Abstract

In many physical experiments, the observations r(m,n) can be represented as the convolution of a model p(m,n) with a source function s(m,n) subject to additive noise e(m,n). The ultimate goal of deconvolution schemes^1,2 is to recover both the model and the source function using observations and all other available information. Let d(m,n) = p(m,n) * s(m,n). The signal-to-noise ratio α is defined as α = Σ Σ d²(m,n)/Σ Σ e²(m,n).

© 1983 Optical Society of America