放射CT重建的平均场技术

Mean Field Annealing for ECT Reconstruction

本文引用了一种图像分布的先验模型，在这种先验模型中引入了两个边界模型，称为组合马尔柯夫随机场模型，用Ｂａｙｅｓ定理将其与放射ＣＴ（ＥＣＴ）问题相结合构造出后验概率，由于这个最大后验估计问题是一个非凸问题，本文采用平均场退火技术对其进行分析，将统计松弛方程转化为一组确定性迭代式，计算机进行的模拟实验证实了这种算法的有效性。

In this paper, a priori model of images, called Compound Markov Random Field (CMRF), is proposed. The CMRF is combined with the ECT reconstruction. which can be indicated as maximization of a likelibood function, and construct a bayesian estimation problem. We find the problem is no longer a convex problem. A mean field annealing relaxation equation is converted into a determinstic iterative algorithm. Simulations are done on a person computer to demonstrate the validity of the algorithm.