ISSN Online: 2377-424X
ISBN Print: 1-56032-797-9
International Heat Transfer Conference 11
TOMOGRAPHIC RECONSTRUCTION OF INTERFEROMETRIC PROJECTIONS: ALGEBRAIC RECONSTRUCTION (ART) VERSUS GENETIC ALGORITHM (GA)
要約
The quality of tomographically reconstructed images of
interferometric projections is compared between the
conventional algebraic reconstruction technique (ART),
and the new genetic algorithm (GA). The examination uses
two different computer synthesized phantom density fields.
Each phantom is interferometrically projected at three or
five equally angled directions. To simulate experimental
uncertainties, random noise levels are imposed on the
projected data before they are used for reconstruction.
When no noise is imposed, the linear ART optimization
provides faster and more accurate reconstruction than the
nonlinear GA-based method. However, when the random
noise is imposed on the projection data, the GA
reconstruction accuracy gradually decreases while the ART
reconstruction accuracy rapidly decreases with increasing
noise levels. Within the model data and range of noise
evaluated in our study, ART accuracy generally exceeded
that for GA. However, the GA method also provided
robust reconstructions and may offer advantages at higher
noise levels or for applications unsuited to ART, such as
non-linear and nonalgebraic projections.