The Simplest Discrete Radon Transform

Pitfalls in the evaluation of the discrete Radon transform are reviewed and methods are identified that overcome the problems. A commonly used discrete form of the Radon integral formula results in two types of artifacts being generated: operator aliasing artifacts and truncation artifacts. Operator aliasing artifacts can be avoided by taking into account the bandlimited nature of the integrand during the conversion of the Radon integral into a discrete summation. Truncation artifacts are due to aperture limitation of the input data (missing offsets) and can be overcome by imposing a minimum entropy constraint in model space that forces the solution to be as simple as possible. Eliminating these two types of artifacts produces a clean, simple, and more highly resolved image in the Radon domain for any application such as multiple elimination, predictive deconvolution or migration.