Haralick texture

In the early 1970s, Bob Haralick computed a family of texture attributes from a grey-level co-occurrence matrix (GLCM; sometimes referred to as a grey tone spatial dependency matrix). The approach produces "texture" attributes, such as contrast and entropy, based on how often pixel values in an image occur next to other pixel values.

Algorithm

 * 1) Read data from an image, seismic horizon, or grid
 * 2) Compute GLCMs in a moving window
 * 3) Compute statistics of GLCMs, and put in new grid
 * 4) Result: various attributes, such as entropy, energy, contrast

Applications to geoscience

 * Seismic facies classification
 * Core facies classification

Implementations

 * OpendTect — 3d seismic textures
 * GeoCraft — 2D horizons only
 * ImageJ — via various plugins, e.g. this one
 * Agile — see Github