“Data fusion best affine unbiased estimation of a deterministic vector, applications to image fusion”

GUSTAVO A. ROITMAN; BRUNO CERNUSCHI FRÍAS

Buenos Aires

Congreso; XXII Congreso Argentino de Control Automático”, AADECA 2010, de la Asociación Argentina de Control Automático, AADECA, Buenos Aires, 31 de Agosto al 3 de Septiembre de 2010.; 2010

Asociación Argentina de Control Automático, AADECA.

In this work we deal with centralized data fusion by comparing two differentapproaches. The first approach uses the minimum square error criterion to find an affineunbiased fusion rule (BAUE). The second approach is that of Maximum Likelihood, statedunder non independent correlated samples, which performance is contrasted to that of BAUEfusion (for Gaussian measurement noise) in terms of the Cramer-Rao lower bound and simulations.Both approaches suppose that the estimated vector is deterministic. The developedfusion rules are then suited to an image fusion example