Full text loading...
-
Coupled Hough-Transform and Deep Learning for Improved Diffraction Delineation
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, NSG2022 4th Conference on Geophysics for Mineral Exploration and Mining, Sep 2022, Volume 2022, p.1 - 5
Abstract
Summary
Deep-learning algorithms are widely used in seismic data processing and interpretation, although mainly in sedimentary settings. In hardrock settings and for mineral exploration, deep-learning algorithms can also be utilized for targeting deep-seated deposits that often are associated with clear diffractions. Due to low signal-to-noise ratio in hardrock environment, deep-learning denoising solutions alone are usually insufficient for diffraction delineation, and additional image processing tools are required. In this study, we use Hough-transform methods to showcase its potential on a synthetic seismic dataset for targeting and removal of coherent noise that passed through a deep-learning denoising algorithm, while preserving and enhancing diffractions. We also applied the Hough-transform to detect diffractions on time slices using their circular pattern. This was showcased using a hardrock 3D seismic dataset where for example a major diffraction was generated from a volcanogenic massive sulphide deposit. Additional diffractions were also identified that have potential to be further studied. A further use of the methods employed in this study can also be coupled into supervised deep-learning algorithms for automatic labelling.
© EAGE Publications BV