Home

Quick Links

Search

 
Impact of grey level transformation and chosen amplitude range on GLCM-based anisotropy estimationNormal access

Author: Christoph Georg Eichkitz
Journal name: First Break
Issue: Vol 37, No 5, May 2019 pp. 67 - 74
Language: English
Info: Article, PDF ( 3.61Mb )
Price: € 30

Summary:
The relative influence of parameters used for seismic anisotropy estimation based on the grey level co-occurrence matrix (GLCM) was examined in a previous article (Eichkitz and Amtmann, 2018) whose objective was a general understanding of the parameters, including the number of grey levels, type of grey level transformation, size of the analysis window, and type of GLCM-based attribute. We now examine the influence of the type of grey level transformation and amplitude range chosen for the transformation. The grey level co-occurrence matrix (GLCM) is a second-order statistical texture classification method originally described by Haralick et al. (1973) for interpretation of a pixelated image. Within a moving analysis window, the co-occurrences of neighbouring pixel values are counted and written into a symmetrical 2D matrix whose size corresponds to the number of grey levels. The counting of neighbouring pixel values is performed in specific directions. For two-dimensional data such as digital photographs, the GLCM is computed in four directions, either independently or all at once. For three-dimensional data such as seismic cubes, the computation algorithm is extended to 13 space directions (Eichkitz et al., 2013). After counting co-occurrences, the elements in the 2D GLCM matrix are divided by the total number of entries, resulting in a relative frequency matrix that can be regarded as a kind of probability matrix. This matrix is then used to compute various attributes (Haralick et al., 1973; Soh and Tsatsoulis, 1999; Wang et al., 2010). Common applications of GLCM attributes include classification of satellite images (Franklin et al. 2001; Tsai et al. 2007) and images based on magnetic resonance or computed tomography (Kovalev et al., 2001; Zizzari et al., 2011). Seismic applications include the direct extraction of GLCM-based attributes to interpret channels systems (Eichkitz et al., 2013, 2014, 2015a, 2015b, 2016; West et al., 2002; Gao, 2007, 2011; de Matos et al., 2011), sedimentary facies (Di and Gao, 2017; Eichkitz et al., 2012; Chopra and Alexeev, 2005, 2006a, 2006b; Yenugu et al., 2010; Wang et al., 2016), salt bodies (Gao, 2003), and fractures (Eichkitz et al., 2015c, 2016; Schneider et al., 2016).


Back to the article list