An algorithm for the reconstruction of discrete sets from two projections in presence of absorption (Q2573236)

The paper concerns the reconstruction of an object that emits radiation through absorbing materials. The object is regarded as a binary matrix which has to be reconstructed from its absorbed projections. The authors prove that the left and right horizontal absorbed projections of a binary matrix \(A=(a_{i,j})\) given by \[ L=\left(\sum_{j=1}^na_{i,j}\beta^{-j}\right)_{i=1,\dots,m}, \;R=\left(\sum_{j=1}^na_{i,j}\beta^{-n+j-1}\right)_{i=1,\dots,m}, \] uniquely determine \(A\) when \(\beta=\beta_0=(1+\sqrt5)/2\). Moreover, a fast algorithm for solving the related reconstruction problem is defined and its complexity is analyzed. Finally, a possibility of using different absorption coefficients \(\mu=\ln \beta\) for \(\beta\neq\beta_0\) is discussed. The paper continues investigations presented by \textit{A. Kuba} and \textit{M. Nivat} [Linear Algebra Appl. 339, No. 1--3, 171--194 (2001; Zbl 1004.65056)].