Multilevel Image Reconstruction with Natural Pixels

Van Emden Henson, Mark A. Limber, Stephen F. McCormick, Bruce T. Robinson

SIAM Journal on Scientific Computing · 1996 · 19 citations · 10 references

Concepts

Abstract

The sampled Radon transform of a two-dimensional (2D) function can be represented as a continuous linear map $A:L_2 (\Omega ) \to {\bf R}^N $, where $(Au)_j = (u,\psi _j )$ and $\psi _j $ is the characteristic function of a strip through $\Omega $ approximating the set of line integrals in the sample. The image reconstruction problem is: given a vector ${\bf b} \in {\bf R}^N $,find an image (or density function) $u(x,y)$ such that $Au = {\bf b}$. In general there are infinitely many solutions; we seek the solution with minimal 2-norm, which leads to a matrix equation $B{\bf w} = {\bf b}$, where B is a square dense matrix with several convenient properties. We analyze the use of Gauss–Seidel iteration applied to the problem, observing that while the iteration formally converges, there exists a near null space into which the error vectors migrate, after which the iteration stalls. The null space and near null space of B are characterized in order to develop a multilevel scheme. Based on the principles of the multilevel projection method (PML), this scheme leads to somewhat improved performance. Its primary utility, however, is that it facilitates the development of a PML-based method for spotlight tomography, that is, local grid refinement over a portion of the image in which features of interest can be resolved at finer scale than is possible globally.

References

10