• JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
 
  Bookmark and Share
 
 
Master's Dissertation
DOI
https://doi.org/10.11606/D.45.2010.tde-20230727-113611
Document
Author
Full name
Thiago Meireles Paixão
E-mail
Institute/School/College
Knowledge Area
Date of Defense
Published
São Paulo, 2010
Supervisor
Title in Portuguese
Rastreamento de objetos utilizando reconhecimento estrutural de padrões: uma abordagem com modelo estático
Keywords in Portuguese
Reconhecimento De Padrões
Vídeo
Visão Computacional
Abstract in Portuguese
Rastreamento de objetos e um problema amplamente conhecido em visao computacional cujo objetivo e localizar um objeto no decorrer de uma sequencia de video, uma das principais tarefas em aplicaçoes de video digital [1]. Dentre tais apliaçoes de video digital. Dentre tais aplicacçoes, podemos citar sistemas de monitoreamento de trafego, analise de videos esportivos, vigilancia automatizada e interaçao humano-computador.Em um projeto de mestrado, Graciano e Cesar [2] propuseram uma metodologia de rastreamento baseada em reconhecimento estrutural de padroes. Em sintese, o rastreamento e produto do reconhecimento das partes do objeto que pode assumir diferentes configuraçoes espaciais ao longo do tempo. A soluçao original para tratar da dinamica espacial do objeto baseia-se na atualizaçao dos atributos espaciais do modelo em cada interaçao do processo de rastreamento, o que induz rapidamente à degeneraçao do modelo. Neste trabalho propomos uma abordagem com modelo estatico para lidar com a dinamica espacial do objeto e evitar a degeneraçao do modelo. Adicionalmente, propomos modificaçoes que visam simplificar a interaçao com o usuario na geraçao do modelo, ampliar o poder de representaçao do modelo incluindo o fundo, tratar casos simples de rastreamento (sem oclusao) na presença de varios objetos, e reduzir o custo computacional do processo de rastreamento.
Title in English
not available
Abstract in English
Object tracking is a widely known topic in the field of computer vision which aims at locating an object troughout a video sequence. it may be considered one of the main tasks embedded in digital video application such as traffic monitoring systems, sport video analysis, automatic surveillance, and huma-computer interaction. Graciano and Cesar [2] proposed an object tracking methodology based on structural pattern recognition as a master project. In summary, tracking is achieved by object parts recognition based on a user-aided generated model of such object. The model is a static description of the object whose spatial configuration is expected to change over time. The original solution to address the object spatial dynamics is based ona model udate scheme of its spatial attributes in each tracking process iteration, which induces quickly degeneration of the model. We propose in this work a static model approach to deal with object spatial dynamics to avoid the model degeneration process. Furthermore, we also propose modifications in order to simplify the user interaction process i model generation, extent the power of representation of the model to include the background, deal with simple situations of tracking (without oclusion) in scenes with multiple objects, and finally reduce computational cost of the tracking process.
 
WARNING - Viewing this document is conditioned on your acceptance of the following terms of use:
This document is only for private use for research and teaching activities. Reproduction for commercial use is forbidden. This rights cover the whole data about this document as well as its contents. Any uses or copies of this document in whole or in part must include the author's name.
Publishing Date
2023-07-27
 
WARNING: Learn what derived works are clicking here.
All rights of the thesis/dissertation are from the authors
CeTI-SC/STI
Digital Library of Theses and Dissertations of USP. Copyright © 2001-2024. All rights reserved.