Enhancement of precision and reduction of measuring points in tomographic reconstructions

Accurate external measurements are required in tomographic problems to obtain a reasonable knowledge of the internal structures. Crucial is the distribution of the external measuring points. We suggest a procedure how to systematically optimize this distribution viz. to increase the precision (i.e. to shrink error bars) of the reconstruction by detecting the important and by eliminating the irrelevant measuring points. In a realistic numerical example we apply our scheme to magnetotomography of fuel cells. The result is striking: Starting from a smooth distribution of measuring points on a surface of a cuboid around the fuel cell, the number of measuring points can systematically be reduced by more than 90%. At the same time the precision increases by a factor of nearly 3.