二值纹理图象重建的平均场技术

Reconstructing Binary Textured Images by Mean Field Technique

根据平均场神经网络，本文提出了一种二值纹理图象重建算法。纹理图象用马尔可夫随机场模型描述，重建算法是一个自适应过程，递归地进行模型参数估计和图象重建，平均场方法将图象搜索中基于模拟降温的统计松驰方程转化为一组确定性方程，从而有效地提高了计算效率。计算机实验结果证实了重建算法的有效性。

In this Paper,we present a reconstruction method for binary textured images using a mean field technique. The textured image is modeled as a realization of Markov random fields. The reconstruction algorithm is an adaptive process that recursively performs the parameter estimation and the image reconstruction.The mean field method converts the stochastic equation characterized by the simulated annealing into the deterministic equation,and thus efficiently reduces the computational time. A set of simulations are given to show its performance.