基于分形理论探究碳酸盐岩CT图像二值化最佳阈值

发布时间：2018-01-11 07:19

  本文关键词：基于分形理论探究碳酸盐岩CT图像二值化最佳阈值　出处：《石油地球物理勘探》2017年05期 　论文类型：期刊论文

  更多相关文章： 碳酸盐岩 微观孔隙 CT图像 分形 孔隙度 最佳阈值

【摘要】：CT扫描构建数字岩心的一个重要环节是灰度图像的二值化。本文基于图像处理分析和分形理论计算孔渗参数,为CT图像二值化提供了更准确、更适用的约束条件。孔洞型碳酸盐岩不同形态和尺度的微观孔隙累积数量与半径分布遵循幂律关系,具有统计意义上的分形特征,且分维值与孔隙度之间存在非线性定量关系。通过自定义四连通像素值梯度突变算法与分形理论相结合的方法,统计分析了孔隙形态、数量等参数并计算了孔隙度,其平均误差小于7.2%;改进"先二值化后边缘识别"的常规方法,运用滤波降噪和边缘识别算子识别标定CT灰度图像微观孔隙,既保证了精度又提高了效率。根据这两种图像处理技术可确定最佳灰度阈值,实现CT图像二值化。该方法处理的二值图像较好地保留了不同尺度微观孔隙结构的形态和分布特征,因此可广泛应用于常规岩心重构及后续的三维数字岩心的构建。
[Abstract]:The binarization of gray image is an important part of constructing digital core by CT scanning. Based on image processing analysis and fractal theory, the calculation of pore and osmotic parameters provides a more accurate method for the binarization of CT images. The microcosmic pore accumulative quantity and radius distribution of pore type carbonate rock in different shape and scale follow the power law relation and have the fractal characteristic in statistical sense. And there is a nonlinear quantitative relationship between fractal dimension and porosity. The pore morphology is statistically analyzed by the combination of self-defined four-connected pixel value gradient mutation algorithm and fractal theory. The average error of porosity is less than 7.2. The conventional method of "first binarization and then edge recognition" is improved. The filter noise reduction and edge recognition operator are used to identify and calibrate the micro pores of CT gray images. It not only ensures the accuracy but also improves the efficiency. According to these two image processing techniques, the optimal gray threshold can be determined. The binarization of CT images is realized. The binary images processed by this method can preserve the morphology and distribution characteristics of micro-pore structures of different scales. Therefore, it can be widely used in conventional core reconstruction and subsequent three-dimensional digital core construction.
【作者单位】： 中国科学院大学渗流流体力学研究所;中国石油勘探开发研究院廊坊分院;兰州工业学院电气工程学院;
【基金】：国家重大科技专项(2011ZX05013-002)资助
【分类号】：P618.13
【正文快照】： *河北省廊坊市44号信箱,065007。Email:guomwu@163.com本项研究受国家重大科技专项(2011ZX05013-002)资助。吴国铭,李熙U，

本文编号：1408596