Ciência da Computação
URI permanente desta comunidadehttps://repositorio.fei.edu.br/handle/FEI/342
Navegar
10 resultados
Resultados da Pesquisa
Artigo de evento 1 Citação(ões) na Scopus Surface reconstruction for generating digital models of prosthesis(2011-03-05) DE AQUINO, L. C. M.; LEITE, D. A. T. Q.; GIRALDI, G. A.; CARDOSO, J. S.; Paulo Rodrigues; NEVES, L. A. P.The restoration and recovery of a defective skull can be performed through operative techniques to implant a customized prosthesis. Recently, image processing and surface reconstruction methods have been used for digital prosthesis design. In this paper we present a framework for prosthesis modeling. Firstly, we take the computed tomography (CT) of the skull and perform bone segmentation by thresholding. The obtained binary volume is processed by morphological operators, frame-by-frame, to get the inner and outer boundaries of the bone. These curves are used to initialize a 2D deformable model that generates the prosthesis boundary in each CT frame. In this way, we can fill the prosthesis volume which is the input for a marching cubes technique that computes the digital model of the target geometry. In the experimental results we demonstrate the potential of our technique and compare it with a related one.- Computing the q-index for tsallis nonextensive image segmentation(2009-10-11) Paulo Rodrigues; GIRALDI, G. A.The concept of entropy based on Shannon Theory of Information has been applied in the field of image processing and analysis since the work of T. Pun [1]. This concept is based on the traditional Boltzaman-Gibbs entropy, proposed under the classical thermodynamic. On the other hand, it is well known that this old formalism fails to explain some physical system if they have complex behavior such as long rang interactions and long time memories. Recently, studies in mechanical statistics have proposed a new kind of entropy, called Tsallis entropy (or non-extensive entropy), which has been considered with promising results on several applications in order to explain such phenomena. The main feature of Tsallis entropy is the q-index parameter, which is close related to the degree of system nonextensivity. In 2004 was proposed [2] the first algorithm for image segmentation based on Tsallis entropy. However, the computation of the q-index was already an open problem. On the other hand, in the field of image segmentation it is not an easy task to compare the quality of segmentation results. This is mainly due to the lack of an image ground truth based on human reasoning. In this paper, we propose the first methodology in the field of image segmentation for q-index computation and compare it with other similar approaches using a human based segmentation ground truth. The results suggest that our approach is a forward step for image segmentation algorithms based on Information Theory. © 2009 IEEE.
- Geometric Data Analysis Based on Manifold Learning with Applications for Image Understanding(2017-07-01) MIRANDA, G. F.; Plinio Thomaz Aquino Junior; GIRALDI, G. A.© 2017 IEEE.Nowadays, pattern recognition, computer vision, signal processing and medical image analysis, require the managing of large amount of multidimensional image databases, possibly sampled from nonlinear manifolds. The complex tasks involved in the analysis of such massive data lead to a strong demand for nonlinear methods for dimensionality reduction to achieve efficient representation for information extraction. In this avenue, manifold learning has been applied to embed nonlinear image data in lower dimensional spaces for subsequent analysis. The result allows a geometric interpretation of image spaces with relevant consequences for data topology, computation of image similarity, discriminant analysis/classification tasks and, more recently, for deep learning issues. In this paper, we firstly review Riemannian manifolds that compose the mathematical background in this field. Such background offers the support to set up a data model that embeds usual linear subspace learning and discriminant analysis results in local structures built from samples drawn from some unknown distribution. Afterwards, we discuss topological issues in data preparation for manifold learning algorithms as well as the determination of manifold dimension. Then, we survey dimensionality reduction techniques with particular attention to Riemannian manifold learning. Besides, we discuss the application of concepts in discrete and polyhedral geometry for synthesis and data clustering over the recovered Riemannian manifold with emphasis in face images in the computational experiments. Next, we discuss promising perspectives of manifold learning and related topics for image analysis, classification and relationships with deep learning methods. Specifically, we discuss the application of foliation theory, discriminant analysis and kernel methods in curved spaces. Besides, we take differential geometry in manifolds as a paradigm to discuss deep generative models and metric learning algorithms.
- CAD system for breast US images with speckle noise reduction and bio-inspired segmentation(2019-10-05) RODRIGUES, P. S. S.; Guilherme Lopes; GIRALDI, G. A.; BARCELOS, C. A. Z.; VIEIRA, L.; GULIATO, D.; KUMAR SINGH, B.© 2019 IEEE.Ultrasound (US) images are highly susceptible to speckle-like noise which makes imperative to use specific techniques for image smoothing. However, this process can lead to undesirable side effects such as the degradation of the real contour of the region of interest (ROI). In such context, this paper presents a new methodology for computer aided diagnosis (CAD) systems whose heart is the combination of a method for speckle noise reduction, with histogram equalization and a technique for image segmentation that uses the bio-inspired firefly algorithm and Bayesian model. The segmentation approach and the equalization are applied in two distinct stages: globally and locally. The global application produces an initial coarse estimate of the ROI, and the local application defines this region more precisely. In the classification step we carried out experiments which show that the combination of features computed both within and below the lesion strongly influences the final accuracy. We show that the gray-scale distribution and statistical moments within the lesion together with gray-scale distribution and contrast of the region below the lesion is the combination that produces the better classification results. Experiments in a database of 250 US images of breast anomalies (100 benign and 150 malignant) show that the proposed methodology reaches performance of 95%.
- A strategy based on non-extensive statistics to improve frame-matching algorithms under large viewpoint changes(2019) LOPES, Guilherme; HORVATH, M.; GIRALDI, G. A.; LOPES, Guilherme
- Improving a firefly meta-heuristic for multilevel image segmentation using Tsallis entropy(2015) RODRIGUES, Paulo; LOPES, Guilherme; ERDMANN, H. R.; RIBEIRO, M. P.; GIRALDI, G. A.In this paper we show that the non-extensive Tsallis entropy, when used as kernel in the bio-inspired firefly algorithm for multi-thresholding in image segmentation, is more efficient than using the traditional crossentropy resented in the literature. The firefly algorithm is a swarm-based meta-heuristic, inspired by fireflies-seeking behavior following their luminescence. We show that the use of more convex kernels, as those based on non-extensive entropy, is more effective at 5 % of significance level than the cross-entropy counterpart when applied in synthetic spaces for searching thresholds in global minimum
- Dimensionality reduction, classification and reconstruction problems in statistical learning approaches(2008) GIRALDI, G. A.; RODRIGUES, Paulo; KITANI, E. C.; THOMAZ, C.
- Non-extensive entropy algorithm for multi-region segmentation: generalization and comparison(2013) RODRIGUES, Paulo; GIRALDI, G. A.
- Improving the non-extensive medical image segmentation based on Tsallis entropy(2011) RODRIGUES, Paulo; GIRALDI, G. A.Thresholding techniques for image segmentation is one of the most popular approaches in Computational Vision systems. Recently, M. Albuquerque has proposed a thresholding method (Albuquerque et al. in Pattern Recognit Lett 25:1059–1065, 2004) based on the Tsallis entropy, which is a generalization of the traditional Shannon entropy through the introduction of an entropic parameter q. However, the solution may be very dependent on the q value and the development of an automatic approach to compute a suitable value for q remains also an open problem. In this paper, we propose a generalization of the Tsallis theory in order to improve the non-extensive segmentation method. Specifically, we work out over a suitable property of Tsallis theory, named the pseudo-additive property, which states the formalism to compute the whole entropy from two probability distributions given an unique q value. Our idea is to use the original M. Albuquerque’s algorithm to compute an initial threshold and then update the q value using the ratio of the areas observed in the image histogram for the background and foreground. The proposed technique is less sensitive to the q value and overcomes the M. Albuquerque and k-means algorithms, as we will demonstrate for both ultrasound breast cancer images and synthetic data.
- Statistical learning approaches for discriminant features selection(2008) GIRALDI, G. A.; RODRIGUES, Paulo; KITANI, E. C.; SATO, J. R.; THOMAZ, C. E.