BDBComp
Parceria:
SBC
Dempster-Shafer Theory as an Inference Method for Corresponding Geometric Distorted Images

José Demisio Simões da SilvaPaulo Ouvera Simoni

This paper presents further results of the application of DempsterShafer Theory for uncertainty reasoning as a computation model to correspond distorted images in Computer Vision. In (Silva and Simoni, 2002), the model was applied to correspond images presenting differences in brightness and contrast, as an extension of the work in (Silva and Simoni, 2001). The results showed the effectiveness and robustness of the method in corresponding non-equalized pairs of images. In new experiments the model is applied to distorted images, that is, images in which one is rotated in relation to the other. The right image was rotated in 7 different angles (0.5; 1; 1.5; 2; 2.5; 3; and 3.5 degrees). The model was applied to correspond only point in the images and it successfully established the correspondence for 5 rotated images. The correspondence is based on contextual and structural features of a point, treated as corresponding evidences whose combination is performed by a modified version of Dempster´s rule of combination. A search process maximizes the Belief on the combined evidences. Clique no link abaixo para buscar o texto completo deste trabalho na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato: bdbcomp@lbd.dcc.ufmg.br
     Mantida por:
LBD