Image segmentation is one of the most important tasks in image analysis. This task can be used on an extensive range of application. However, some traditional techniques exhibit high computational costs, hindering their application because the complexity of such approaches is intrinsically related to the nature of the image. This work proposal explores a new area of application of the theory of complex networks until then little explored, the analysis of images. It is a relatively recent area of study that has attracted the attention of the scientific community and has been successfully applied to a large scale in the analysis of real-world problems.
André Pilastri, João Papa, João Tavares
In USNCCM, July 28-August 1,
2019
Studies of complex networks have been an important topic of interest to many researchers, in part due to its potential for a simple representation of complex systems in various fields of science. This work aims to develop and study some methods used for the characterization of complex networks. Thus, a new approach to the dermatoscopic image analysis that combines extraction of superpixels and detection of communities of a complex network is expected to be developed. To reduce the computational cost, a SLIC preprocessing algorithm will be used to group several pixels of the image into a uniform region (superpixels), which will decrease the size of the network and, consequently, minimize the computational cost of the cluster.
André Pilastri, João Papa, João Tavares
In CMBBE,
2018
In this paper, the study and application of data analysis techniques for extracting information is proposed. The contribution of this work targets the process of identification of relevant literature from a collection of crawled documents. Novel functions, called social network features, are described and evaluated on documents crawled on ArXiv, to examine their relevance. The results highlight the data analysis process and the performance of the classification of the data mining algorithms used.
André Pilastri, Daniel F. Nogueira, Danilo Samuel Jodas
In RECPAD,
2017
The theory Compressive Sensing (CS) has provided a new acquisition strategy and recovery with good in the image processing area. This theory guarantees to recover a signal with high probability from a reduced sampling rate below the Nyquist-Shannon limit. The problem of recovering the original signal from the samples consists of solving an optimization problem. This article presents an overview of reconstruction algorithms for sparse signal recovery in CS, these algorithms may be broadly divided into six types. We have provided a comprehensive survey of the numerous reconstruction algorithms in CS aiming to achieve computational efficiency.
André Pilastri, João Tavares
In DSIE|16,
2016
In the pattern recognition research field, Support Vector Machines (SVM) have been an effectiveness tool for classification purposes, being successively employed in many applications. The SVM input data is transformed into a high dimensional space using some kernel functions where linear separation is more likely. However, there are some computational drawbacks associated to SVM. One of them is the computational burden required to find out the more adequate parameters for the kernel mapping considering each non-linearly separable input data space, which reflects the performance of SVM. This paper introduces the Polynomial-Powers of Sigmoid for SVM kernel mapping, and it shows their advantages over well-known kernel functions using real and synthetic datasets.
S. E. N. Fernandes, A. L. Pilastri, L. A. M. Pereira, R. G. Pires, J. P. Papa
In SIBGRAPI,
2014