抄録
Recently, Maximum A Posteriori estimation using Expectation Maximization algorithm ( MAP EM )has been widely studied in the image reconstruction of Emission Computed Tomography (ECT). Green (1990) and Lange (1990) proposed the MAP EM methods using the prior based on the Gibbs distribution which assumed a locally correlated Markov Random Field ( MRF). In the MAP EM method, we have to decide the Gibbs parameter which has an influence on quality of reconstructed images. However, it is hard to decide the optimal Gibbs parameter uniquely for a given acquired data set. In this paper, we made a comparative study for the Gibbs parameter about the number of projections and the number of pixels in reconstructed images using the MAP EM method.
