Catálogo de publicaciones - libros

A new model to compute similarity is presented. The representation of a 3D object is reviewed; sequence of vertices and index of vertices are the basic information about the of any 3D object. A linear function called is introduced to create a new sequence or time series from a 3D object. A method to create 3D objects is also described. Experimental results show viability to compute similarity among 3D objects using the extracted sequences and the Dynamic Time Warping algorithm.

In this paper we explore the use of complex numbers as means of representing angular statistics for surface normal data. Our aim is to use the representation to construct a statistical model that can be used to describe the variations in fields of surface normals. We focus on the problem of representing facial shape. The fields of surface normals used to train the model are furnished by range images. We compare the complex representation with one based on angles, and demonstrate the advantages of the new method. Once trained, we illustrate how the model can be fitted to brightness images by searching for the set of parameters that both satisfy Lambert’s law and minimize the integrability error.

Virtual colonoscopy is a non-invasive method for diagnosing colon diseases such as diverticulosis and cancer using digitized tomographic images to produce 3D images of the colon. In virtual colonoscopy, it is crucial to generate the camera path rapidly and accurately for an efficient examination. Most of the existing path-generation methods are computationally expensive since they require preliminary data structures and the 3D positions of all path points should be calculated. In this paper, we propose an automated pathgeneration method that secures visibility by emulating ray propagation through the colon conduit. The proposed method does not require any preliminary data preprocessing steps, which takes several minutes and it also dramatically reduces the number of points needed to represent the camera path. The experimental result is a perceivable increase in computational efficiency and a simpler approach to colon navigation. The proposed method can also be used in other applications that require efficient virtual navigation.

Edition is an important and useful task in supervised classification specifically for instance-based classifiers because edition discards from the training set those useless or harmful objects for the classification accuracy and it helps to reduce the size of the original training sample and to increase both the classification speed and accuracy. In this paper, we propose two edition schemes that combine edition methods and sequential search for instance selection. In addition, we present an empirical comparison between these schemes and some other edition methods.

The conceptual k-means algorithm consists of two steps. In the first step the clusters are obtained (aggregation step) and in the second one the concepts or properties for those clusters are generated (characterization step). We consider the conceptual k-means management of mixed, qualitative and quantitative, features is inappropriate. Therefore, in this paper, a new conceptual k-means algorithm using similarity functions is proposed. In the aggregation step we propose to use a different clustering strategy, which allows working in a more natural way with object descriptions in terms of quantitative and qualitative features. In addition, an improvement of the characterization step and a new quality measure for the generated concepts are presented. Some results obtained after applying both, the original and the modified algorithms on different databases are shown. Also, they are compared using the proposed quality measure.

In this paper we present an image segmentation technique based on the concepts of circulation and topological control. Circulation is a mathematical tool widely used for engineering problems, but still little explored in the field of image processing. On the other hand, by controlling the topology it is possible to dictate the number of regions in the segmentation process. If we take very small regions as noise, the mechanism can be seen as an efficient means for noise reduction. This has motivated us to combine both mathematical tool in our algorithm. As a result, we obtained an automatic segmentation algorithm that generates segmented regions bounded by simple closed curves; a desireable characteristic in many applications.

In this paper we present a novel approach for microcalcification detection in Digital Breast Tomosynthesis (DBT) datasets. A reconstruction-independent approach, working directly on the projected views, is proposed. Wavelet filter responses on the projections are thresholded and combined to obtain candidate microcalcifications. For each candidate, we create a fuzzy contour through a multi-level thresholding process. We introduce a fuzzy set definition for the class microcalcification contour that allows the computation of fuzzy membership values for each candidate contour. Then, an aggregation operator is presented that combines information over the complete set of projected views, resulting in 3D fuzzy particles. A final decision is made taking into account information acquired over a range of successive processing steps. A clinical example is provided that illustrates our approach. DBT still being a new modality, a similar published approach is not available for comparison and limited clinical data currently prevents a clinical evaluation of the algorithm. .

In this paper we introduce a simple method for the detection of hard cuts using only interframe differences. The method is inspired in the computational gestalt theory. The key idea in this theory is to define a meaningful event as large deviation from the expected background process. That is, an event that has little probability to occur given a probabilistic background model. In our case we will define a hard cut when the interframe differences have little probability to be produced by a given model of interframe differences of non-cut frames. Since we only use interframe differences, there is no need to perform motion estimation, or other type of processing, and the method turns to be very simple with low computational cost. The proposed method outperforms similar methods proposed in the literature.

The Chagas disease or affects between 16 and 18 million people in endemic areas. This disease affects the beating rate of infected patients’ cardiomyocytes. At the Molecular Biology of Chagas Disease Laboratory in Argentina the effect of isolated patient’s serum antibodies is studied over rat cardiomyocyte cultures. In this work an image processing application to measure the beating rate of this culture over video sequences is presented. This work is organized as follows. Firstly, a preliminary analysis of the problem is introduced, isolating the main characteristics of the problem. Secondly, a Monte Carlo experiment is designed and used to evaluate the robustness and validity of the algorithm. Finally, an algorithm of order (( log + )) for tracking cardiomyocyte membranes is presented, where is the number of frames and is the maximum area of the membrane. Its performance is compared against the standard beating rate measure method.

Recent advances in digital processing of biological signals have made it possible to incorporate more extensive signals, generating a large number of features that must be analyzed to carry out the detection, and thereby acting against the performance of the detection methods. This paper introduces a simple feature reduction method based on correlation that allows the incorporation of very extensive signals to the new biological signal detection algorithms. To test the proposed technique, it was applied to the detection of Functional Dyspepsia (FD) from the EGG signal, which is one of the most extensive signals in clinical medicine. After applying the proposed reduction to the wavelet transform coefficients extracted from the EGG signal, a neuronal network was used as a classifier for the wavelet transform coefficients obtained from the EGG traces. The results of the classifier achieved sensitivity, and specificity for a universe of 56 patients studied.