences. There is also an increasing interest in using Internet technology in cellular networks, such as mobile phones. However, the link characteristics for a wireless network are very different from the characteristics of a wired network, which the Internet originally was designed for. The main difference is the time varying link characteristics of the wireless link channel. The variation are due to multipath fading and interference. Herein it is investigated how the wireless link can cooperate with the source coder and network transportation to increase the end-to-end quality, without increasing the delay. Under the assumption that the wireless link is the most likely bottleneck of the system. It is shown that by using a progressive source coder and extending the Internet protocols to enable quick rate adaption by routers and wireless links it is possible to increase the end-to-end quality. / Godkänd; 2003; 20070215 (ysko)

<p>We have attended an online presentation through Adobe connect.</p>

The estimation of frequencies and corresponding harmonic overtones is a problem of great importance in many situations. Applications can, for example, be found in supervision of electrical power transmission lines, in seismology and in acoustics. Generally, a periodic function with an unknown fundamental frequency in cascade with a parameterized and unknown nonlinear function can be used as a signal model for an arbitrary periodic signal. The main objective of the proposed modeling technique is to estimate the fundamental frequency of the periodic function in addition to the parameters of the nonlinear function. The thesis is divided into four parts. In the first part, a general introduction to the harmonic signal modeling problem and different approaches to solve the problem are given. Also, an outline of the thesis and future research topics are introduced. In the second part, a previously suggested recursive prediction error method (RPEM) for harmonic signal modeling is studied by numerical examples to explore the ability of the algorithm to converge to the true parameter vector. Also, the algorithm is modified to increase its ability to track the fundamental frequency variations. A modified algorithm is introduced in the third part to give the algorithm of the second part a more stable performance. The modifications in the RPEM are obtained by introducing an interval in the nonlinear block with fixed static gain. The modifications that result in the convergence analysis are, however, substantial and allows a complete treatment of the local convergence properties of the algorithm. Moreover, the Cramér–Rao bound (CRB) is derived for the modified algorithm and numerical simulations indicate that the method gives good results especially for moderate signal to noise ratios (SNR). In the fourth part, the idea is to give the algorithm of the third part the ability to estimate the driving frequency and the parameters of the nonlinear output function parameterized also in a number of adaptively estimated grid points. Allowing the algorithm to automatically adapt the grid points as well as the parameters of the nonlinear block, reduces the modeling errors and gives the algorithm more freedom to choose the suitable grid points. Numerical simulations indicate that the algorithm converges to the true parameter vector and gives better performance than the fixed grid point technique. Also, the CRB is derived for the adaptive grid point technique.

In many applications such as heat diffusion and flow problems, it is of interest to describe the process behavior inside a particular medium. An example can be the strive for estimating certain parameters related to the material. These processes are often modeled by a partial differential equation. Certain methods for identifying unknown material constants require the model to be of finite order. This thesis describes how the diffusion process can be approximated with finite order model, and how the accuracy of an estimated model depends on the model order. In particular, a detailed analysis is carried out for the case when the approximate model accounts for solving the diffusion by a difference method.

The problem of identifying dynamic errors-in-variables models is of fundamental interest in many areas like process control, array signal processing, astronomical data reduction. In recent years, this field has received increased attention of the research community. In this thesis, some time domain and frequency domain approaches for identifying the errors-in-variables model is studied. The first chapter gives an overview of various methods for identifying dynamic errors-in-variables systems. Several approaches are classified and a qualitative comparison of different existing methods is also presented. The second chapter deals with instrumental variables based approaches. The least squares and the total least squares methods of solving the Yule–Walker equation is of central interest here. The methods are compared from the view point of asymptotic performance, numerical robustness and computation. The method presented in the third chapter uses prefiltered data. The input-output data is passed through a pair of user defined prefilters and the output data from the prefilters is subjected to a least-squares like algorithm. Compared to the IV approach, the proposed method shows a significant improvement in the small-sample properties of the MA parameter estimates, without any increase in the computational load. In the fourth chapter, we show that the two-dimensional process composed of the input-output data admits a finite order ARMA representation. Then we propose a parametric identification algorithm and another non-parametric identification method based on the ARMA representation.

Identification of dynamic errors-in-variables systems, where both inputs and outputs are affected by errors (measurement noises), is a fundamental problem of great interest in many areas, such as process control, econometrics, astronomical data reduction, image processing, etc. This field has received increased attention within several decades. Many solutions have been proposed with different approaches. In this thesis, the focus is on some specific problems concerning two time domain methods for identifying linear dynamic errors-in-variables systems. The thesis is divided into four parts. In the first part, a general introduction to the problem of identifying errors-in-variables systems and different approaches to solve the problem are given. Also, a summary of the contributions and some topics for future works are presented. The second part of the thesis considers the instrumental variables based approaches. They are studied under the periodic excitation condition. The main motivation is to analyze what type of instrumental variables should be chosen to maximally utilize the information of the periodic measurements. A particular overdetermined instrumental variable estimator is proposed, which can achieve optimal performance without weighting. The asymptotic convergence properties of the Bias-eliminating least squares (BELS) methods are investigated in the third part. By deriving an error dynamics equation for the parameter estimates, it is shown that the convergence of the bias-eliminating algorithms is determined by the largest magnitude of the eigenvalues of the system matrix. To overcome the possible divergence of the iteration-type bias-eliminating algorithms under very low signal-to-noise ratio, we reformulate the bias-elimination problem as a minimization problem and develop a variable projection algorithm to perform consistent parameter estimation. Part four contains an analysis of the accuracy properties of the BELS estimates. It is shown that the estimated system parameters and the estimated noise variances are asymptotically Gaussian distributed. An explicit expression for the normalized asymptotic covariance matrix of the estimated system parameters and the estimated noise variances is derived.

No description available.

In recent years, the three-dimensional (3D) movie industry has reaped massive commercial success in the theaters. With the advancement of display technologies, more experienced capturing and generation of 3D contents, TV broadcasting, movies, and games in 3D have entered home entertainment, and it is likely that 3D applications will play an important role in many aspects of people's life in a not distant future. 3D video contents contain at least two views from different perspectives for the left and the right eye of viewers. The amount of coded information is doubled if these views are encoded separately. Moreover, for multi-view displays (i.e. different perspectives of a scene in 3D are presented to the viewer at the same time through different angles), either video streams of all the required views must be transmitted to the receiver, or the displays must synthesize the missing views with a subset of the views. The latter approach has been widely proposed to reduce the amount of data being transmitted and make data adjustable to 3D-displays. The virtual views can be synthesized by the Depth Image Based Rendering (DIBR) approach from textures and associated depth images. However, it is still the case that the amount of information for the textures plus the depths presents a significant challenge for the network transmission capacity. Compression techniques are vital to facilitate the transmission. In addition to multi-view and multi-view plus depth for reproducing 3D, light field techniques have recently become a hot topic. The light field capturing aims at acquiring not only spatial but also angular information of a view, and an ideal light field rendering device should be such that the viewers would perceive it as looking through a window. Thus, the light field techniques are a step forward to provide us with a more authentic perception of 3D. Among many light field capturing approaches, focused plenoptic capturing is a solution that utilize microlens arrays. The plenoptic cameras are also portable and commercially available. Multi-view and refocusing can be obtained during post-production from these cameras. However, the captured plenoptic images are of a large size and contain significant amount of a redundant information. An efficient compression of the above mentioned contents will, therefore, increase the availability of content access and provide a better quality experience under the same network capacity constraints. In this thesis, the compression of depth images and of plenoptic contents captured by focused plenoptic cameras are addressed. The depth images can be assumed to be piece-wise smooth. Starting from the properties of depth images, a novel depth image model based on edges and sparse samples is presented, which may also be utilized for depth image post-processing. Based on this model, a depth image coding scheme that explicitly encodes the locations of depth edges is proposed, and the coding scheme has a scalable structure. Furthermore, a compression scheme for block-based 3D-HEVC is also devised, in which diffusion is used for intra prediction. In addition to the proposed schemes, the thesis illustrates several evaluation methodologies, especially the subjective test of the stimulus-comparison method. This is suitable for evaluating the quality of two impaired images, as the objective metrics are inaccurate with respect to synthesized views. For the compression of plenoptic contents, displacement intra prediction with more than one hypothesis is applied and implemented in the HEVC for an efficient prediction. In addition, a scalable coding approach utilizing a sparse set and disparities is introduced for the coding of focused plenoptic images. The MPEG test sequences were used for the evaluation of the proposed depth image compression, and public available plenoptic image and video contents were applied to the assessment of the proposed plenoptic compression. For depth image coding, the results showed that virtual views synthesized from post-processed depth images by using the proposed model are better than those synthesized from original depth images. More importantly, the proposed coding schemes using such a model produced better synthesized views than the state of the art schemes. For the plenoptic contents, the proposed scheme achieved an efficient prediction and reduced the bit rate significantly while providing coding and rendering scalability. As a result, the outcome of the thesis can lead to improving quality of the 3DTV experience and facilitate the development of 3D applications in general.