Abstract

Iterative convolution, a technique for limited-data tomographic reconstruction, is based on constrained iteration between a source function’s estimated image and its Radon transform. For certain source functions, the estimated image diverges from the source function. This divergence is shown to result from the algorithm’s failure to enforce consistency between the estimated image and its measured Radon transform. To correct for this, consistency can be enforced by using a routine based on the direct-inversion formula and on decomposing the image into components generated from its measured and missing Radon-transform integrals. This routine was used to develop a new iterative scheme that converges absolutely for three test objects for which simple iterative convolution diverges.

© 1989 Optical Society of America

Iterative convolution backprojection algorithms for image reconstruction from limited data

Barry P. Medoff, William R. Brody, Menahem Nassi, and Albert Macovski
J. Opt. Soc. Am. 73(11) 1493-1500 (1983)

Sinogram recovery with the method of convex projections for limited-data reconstruction in computed tomography

Hiroyuki Kudo and Tsuneo Saito
J. Opt. Soc. Am. A 8(7) 1148-1160 (1991)

Tomography by iterative convolution: empirical study and application to interferometry

Charles M. Vest and Ivan Prikryl
Appl. Opt. 23(14) 2433-2440 (1984)

Cone-beam reconstruction by backprojection and filtering

Andrei V. Bronnikov
J. Opt. Soc. Am. A 17(11) 1993-2000 (2000)

Iterative reconstruction algorithms: convergence as a function of spatial frequency

Stephen J. Norton
J. Opt. Soc. Am. A 2(1) 6-13 (1985)