一种全变分二范数有限差分小波域超声成像测井修正算法  

An total variational two norm finite difference wavelet domain inpainting algorithm in ultrasonic imaging logging

作　　者：张健[1,2] 王一举[1] 余华平[1] 

机构地区：[1]长江大学计算机科学学院,湖北荆州434023 [2]武汉理工大学光纤传感技术国家工程实验室,武汉430070

出　　处：《中国科技论文》2015年第20期2457-2462,共6页China Sciencepaper

基　　金：国家自然科学基金资助项目(41374148);湖北省教育厅科技计划项目(Q20111306)

摘　　要：针对超声成像测井中由于各种不确定因素而导致测井图像中出现带状异常区域的现象,通过分析目前修复异常区域所采用的常规方法,尤其是小波域修复方法的不足,提出一种能够有效解决该类问题的全变分二范数有限差分小波域超声测井图像修正算法。首先,设计了基于图像基于相邻行相关值与绝对误差值相结合的"链条"异常区域自动识别算法,实现对异常区域的准确定位。随后针对全变分小波域图像修复过程中出现的"分片常数"效应的影响,设计了一种全变分二范数有限差分小波域修复算法来有效消除该影响。通过实际测井资料处理结果与标准测井图像进行的效果比对证明,所提算法进一步提升了针对"链条"异常区域的修复效果,为后期图像资料中该类状况的修复提供了有效的解决方案。Aiming at the phenomenon that in the process of ultrasonic image logging,some zonal anomaly area will appear in the images due to several uncertain factors,this paper analyzes the inefficiency of normal ways including wavelet domain restoration way especially to clear away the zonal anomaly area so as to put forward an effective total-variation norm two finite difference ultrasonic logging images in wavelet domain inpainting algorithm to solve this kind of issue.Firstly,a kind of automatic recognition algorithm about chain abnormal region based on adjacent rows and absolute error value combination was proposed,which realize the accurate location of imaging zonal region effectively.Then,Aiming at the total-variation wavelet domain image restoration in the presence of ＂piecewise constant＂ effect,for which the total variational norm two finite difference ultrasonic logging images in wavelet domain inpainting algorithm was proposed to eliminate the impact effectively.By comparing the experimental results in the actual logging image data processing with the standard logging image data respectively.It shows that this algorithm improves the restorative effects further and provides an effective solution for the inpainting and management of such circumstances when similar problems appear in image materials in the later stage.

关 键 词：图像处理 全变分 二范数 有限差分 小波域 成像修正 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]