基于局部二值法的断层检测方法应用研究

Application research of fault detection method based on local binary pattern

利用地震资料检测断层对油气勘探具有重要的意义,也是地震资料解释的重点.文中发展了局部二值模式(Local Binary Pattern,LBP)方法来检测断层,并结合地震剖面的旋转不变局部方差(Rotation Invariant Variance,VAR)信息,提出一种基于LBP/VAR特征的断层检测方法.首先利用相干属性、曲率属性和边缘检测等常规的断层检测方法和LBP/VAR算子对原始地震数据进行断层刻画,说明LBP/VAR算子的有效性,再对LBP、LBP/VAR进行阈值优化处理,使其检测断层的效果更佳.其次用LBP/VAR算子实现对含噪声的理论断层模型以及实际断层图片检测并对比分析.试验结果表明LBP/VAR算子对含噪声的断层理论模型仍然有效,具有一定的鲁棒性.实际断层图片检测证明该方法对断层检测具有更好的效果.研究结果表明LBP/VAR算子具有连续性好、一定的抗干扰能力,能较好的刻画断层,为地震资料断层检测提供了一种有效的检测工具.

Detecting faults with seismic data is of great significance to oil and gas exploration and is also the focus of seismic data interpretation. Although there are many fault detection methods at present, each method has its application limitations. It is necessary to continue to study and develop new fault detection methods. An innovative fault detection method based on LBP/VAR feature is proposed. LBP/VAR method is based on Local Binary Pattern(LBP) and combined with the Rotation-invariant Local Variance(VAR)information of seismic profile. Firstly, the principle of LBP is introduced and the LPB/VAR operator suitable for seismic data is emphatically expounded. Then LBP/VAR operator was used to detect the original seismic data, comparisons with the conventional fault detection methods such as coherence attribute, curvature attribute and edge detection, indicating the effectiveness of LBP/VAR operator. In addition, LBP and LBP/VAR were optimized to make their fault detection effect more precision accuracy. Secondly, the LBP/VAR operator after threshold optimization was used to detect the synthetic fault seismic data established by the forward convolution model, and the detection results were consistent with the model. Thirdly, in order to demonstrate the effectiveness of the LBP/VAR operator to detect faults and robustness to noise, the different proportions of noise are added to synthetic fault seismic data. Even if the edge detection operator depicts blurred faults, the LBP/VAR operator can detect them better. It showed that the LBP/VAR operator had a certain robustness for noise and the more effective in fault detection. The validity of the LBP/VAR operator is also verified by testing with actual geological fault images. Through the analysis of three examples and comparison with the conventional methods, the results showed that the LBP/VAR operator was effective and robust to a certain extent. The LBP/VAR operator provides an effective tool for fault detection of seismic data.