Filtered back projection method for inversion of incomplete tomographic data (Q1195622): Difference between revisions

Let \(Rf\) be the Radon transform of \(f\in L_ 2(| x|<a)\). The problem is to recover \(f\) from a knowledge of \(Rf(p,\alpha)\) for \(\alpha\in S_ +\cup S_ -\) where \(S_ +\subseteq S^{n-1}\) and \(S_ -=-S_ +\). Starting out from the formula \(W*f=R^ \#(w*Rf)\), \(W=R^ \# w\), where \(R^ \#\) is the backprojection operator, the authors give a filtered backprojection algorithm by choosing an angle dependent convolving function \(w\). They point out that there will be numerical difficulties due to large oscillations in \(w\).