This thesis investigates the improvements that can be made to Bayesian passive sonar tracking in the context of active-passive sonar data fusion. Performance improvements are achieved by exploiting the prior information available within a typical Bayesian data fusion framework. The algorithms developed are tested against both simulated data and data measured during the SEABAR 07 sea trial. Results show that the proposed approaches achieve improved detection, decreased estimation error, and the ability to track quiet targets in the presence of loud interferers. / text

No description available.

Using cross-correlation analysis, we demonstrate that flux measurements at observation locations during hydraulic tomography (HT) surveys carry non-redundant information about heterogeneity that are complementary to head measurements at the same locations. We then hypothesize that a joint interpretation of head and flux data can enhance the resolution of HT estimates. Subsequently, we use numerical experiments to test this hypothesis and investigate the impact of stationary and non-stationary hydraulic conductivity field, and prior information such as correlation lengths, and initial mean models (uniform or distributed means) on HT estimates. We find that flux and head data from HT have already possessed sufficient heterogeneity characteristics of aquifers. While prior information (as uniform mean or layered means, correlation scales) could be useful, its influence on the estimates is limited as more non-redundant data are used in the HT analysis (see Yeh and Liu [2000]). Lastly, some recommendation for conducting HT surveys and analysis are presented.

No description available.

No description available.

京都大学 / 新制・課程博士 / 博士(工学) / 甲第24146号 / 工博第5033号 / 新制||工||1786(附属図書館) / 京都大学大学院工学研究科航空宇宙工学専攻 / (主査)教授 藤本 健治, 教授 松野 文俊, 教授 森本 淳 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

This research focuses on the development of a deterministic-probabilistic framework for agricultural land use and management, specifically for both annual crops, such as wheat, barley and maize, and permanent crops, such as vineyards. The goal is to predict crop greening and peak crop development progressively through the growing season, based on accumulating information as the crop develops and matures, and to provide an accompanying uncertainty statement (credible interval) with each prediction. The integrated area underneath the phenology curve can be associated, although not explicitly in our example, with per-area crop yield. The prediction model relies on remotely sensed data, including science data products from the Landsat and MODIS (Moderate Resolution Imaging Spectroradiometer) spaceborne instruments, field data from agro-meteorological stations, and statistical data from prior years. The development of the deterministic-probabilistic model focuses on northeastern Italy, a region of small agricultural plots set in a diverse physical landscape, which is typical of many areas of old-world and developing-nation agriculture. The estimation process uses the phenological cycle of the MODIS Enhanced Vegetation Index (EVI), extracted from the satellite imagery at 500 m spatial resolution. Landsat data, at 30-m spatial resolution, are fused with MODIS data, to provide fine-scale information better suited to small-field agriculture. By applying a piecewise logistic function to model the time trajectory of EVI values, crop development and peak greenness are estimated and characterized based on the main phenological stages determined from the remote imagery trained with ground station observations. The deterministic-probabilistic model is later validated with observations from reference testing stations and statistical crop and yield data obtained independently by administrative districts such as regional and national organizations. A temporal filter of the main phenological stages, here called a crop calendar, plays a critical role. A Bayesian approach to integrate stochastically the parameters related to a certain area provides a way to include the different datasets at the different dimensions and scales and to assess the probability to obtain a vegetation index within a given uncertainty. The model becomes, therefore, a typical generalized linear model problem, deterministically described by a piecewise logistic function, with the parameters describing the peak phenological curve estimated probabilistically, with their own uncertainty. / 2026-01-31T00:00:00Z

Labeling problem, due to its versatile modeling ability, is widely used in various image analysis tasks. In practice, certain prior information is often available to be embedded in the model to increase accuracy and robustness. However, it is not always straightforward to formulate the problem so that the prior information is correctly incorporated. It is even more challenging that the proposed model admits efficient algorithms to find globally optimal solution. In this dissertation, a series of natural and medical image segmentation tasks are modeled as labeling problems. Each proposed model incorporates different useful prior information. These prior information includes ordering constraints between certain labels, soft user input enforcement, multi-scale context between over-segmented regions and original voxel, multi-modality context prior, location context between multiple modalities, star-shape prior, and gradient vector flow shape prior. With judicious exploitation of each problem's intricate structure, efficient and exact algorithms are designed for all proposed models. The efficient computation allow the proposed models to be applied on large natural and medical image datasets using small memory footprint and reasonable time assumption. The global optimality guarantee makes the methods robust to local noise and easy to debug. The proposed models and algorithms are validated on multiple experiments, using both natural and medical images. Promising and competitive results are shown when compared to state-of-art.

Segmentation of individual facial soft tissues has received relatively little attention in the literature due to the complicated structures of these tissues. There is a need to incorporate the prior information, which is usually in the form of atlases, in the segmentation process. In this thesis we performed several segmentation methods that take advantage of prior knowledge for facial soft tissue segmentation. An atlas based method and three expectation maximization &ndash / Markov random field (EM-MRF) based methods are tested for two dimensional (2D) segmentation of masseter muscle in the face. Atlas based method uses the manually labeled atlases as prior information. We implemented EM-MRF based method in different manners / without prior information, with prior information for initialization and with using labeled atlas as prior information. The differences between these methods and the influence of the prior information are discussed by comparing the results. Finally a new method based on EM-MRF is proposed in this study. In this method we aim to use prior information without performing manual segmentation, which is a very complicated and time consuming task. 10 MRI sets are used as experimental data in this study and leave-one-out technique is used to perform segmentation for all sets. The test data is modeled as a Markov Random Field where unlabeled training data, i.e., other 9 sets, are used as prior information. The model parameters are estimated by the Maximum Likelihood approach when the Expectation Maximization iterations are used to handle hidden labels. The performance of all segmentation methods are computed and compared to the manual segmented ground truth. Then we used the new 2D segmentation method for three dimensional (3D) segmentation of two masseter and two temporalis tissues in each data set and visualize the segmented tissue volumes.

A Tomografia por Impedância Elétrica tem como objetivo estimar a distribuição de impedância elétrica dentro de uma região a partir de medidas de potencial elétrico coletadas apenas em seu contorno externo quando corrente elétrica é imposta neste mesmo contorno. Uma das aplicações para esta tecnologia é o monitoramento das condições pulmonares de pacientes em Unidades de Tratamento Intensivo. Dentre vários algoritmos, destacam-se os filtros de Kalman que abordam o problema de estimação sob o ponto de vista probabilístico, procurando encontrar a distribuição de probabilidade do estado condicionada à realização das medidas. Para que estes filtros possam ser utilizados, um modelo de evolução temporal do sistema sendo observado deve ser adotado. Esta tese propõe o uso de um modelo de evolução para a variação de volume de ar nos pulmões durante a respiração de um paciente sob ventilação artificial. Este modelo é utilizado no unscented Kalman filter, uma extensão não linear do filtro de Kalman. Tal modelo é ajustado em paralelo à estimação do estado, utilizando um esquema dual de estimação. Um algoritmo de segmentação de imagem é proposto para identificar as regiões pulmonares nas imagens estimadas e assim utilizar o modelo de evolução. Com o intuito de melhorar as estimativas, o método do erro de aproximação é utilizado no modelo de observação para mitigar os erros de modelagem e informação a priori é adicionada na solução do problema inverso mal-posto. O método é avaliado através de simulações numéricas e ensaio experimental coletado em um voluntário. Os resultados mostram que o método proposto melhora as estimativas feitas pelo filtro de Kalman, propiciando a visualização de imagens absolutas, dinâmicas e com bom nível de contraste entre os tecidos e órgãos internos. / Electrical impedance tomography estimates the electrical impedance distribution within a region given a set of electrical potential measurements acquired along its boundary at the same time that electrical currents are imposed on the same boundary. One of the applications of this technology is lung monitoring of patients in Intensive Care Units. One class of algorithms employed for the estimation are the Kalman filters which deal with the estimation problem in a probabilistic framework, looking for the probability density function of the state conditioned to the acquired measurements. In order to use such filters, an evolution models of the system must be employed. This thesis proposes an evolution model of the variation of air in the lungs of patients under artificial ventilation. This model is used on the Unscented Kalman Filter, a nonlinear extension of the Kalman filter. This model is adjusted in parallel to the state estimation, in a dual estimation scheme. An image segmentation algorithm is proposed for identifying the lungs in the images. In order to improve the estimate, the approximation error method is employed for mitigating the observation model errors and prior information is added for the solution of the ill-posed inverse problem. The method is evaluated with numerical simulations and with experimental data of a volunteer. The results show that the proposed method increases the quality of the estimates, allowing the visualization of absolute and dynamic images, with good level of contrast between the tissues and internal organs.