Borehole image interpretation and analysis
AssigneeMatrix Jvco Ltd
InventorsAbdelwahab Noufal, Abed Benaichouche, Rachid Belmeskine, Mohamed Amri
Life of this patent
Estimating the term…
Abstract
Computer vision system and method for detecting and characterizing heterogeneities from a representation of a borehole. A memory stores processor-executable modules executed by a computer vision processor. The modules include a preprocessing module to extract a plurality of borehole images from the representation of the borehole, a computer vision module to perform semantic segmentation for automated classification of pixels in the borehole images to identify the heterogeneities, and a characterization module to execute a fitting algorithm for extracting dimensional information of the identified heterogeneities.
Claims16
A system comprising:
- a computer vision processor configured to receive at least one representation of a borehole and survey data corresponding thereto;
- and
- a memory storing processor-executable modules that, when executed by the processor, detect and characterize one or more heterogeneities from the representation of the borehole, the processor-executable modules comprising:
- a preprocessing module configured to extract a plurality of borehole images from the representation of the borehole, the extracted borehole images spanning a width of the borehole, wherein the preprocessing module includes a gap filling module configured to inpaint gaps in the borehole images;
- a computer vision module configured to perform semantic segmentation for automated classification of pixels in the borehole images to identify the heterogeneities;
- and
- a characterization module configured to execute a fitting algorithm for extracting dimensional information of the identified heterogeneities, wherein the fitting algorithm executed by the characterization module comprises a sinusoids fitting and a polygon fitting.
A method of detecting and characterizing one or more heterogeneities from a representation of a borehole, the method comprising:
- receiving at least one representation of a borehole and survey data corresponding thereto;
- extracting a plurality of borehole images from the representation of the borehole, the extracted borehole images spanning a width of the borehole;
- inpainting gaps in the borehole images;
- performing semantic segmentation for automated classification of pixels in the borehole images to identify the heterogeneities;
- and
- executing a fitting algorithm for extracting dimensional information of the identified heterogeneities, wherein the fitting algorithm includes both a sinusoids fitting and a polygon fitting.
One or more tangible computer-readable media comprising:
- a plurality of processor-executable modules that, when executed by a processor, detect and characterize one or more heterogeneities from the representation of the borehole, the processor-executable modules comprising:
- a preprocessing module configured to extract a plurality of borehole images from the representation of the borehole, the extracted borehole images spanning a width of the borehole, wherein the preprocessing module includes a gap filling module configured to inpaint gaps in the borehole images;
- a computer vision module configured to perform semantic segmentation for automated classification of pixels in the borehole images to identify the heterogeneities;
- and
- a characterization module configured to execute a fitting algorithm for extracting dimensional information of the identified heterogeneities, wherein the fitting algorithm executed by the characterization module comprises a sinusoids fitting and a polygon fitting.
Dependencies are inferred from the claim text.