Image Reconstruction and Segmentation from Noisy Projections

Abstract

The minimal message length (MML) approach for inductive inference has been effectively used to picture segmentation using Markov random fields as models (MRF). This technique has been expanded to be capable of concurrently reconstructing and segmenting pictures viewed exclusively through noisy projections. The amount of noise added to each projection is determined by the types of pixels (material) that it passes through. The intended use is in low-dose (low-flux) X-ray computed tomography (CT) with irregular projections.