一种多小波电容层析成像图像融合方法  

Image Fusion Method Based on Multi Wavelet Transform for Electrical Capacitance Tomography

作　　者：姚玉梅[1] 陈德运[1] 林甲楠 宋蕾[1] 王莉莉[1] 

机构地区：[1]哈尔滨理工大学计算机科学与技术学院,黑龙江哈尔滨150080

出　　处：《哈尔滨理工大学学报》2014年第5期88-93,共6页Journal of Harbin University of Science and Technology

基　　金：国家自然科学基金(60572153;60972127);高等学校博士学科点专项科研基金(200802140001);黑龙江省自然科学基金(QC2012C059);黑龙江省教育厅计划项目(11541040;12511097)

摘　　要：针对电容层析成像系统的"软场"效应和病态问题对重建图像精度的影响,在分析电容层析成像基本原理和成像算法的基础上,基于滤波LBP和神经网络RBF特性,提出了一种基于多小波变换的ECT图像融合方法.该方法首先对待融合图像进行多小波分解,再分别采用最大值法和区域梯度法融合规则对分解后图像低频部分和高频部分进行图像的融合,最后经过多小波重构得到最终融合图像.仿真实验结果表明,融合后成像精确度得到明显提高,减少了误差,图像更接近原型,为ECT图像重建的研究提供了一个新的方法.On the basis of analyzing ECT basic principle and imaging algorithm,based on the characteristic of filter LBP and neural network RBF,an image fusion method based onmultiwavelet transform is proposed about the effect of ＂soft-field＂ nature and ill-posed problem on the reconstructed image accuracy.This method first applies multiwavelet to decompose the reconstructed images,then fuses the images through the integration of low frequency coefficients and high frequency coefficients based on fusion rules of maxinum and regional gradient.Finally,the fused image is reconstructed by multiwavelet.The simulation results show that image accuracy significantly improved and removed redundant information after integration.This study provided a new method on ECT image reconstruction to make image closer to the prototype.

关 键 词：电容层析成像 图像重建 图像融合 多小波变换 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]