Projection allocation for image reconstruction

The author presents a quantitative appraisal of image reconstruction as a function of image field smoothness (spatial correlation), image area, and the number of available projections of a field with predominant Gauss-Markov statistics. These parameters are used in the derivation of a projection allocation algorithm that determines, a priori, an optimal set of projections for reconstructing an image ensemble so that the average reconstruction distortion is minimized over the ensemble. Results of performance tests of the algorithm are presented, and an expression for the associated lower bound of reconstruction distortion is given. Conditions for separability of smoothness and area dependence of the algorithm are established. It is noted that the concept of projection allocation is of particular benefit to tomographic-based data compression algorithms.<
>