Automatic Salt Deposit Identification from Seismic Images
Authors: Oyebode K, Iruansi U
DOI Info: http://doi.org/10.5281/zenodo.5047558
Identifying mineral deposits from seismic data has been receiving a lot of research attention of late most especially salt identification. Attempts have been made to automate this process to phase out its manual identification. To this end, artificial intelligence models have been leveraged, and one of such is the UNET model. This work puts forward an improved UNET model by adding a preprocessing image layer (diffusion filtering) on top of the UNET model. Results on 2011 salt images showed improved performance over the traditional UNET model.
Affiliations: Department of Computer Engineering, Faculty of Engineering, University of Benin, PMB 1154, Benin City, Nigeria
Keywords: Deep Learning, Image Processing, Image Enhancement, Image Segmentation, Image Feature Engineering
Published date: 2021/06/30