Automated fracture detection and characterization from unwrapped drill-core images using Mask R-CNN
Creators
- 1. School of Minerals and Energy Resources, University of New South Wales, Sydney, Australia, f.al-zubaidi@unsw.edu.au
- 2. School of Minerals and Energy Resources, University of New South Wales, Sydney, Australia
- 3. Lundin Energy Norway AS, Lysaker, Norway
Description
Drill cores provide the most reliable fracture information in subsurface formations as they present a clear and direct view of fractures. Core observation and image log interpretation are usually integrated for fracture analysis of underground layers. There has been a strong move towards developing automated fracture detection methods, however, the focus has been on extracting fracture information from log images, such as acoustic or resistivity image logs. Such efforts using core images are significantly less. This study presents a machine learning-based approach for automatic fracture recognition from unwrapped drill-core images. The proposed method applies a state-of-the-art convolutional neural network for object identification and segmentation. The study also investigates the feasibility of using synthetic fracture images for training the model by creating two types of synthetic data using masks of real fractures and creating sinusoidal shaped fractures. The trained model is then used to detect fractures in real core images from two different boreholes and achieved a precision of 94.80%. The identified fractures are further analyzed and compared to manually segmented fractures in terms of fracture dip angle and dip direction, which achieved average absolute errors of 2.18 and 10.58, respectively. Overall, the study presents a novel application of an advanced machine learning algorithm for fracture detection and analysis from unwrapped core images.
Notes
Files
ID108.pdf
Files
(489.9 kB)
Name | Size | Download all |
---|---|---|
md5:0e20a2c1ddede0f6be6d8c99dfa37948
|
489.9 kB | Preview Download |