Reducing Computational Burden in the Gradient Calculation of a Time-Domain Microwave Image Reconstruction Algorithm

Abstract

Microwave image reconstruction is usually made with very computationally demanding algorithms. In this work a time domain image reconstruction algorithm is investigated. With this algorithm the memory requirement for the gradient calculation easily can expand beyond what can be fitted into the RAM memory of a computer. As a consequence, the hard drive has to be used for intermediate storage of computation data. This has serious implications on the execution time of the reconstruction algorithm. In this work we investigate what effect a reduction of the number of data points in the gradient calculation has on the reconstruction accuracy. The results show that significant reduction is possible without apparent effect on the reconstructed image.