Adoption of a genetic algorithm (GA) for tomographic reconstruction of line-of-sight optical images

A new tomographic reconstruction scheme is proposed that uses a genetic algorithm (GA), a robust and combinatorial function optimization based on the mechanics of the genetic principle. The paper first discusses the implicitly parallel and scaled random nature of the GA optimization using an illustrative example. An introduction of the elementary distribution function (EDF) to constitute the crosssectional field shows a successful adoption of GA for optical tomography. The GA-based tomography was examined for interferometric projections of computer-synthesized phantom density fields. The GA-based tomography shows accurate and stable image reconstruction, particularly for limited projections. List of symbols Ai magnification factor of the ith EDF ART algebraic reconstruction technique CT computer tomography D diameter of an axisymmetric test field or diffusion coefficient (Eq. (2a)) EDF elementary distribution function f type of elementary distribution function F fitness value (Eq. (4)) FC Fourier convolutions GA genetic algorithm I number of individuals in population K Gladstone-Dale constant (0.226]10~3 m3/kg at j\632.8 nm for air at 288 K and 101.3 kPa) MRI magnetic resonance imaging n index of refraction or the number of EDFs no index of refraction in vacuum N fringe shift number N R number of children created in each generation p measured projection Received: 6 November 1995 / Accepted: 6 May 1996 K. D. Kihm, H. S. Ko Department of Mechanical Engineering, Texas A&M University College Station, TX 77843-3123, USA K. Okamoto, D. Tsuru Nuclear Engineering Research Laboratory, University of Tokyo Tokai-mura, Ibaraki 319-11, Japan Correspondence to: K. D. Kihm p* virtual projection PIV particle image velocimetry r radial coordinate of axisymmetric test field R coordinate perpendicular to the line-of-sight direction (Fig. 6) RMS root mean square S coordinate parallel to the line-of-sight direction (Fig. 6) ti shape factor of the ith EDF t time x, y rectangular coordinate system defined on the test field (Fig. 6) d Dirac-delta function h projection angle j laser wave length (632.8 nm for helium-neon laser) o gas density