Data automation for image logs texture analysis: supervised methodology and case studies

A new methodology for the recognition of geological and lithological textures in borehole image logs has been developed and tested. The workflow consists of a features-extraction step, to provide the initial segmentation, inspired by the multi-channel filtering theory, and a classification step, to provide the texture map. The workflow is simple and parametric, making it suitable for automatic interpretation of a wide range of geological images. The goal of this work was the automatic recognition of textural facies in high-resolution microresistivity image logs, using information derived from core interpretation to improve the quality of the classification in intervals biased by poor log quality.