全变分2范数有限差分小波域成像测井图像缺陷修复算法

发布时间：2018-08-07 07:12

【摘要】：针对超声成像测井中因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,文章提出了一种能够有效解决该类问题的全变分2范数有限差分小波域超声测井图像修复算法。首先,设计基于图像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;然后,针对全变分小波域图像修复过程中出现的"分片常数"效应,设计一种全变分2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效果的比对,证明该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。
[Abstract]:In view of the phenomenon that the abnormal zone appears in the logging image due to various uncertain factors in ultrasonic imaging logging, the shortcomings of the conventional methods used in repairing abnormal areas, especially in wavelet domain, are analyzed. In this paper, an ultrasonic image restoration algorithm based on total variational 2-norm finite-difference wavelet domain is proposed, which can effectively solve this kind of problems. First of all, an algorithm of "chain" anomaly area recognition based on the correlation value of adjacent row and absolute error value is designed to locate the abnormal region accurately. Aiming at the "piecewise constant" effect in image restoration process in total variational wavelet domain, a new algorithm of total variational 2-norm finite-difference wavelet domain restoration is designed to eliminate this effect effectively. The comparison of the actual log data processing results with the standard well logging images shows that the algorithm can further improve the repair effect for the abnormal areas of the "chain". It provides an effective solution for the restoration of this kind of condition in the later stage image data.
【作者单位】：武汉理工大学计算机科学与技术学院;武汉理工大学信息工程学院;华北油田测井公司;
【基金】：国家自然科学基金资助项目(41374148;41372155) 湖北省自然科学基金青年基金资助项目(2014CFB248) 湖北省教育厅科技计划资助项目(Q20111306)
【分类号】：P631.81;TP391.41

【相似文献】