基于D-S证据理论的CT/ECT图像融合实验方法  

An experimental method of CT/ECT image fusion based on D-S evidence theory

作　　者：司志宁 田沛[1] 张立峰[1] 

机构地区：[1]华北电力大学自动化系,河北保定071003

出　　处：《中国测试》2018年第1期108-112,共5页China Measurement & Test

基　　金：国家自然科学基金项目(51306058)

摘　　要：为提高多相流过程层析成像系统重建图像质量,分析CT、ECT成像机理及单一模态的局限性,提出一种基于D-S证据理论的CT/ECT图像融合实验方法。首先,将CT、ECT待融合图像的像素值聚类化,运用模糊聚类算法中的隶属度函数表示证据理论中的基本概率赋值函数;并通过D-S合成规则求解融合图像基本概率赋值函数,即融合图像隶属度函数。最后由融合图像隶属度函数反推融合图像像素值,实现"像素-概率-像素"之间的转化。实验结果表明:该方法可有效提高单模态重建图像质量,并优于传统像素融合算法。In order to improve the image quality of muhiphase flow process tomography system and analyze the limitation of CT and ECT imaging mechanism and single modality, a new experimental method of CT/ECT image fusion based on D-S evidence theory is proposed. Firstly, the pixel values of CT and ECT image to be fused are clustered, and the membership function in fuzzy clustering algorithm is used to represent the basic probability assignment function in evidence theory. Then, the basic probability assignment function of fusion image is solved by D-S composition rule; Finally, the pixel values of the image are inversely converged by the membership function of the fusion image. The transformation between ＂pixel-probability-pixel＂ is realized. Experimental results show that this method can effectively improve the single-mode imaging quality and is superior to other traditional pixel fusion algorithms.

关 键 词：过程层析成像 图像融合 D—S证据理论 基本概率赋值函数 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]