1. Inicio
  2. Atrás
Título: World modeling in radar : a regularization by synthesis
Autores: Maluf, David Ameen
Fecha: 1995
Publicador: McGill University - MCGILL
Fuente:
Tipo: Electronic Thesis or Dissertation
Tema: Engineering, Electronics and Electrical.
Descripción: In Radio Detection and Ranging, the inverse problem is that of acquiring knowledge of the physical features of a body by making observations of the reflected energy and synthesising the model from the measured data. This procedure is in contrast to the forward problem, which consists of calculating the observable effects from a given model. The forward problem has unique solution whereas the inverse process, being carried out on the basis of hypotheses, is always characterized by a lack of uniqueness.
The approach taken towards developing the synthesis framework is to consider a logic argumentation of belief functions which is used to represent the various symbolic aspects of belief and uncertainty. The logic extends so that not just one argument, but all arguments, supporting or opposing a hypothesis are considered. That is the logic used to solve the inverse problem. As arguments are identified among measurements, the support they confer on a hypothesis or its negation is aggregated to provide a measure of the degree of belief in the hypotheses of interest. The aggregation operation, or the synthesis regularization, will depend on an entropy calculation to represent the uncertainty associated with the arguments.
Based on the theory of Kalman filters, sensor fusion is used to finally establish probabilistic models of the hypotheses. In conjunction with the synthesis regularization, consistent estimates will converge to a qualitative image reconstruction. The synthesis framework is compared to current solutions to the inverse problem in radio detection and ranging and applied to Ground Penetrating Radar image reconstruction.
Idioma: en