Orientador: Clésio Luis Tozzi / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenhaia Elétrica e de Computação / Made available in DSpace on 2018-08-16T00:09:42Z (GMT). No. of bitstreams: 1 SoutoJunior_CarlosAlberto_M.pdf: 16501917 bytes, checksum: 490a2364c9bd25c00b6cfa939af84889 (MD5) Previous issue date: 2010 / Resumo: Este trabalho consiste em uma avaliação de descritores de atributos de textura para o caso totalmente não-supervisionado, na qual nada se conhece anteriormente sobre a natureza das texturas ou o número de regiões presentes na imagem. Escolheram-se para descrever as texturas decomposição por filtros de Gabor, descritores escalares baseados em matrizes de co-ocorrência de níveis de cinza e campos aleatórios de Gauss-Markov; e aplicou-se um procedimento baseado no algoritmo k-means, onde o valor ótimo do parâmetro k foi estimado a partir de uma métrica de qualidade calculada nos resultados da execução do algoritmo k-means para vários valores de k. O k ótimo foi obtido pelo "método do cotovelo". Aplicou-se o procedimento em imagens sintéticas e naturais e confrontou-se com uma segmentação manual. Obtiveram-se melhores resultados para imagens agrícolas de baixa altitude e tipo frente-fundo quando usados descritores baseados em matrizes de co-ocorrência; nas imagens de satélite, o método que emprega campos aleatórios foi melhor sucedido / Abstract: This work comprises a texture features descriptors evaluation focusing the fully unsupervised case, where neither the texture nature nor the numbers of regions in the image are previously known. Three distinct texture descriptors were chosen: Image decomposition with Gabor filters, scalar descriptors based in gray-level co-occurrence matrix and Gauss-Markov random fields; and an automatic region number determination framework was applied. For performance evaluation, the procedure was applied in both synthetic and natural images / Mestrado / Engenharia de Computação / Mestre em Engenharia Elétrica

Scene parsing, or segmenting all the objects in an image and identifying their categories, is one of the core problems of computer vision. In order to achieve an object-level semantic segmentation, we build upon the recent superparsing approach by Tighe and Lazebnik, which is a nonparametric solution to the image labeling problem. Superparsing consists of four steps. For a new query image, the most similar images from the training dataset of labeled images is retrieved based on global features. In the second step, the query image is segmented into superpxiels and 20 di erent local features are computed for each superpixel. We propose to use the SLICO segmentation method to allow control of the size, shape and compactness of the superpixels because SLICO is able to produce accurate boundaries. After all superpixel features have been extracted, feature-based matching of superpixels is performed to nd the nearest-neighbour superpixels in the retrieval set for each query superpxiel. Based on the neighbouring superpixels a likelihood score for each class is calculated. Finally, we formulate a Conditional Random Field (CRF) using the likelihoods and a pairwise cost both computed from nonparametric estimation to optimize the labeling of the image. Speci cally, we de ne a novel pairwise cost to provide stronger semantic contextual constraints by incorporating the similarity of adjacent superpixels depending on local features. The optimized labeling obtained with the CRF results in superpixels with the same labels grouped together to generate segmentation results which also identify the categories of objects in an image. We evaluate our improvements to the superparsing approach using segmentation evaluation measures as well as the per-pixel rate and average per-class rate in a labeling evaluation. We demonstrate the success of our modi ed approach on the SIFT Flow dataset, and compare our results with the basic superparsing methods proposed by Tighe and Lazebnik.

Campos Aleatórios Condicionais são modelos probabilísticos discriminativos que tem sido utilizados com sucesso em diversas áreas como processamento de linguagem natural, reconhecimento de fala e bioinformática. Entretanto, implementar algoritmos eficientes para esse tipo de modelo não é uma tarefa fácil. Nesse trabalho apresentamos um arcabouço que ajuda no desenvolvimento e experimentação de Campos Aleatórios Condicionais Semi Markovianos (semi-CRFs). Desenvolvemos algoritmos eficientes que foram implementados em C++ propondo uma interface de programação flexível e intuitiva que habilita o usuário a definir, treinar e avaliar modelos. Nossa implementação foi construída como uma extensão do arcabouço ToPS que, inclusive, pode utilizar qualquer modelo já definido no ToPS como uma função de característica especializada. Por fim utilizamos nossa implementação de semi-CRF para construir um preditor de promotores que apresentou performance superior aos preditores existentes. / Conditional Random Fields are discriminative probabilistic models that have been successfully used in several areas like natural language processing, speech recognition and bioinformatics. However, implementing efficient algorithms for this kind of model is not an easy task. In this thesis we show a framework that helps the development and experimentation of Semi-Markov Conditional Random Fields (semi-CRFs). It has an efficient implementation in C++ and an intuitive API that allow users to define, train and evaluate models. It was built as an extension of ToPS framework and can use ToPS probabilistic models as specialized feature functions. We also use our implementation of semi-CRFs to build a high performance promoter predictor.

This thesis explores metric properties of Liouville quantum gravity (LQG), a random geometry with conformal symmetries introduced in the context of string theory by Polyakov in the 80’s. Formally, it corresponds to the Riemannian metric tensor “e^{γh}(dx² + dy²)” where h is a planar Gaussian free field and γ is a parameter in (0, 2). Since h is a random Schwartz distribution with negative regularity, the exponential e^{γh} only makes sense formally and the associated volume form and distance functions are not well-defined. The mathematical language to define the volume form was introduced by Kahane, also in the 80’s. In this thesis, we explore a renormalization approach to make sense of the distance function and we study its basic properties.

No description available.

Thermal cameras are used in numerous computer vision applications, such as human detection and scene understanding. However, the cost of high quality and high resolution thermal sensors is often a limiting factor. Conversely, high resolution visual spectrum cameras are readily available and generally inexpensive. Herein, we explore the creation of higher quality upsampled thermal imagery using a high resolution visual spectrum camera and Markov random fields theory. This paper also presents a discussion of the tradeoffs from this approach and the effects of upsampling, both from quantitative and qualitative perspectives. Our results demonstrate the successful application of this approach for human detection and the accurate propagation of thermal measurements within images for more general tasks like scene understanding. A tradeoff analysis of the costs related to performance as the resolution of the thermal camera decreases are also provided.

Recently, Bayesian methods have become the essence of modern statistics, specifically, the ability to incorporate hierarchical models. In particular, correlated data, such as the data found in spatial and temporal applications, have benefited greatly from the development and application of Bayesian statistics. One particular application of Bayesian modeling is Gaussian Markov Random Fields. These methods have proven to be very useful in providing a framework for correlated data. I will demonstrate the power of GMRFs by applying this method to two sets of data; a set of temporal data involving car accidents in the UK and a set of spatial data involving Provo area apartment complexes. For the first set of data, I will examine how including a seatbelt covariate effects our estimates for the number of car accidents. In the second set of data, we will scrutinize the effect of BYU approval on apartment complexes. In both applications we will investigate Laplacian approximations when normal distribution assumptions do not hold.

The undirected graphical model or Markov Random Field (MRF) is one of the more popular models used in computer vision and is the type of model with which this work is concerned. Models based on these methods have proven to be particularly useful in low-level vision systems and have led to state-of-the-art results for MRF-based systems. The research presented will describe a new discriminative training algorithm and its implementation. The MRF model will be trained by optimizing its parameters so that the minimum energy solution of the model is as similar as possible to the ground-truth. While previous work has relied on time-consuming iterative approximations or stochastic approximations, this work will demonstrate how implicit differentiation can be used to analytically differentiate the overall training loss with respect to the MRF parameters. This framework leads to an efficient, flexible learning algorithm that can be applied to a number of different models. The effectiveness of the proposed learning method will then be demonstrated by learning the parameters of two related models applied to the task of denoising images. The experimental results will demonstrate that the proposed learning algorithm is comparable and, at times, better than previous training methods applied to the same tasks. A new segmentation model will also be introduced and trained using the proposed learning method. The proposed segmentation model is based on an energy minimization framework that is iii novel in how it incorporates priors on the size of the segments in a way that is straightforward to implement. While other methods, such as normalized cuts, tend to produce segmentations of similar sizes, this method is able to overcome that problem and produce more realistic segmentations.

Hypothesis-free learning is increasingly popular given the large amounts of data becoming available. Structure learning, a hypothesis-free approach, of graphical models is a field of growing interest due to the power of such models and lack of domain knowledge when applied on complex real-world data. State-of-the-art techniques improve on scalability of structure learning, which is often characterized by a large problem space. Nonetheless, these techniques still suffer computational bottlenecks that are yet to be approached. In this work, we focus on two popular models: dynamical linear systems and Markov random fields. For each case, we investigate major computational bottlenecks of baseline learning techniques. Next, we propose two frameworks that provide higher scalability using appropriate problem reformulation and efficient structure based heuristics. We perform experiments on synthetic and real data to validate our theoretical analysis. Current results show that we obtain a quality similar to expensive baseline techniques but with higher scalability. / Master of Science

Making full use of spatial information is an important problem in information-processing and decision making. In this dissertation, two Bayesian decision theoretic frameworks for context classification are developed which make full use of spatial information. The first framework is a new multispectral image context classification technique which is based on a recursive algorithm for optimal estimation of the state of a two-dimensional discrete Markov Random Field (MRF). The implementation of the recursive algorithm is a form of dynamic programming. The second framework is based on a stochastic relaxation algorithm and Markov-Gibbs Random Fields. The relaxation algorithm constitutes an optimization using annealing. We also discuss how to estimate the Markov Random Field Model parameters, which is a key problem in using MRF in image processing and pattern recognition. The estimation of transition probabilities in a 2-D MRF is converted into two 1-D estimation problems. Then a Space-varying estimation method for transition probabilities is discussed. / Ph. D.