Fundamentos geometricos e algebricos da calibração e reconstrução tridimensional : aplicação na analise cinematica de movimentos humanos / Geometric and algebraic basis for calibra tion and threedimensional re construcion. application in the kinematic analysis of human movements

AUTOR(ES)
DATA DE PUBLICAÇÃO

1999

RESUMO

To get information such as position, orientation and size of the object of interest by measuring its image are the basis of the kinematic analysis of the human movement. Video cameras are frequently used for this purpose. The aim of this work is to study Geometric and AIgebraic concepts of the procedures used to calibrate the video cameras and of the reconstruction of the three-dimensional coordinates, applied in the kinematic analysis of the human movements. The calibration of the cameras consists basically of establishing the geometric relationship between the object of interest and its image. Once the operator for this relationship is known, in two or more cameras, the space coordinates of the studied object can be achieved from the images through threedimensional reconstruction. This way the object-image, and vice versa, transformations assume a role of prominence in these systems. The projective operator that acts in these transformations is based on the Basic Theorem of Projective Geometry. The algebraic structure ofthis theorem is formulated by the projective equation, from where we get the matrix of the projective operator, whose elements are the calibration parameters of the video cameras. Considering the characteristics of the threedimensional object transformation into bi-dimensional images, the reconstruction of the three-dimensional coordinates of this object needs the images of at least two cameras. At least six control points with known coordinates in the 3D space and in the 2D image are necessary to estimate the calibration parameters of each camera. These control points are used in a system of normal equations and estimated by the least-squares thecnique. The question of the selection of the calibration control points, according to some authors, is still not sufficiently resolved. In order to optimize the selection of the control points we present an algorithm, which was used in an experiment to determine the optimal configuration of the control points reIative to the position of the cameras. The set ofprocedures considered allows us to classify the control points as a function of the contribution that each one brings to the quality ofthe calibration ofthe cameras

ASSUNTO(S)

geometria projetiva processamento de imagens calibração biomecanica

ACESSO AO ARTIGO

Documentos Relacionados