Improved seismic texture analysis based on nonlinear gray-level transformation (original) (raw)

2016, SEG Technical Program Expanded Abstracts 2016

Seismic texture analysis is a common and useful tool in delineating depositional features from three-dimensional (3D) seismic surveys, and various texture attributes have been presented for robust facies interpretation, including the popular gray-level co-occurrence matrix (GLCM) and its derived attributes. When applied for attribute extraction from a seismic volume, most texture algorithms perform a gray-level transformation that rescales seismic amplitude into a user-defined range linearly. However, most features of interpretation interest in a seismic dataset often cover only a small portion of its whole amplitude range. For improved texture delineation of such features, they are expected to be represented by more gray levels while the rest by fewer levels, which is non-linear. This study proposes implementing three non-linear (logarithmic, exponential and sigmoid) transformations for seismic texture attribute extraction and interpretation. Applications to a deep marine depositional system from Angola demonstrates an improved resolution of GLCM homogeneity and contrast, which help better delineate channels and other features such as fans and lobes compared to the traditional linear transformation.