检测地震不连续性结构的多分辨局部结构熵算法

根据地质构造的变化特点和地震数据的时间-尺度特性,提出了基于多尺度瞬时相位的多分辨率局部结构熵算法.首先在小波域的多尺度瞬时相位体上,利用局部结构熵检测不同分辨率下的局部不连续性结构,然后利用图像融合中的像素灰度极大值法,将不同分辨率下的局部结构熵值进行融合,得到一个最终的不连续性检测结果.在具体的算法实现中,利用螺旋坐标系对分析单元内的地震道进行排列组合,构建 4 阶协方差矩阵,减少了计算量.模型和实际资料算例的计算结果表明,与基于特征结构的相干算法相比,该方法既可以清晰地显示出大的结构突变,又能加强微弱、缓变的岩性变化,计算效率提高了近4倍.

On the Method of Multi-Resolution Local Structural Entropy for Enhancing Local Discontinuities of Seismic Data

In terms of the variation of geological feature and the time-scale distribution of seismic data,a multi-resolution local structural entropy(LSE) measure based on the multi-scale instantaneous phase is proposed.At first,based on the multi-scale instantaneous phase in wavelet domain,the local structural entropy measure is used to detect different-resolution local discontinuities of seismic data.Then the method of maximizing pixelgrey of image is applied to fuse the local structural entropies in different resolution.And the LSE is finally obtained.In order to reduce the computational complexity the helix coordinate is introduced to arrange and combine the seismic traces in the analyzing cube and to construct the fourth-order covariance matrix.Synthetic results and real data show that not only the large-scale structural discontinuities are displayed clearly,but also small-scale and slow structural variations are enhanced.The approach improves the interpretation of seismic data,and the computational efficiency increases nearly four times.