Computed tomographic CT reconstuction from a limited number of views results in a null subspace in the space of reconstructed functions. Typical reconstruction algorithms set the null space part of the reconstruction to zero, which results in artifacts. The Bayesian approach provides one way to make use of prior information about the reconstructed scene that can help overcome the lack of data. A maximum a posteriori (MAP) reconstruction algorithm is shown to produce dramatically improved reconstruction compared to the algebraic reconstruction technique (ART).
Keywords: tomographic reconstruction, ill-posed problem, null space, measurement space, maximum a posteriori (MAP) reconstruction, algebraic reconstruction technique (ART)
Return to publication list
Send e-mail to author at kmh@hansonhub.com