Texture-Based-Similarity Graph to Aid Seismic Interpretation

Abstract

Seismic interpreters use their trained eyes to assess the similarity between seismic datasets. However, to find and evaluate all relevant parts of a seismic cube can be a time-consuming task. We have developed a method based on texture analysis and graph theory that can automatically compare seismic sections. Such a method has the potential to help experts in several tasks and, to the best of our knowledge, it is the first one to tackle seismic image similarity combining texture descriptors and graphs.

This work proposes a Texture-Based-Similarity Graph that represents the seismic survey as a graph whose nodes represent seismic sections and whose edges represent the Euclidean distance between Local Binary Pattern (LBP) feature vectors computed for each section. By using this representation, it is possible to calculate the similarity between any two seismic sections. Based on the proposed technique, we created a system to accelerate the inspection of a seismic cube. Using our method, we suggest which seismic sections (key-sections) should be considered in the interpretation process taking into consideration their distance to neighboring lines in one specific cube. Key-sections computed with this method could be used to build a non-regular grid that is more likely to capture the underlying structures present in a survey, allowing for a faster and more precise interpretation.

Experiments conducted on two public datasets (Netherlands F3 and Penobscot) indicated that the methodology has a great potential to speed up the seismic interpretation process. Other possible applications would be to use key-sections distance map to highlight different areas within a seismic survey or to compare 3D surveys in the search for analogs.

AAPG Datapages/Search and Discovery Article #90323 ©2018 AAPG Annual Convention and Exhibition, Salt Lake City, Utah, May 20-23, 2018