摘要
CT技术图象重建方法可分为变换法和代数法两大类。本文讨论了几种代数重建法的特点,开展了地球物理CT技术在岩石实验中应用的探讨性工作。研究表明,实验数据的测量误差直接影响着反演结果,当测量误差比较大时,即使数据量很大,也难以提高反演结果的分辨。岩样的吸收、换能器与样品的耦合、岩样均匀性等因素对实验数据的测量精度都有着重要的影响。文中还针对实验条件进行了数值模拟,以检验本方案的可行性,并对当前流行的BPT、ART和SIRT算法及其组合算法的成象能力结合本实验的条件进行了性能的比较。结果表明:BPT方法计算简单,但所给出的图象偏粗糙,特别是在井间观测时水平方向的分辨率较差;ART方法收敛速度快,图象的反差大,反演结果的高频成分比较丰富,但边界异常大;SIRT方法收敛速度慢,图象的低频比较好,结构平滑,边界异常小,反演受测量误差的影响比较小。综合了ART和SIRT两种反演方法特点的ART-SIRT联合反演方法比较理想,发挥出了各自的长处。
The computerized tomographical reconstruction methods include transform method and algebraic iteration one. In this paper, the algebruction methods include transform method and ART method has quick convergence and largeock experiments. BPT gives only a rough image with poor resolution in horizontal direction. ART method has quirk convergence and large image contrast, but the boundary effect is high, and time-consuming in inversion computation. Though SIRT method converges slowly and provides a low image contrast, it has less boundary effect and is fast in computation. No serious effect comes from measurement error during image reconstruction. The combined method of ART-SIRT inversion assimilates the advantages of above two methods and always leads to a satisfactory result. The present study indicates that the measurement error in experiment should be highly taken into account in image reconstruction. CT inversion requires huge amount of accurate observational data, which means that poor data will llead to worse result. The attenuation of the sample, the coupling of the sensor with the sample and homogeneity of the sample are key facors effecting the accuracy of the observational data.
出处
《地球物理学报》
SCIE
EI
CAS
CSCD
北大核心
1989年第4期409-416,共8页
Chinese Journal of Geophysics
基金
地震学联合科学基金会
关键词
岩样
包体构造
CT技术
岩石实验
Computerized tomography, Rock experiment, Algebraic reconstruction technique, Crossborehole observation.
