Tecnicas de rastreamento e aplicações em analise cinematica de movimentos humanos

AUTOR(ES)
DATA DE PUBLICAÇÃO

2004

RESUMO

The problem of tracking objects from a sequence of video images has been of great interest in the recent years. This interest is motivated, especially, by the wide range of applications such as virtual reality surveillance systems, robotic, human motion analysis, etc. In the human motion analysis, the kinematic variables, (e.g., the covered distance, velocities and accelerations) are important parameters for clinical evaluation of the subjects with motion disabilities, as well as measurement of motion performance of athletes. Among the different ways of human motion description, the kinematic analysis by videogrammetry represents a wide of aplication due the simplicity of the used devices (video cameras) for their implementation and the variety of problem that can be considered. However, the automatic determination of these parameters is a complex omputational problem because of the great number of aspects that make difficult the identification and extraction of the objects of interest. The aim of this work is to develop algorithms for tracking of objects allowing an automatic determination of their 2D or 3D coordinates, as well as the quantification of the kinematic variables of human motion. In this work, the problems related to tracking are associated with two applications: the clinical analysis of human motion and athlete performance analysis. In the first case, we present algorithms for tracking of markers fixed to the human body in the regions of interest (e.g., in the articulations). This approach takes into account aspect such as placement and quantity of markers. In the second case, we consider the problem of tracking soccer players during a match, taking into consideration the possible changes of illumination, players? occlusion, and the use of multiple cameras covering the whole playing field

ASSUNTO(S)

processamento de imagens biomecanica rastreamento automatico

Documentos Relacionados