APPLICATION OF AN IMPROVED U-NET NEURAL NETWORK ON FRACTURE SEGMENTATION FROM OUTCROP IMAGES

Zhibao Wang, Ziming Zhang, Lu Bai, Yuze Yang, Qiang Ma

Research output: Contribution to conferencePaperpeer-review

Abstract

Outcrop records contain very rich geological historical information, and the study of fractures in outcrop areas is an important part of geological exploration work. The accurate fracture information can provide useful technical support for the development and exploration of subsurface oil and gas. The outcrop images usually include unclear boundaries, complex structure and inconspicuous features, which make fracture detection from outcrop images a difficult task. To tackle these challenges, an improved U-Net algorithm based on the ResNeXt module is proposed in this paper to segment the fractures from the outcrop images. Experiments are conducted on the outcrop images from Yijianfang area in the Tarim Basin in China, and the results show that the proposed algorithm has improved the accuracy and IoU in fracture segmentation from the outcrop images.
Original languageEnglish
Publication statusAccepted/In press - 5 Apr 2022
Event2022 IEEE International Geoscience and Remote Sensing Symposium IGARSS. - Kuala Lumpur, Malaysia
Duration: 17 Jul 202222 Jul 2022

Conference

Conference2022 IEEE International Geoscience and Remote Sensing Symposium IGARSS.
Country/TerritoryMalaysia
CityKuala Lumpur
Period17/07/2222/07/22

Fingerprint

Dive into the research topics of 'APPLICATION OF AN IMPROVED U-NET NEURAL NETWORK ON FRACTURE SEGMENTATION FROM OUTCROP IMAGES'. Together they form a unique fingerprint.

Cite this