Granted Patent B1

US 12,372,369 · App. 18/770,321 · Granted Jul 29, 2025

Inventors: Patrick Loughrin Neary (Hyde Park, UT); Kimberly Margaret Moore (Redwood City, CA); Regupathi Angappan (Baltimore, MD)

Assignee: SB Technology, Inc.

G01C21/3804G01C21/08

Abstract

Claims (50)

1. A method, comprising:

obtaining a first geophysical magnetic map of a navigable region of the earth;

determining, based on a first machine learning model, one or more areas in the first geophysical magnetic map, wherein each of the one or more areas includes one or more artifacts in the first geophysical magnetic map, and determining the one or more areas comprises:

generating, based on the first machine learning model and the magnetic map, a second geophysical magnetic map by the first machine learning model; and

comparing the second geophysical magnetic map with the first geophysical magnetic map to determine the one or more areas; and

removing, from the first geophysical magnetic map, the one or more artifacts in each of the one or more areas, wherein removing the one or more artifacts comprises;

determining, based on a second machine learning model, a respective set of revised map values for at least one of the one or more artifacts; and

replacing the map values associated with at least one of the one or more artifacts with the respective set of revised map values,

wherein the second machine learning model comprises a variational autoencoder that takes as input first geophysical magnetic map data and processes the one or more artifacts in the first geophysical magnetic map to generate the respective set of revised geophysical magnetic map values for the at least one of the one or more artifacts.

2. The method according to claim 1 , further comprises training the second machine learning model based on a plurality of magnetic maps that are free of artifacts and a plurality of magnetic maps that include artifacts.

3. The method according to claim 1 , further comprising:

sending the respective set of revised map values for each of the one or more artifacts to a user;

receiving, from the user, a respective second set of revised map values for each of the one or more artifacts; and

re-training the second machine learning model based on the respective second set of revised map values for each of the one or more artifacts.

4. The method according to claim 1 , wherein an input to the first machine learning model comprises the magnetic map and at least one of a topographic map or a gravity-based map of the region.

5. The method according to claim 1 , wherein an output from the first machine learning model comprises the one or more areas or metadata associated with the one or more areas.

6. The method according to claim 1 , further comprising:

sending the one or more artifacts to a user; and

receiving, from the user, the respective set of revised map values for each of the one or more artifacts.

7. The method according to claim 6 , wherein sending the one or more artifacts to the user comprises displaying on a graphic user interface (GUI) the one or more artifacts.

8. The method according to claim 1 , wherein the second machine learning model comprises a generative adversarial network (GAN) that processes the one or more artifacts in the magnetic map to generate the respective set of revised map values for each of the one or more artifacts.

9. The method according to claim 8 , further comprises:

generating a plurality of synthetic magnetic maps based on the magnetic map of the region and a plurality of simulated artifacts; and

training the GAN based on the plurality of synthetic magnetic maps.

10. The method according to claim 1 , wherein the first machine learning model comprises an autoencoder, the autoencoder comprises a plurality of encoding blocks and a plurality of decoding blocks, and determining the one or more areas comprises processing, based on the plurality of encoding blocks and the plurality of decoding blocks, the magnetic map to determine the one or more areas.

11. The method according to claim 10 , wherein the autoencoder is based on a convolutional neural network (CNN).

12. The method according to claim 10 , further comprises training the first machine learning model based on a plurality of magnetic maps that are free of artifacts and a plurality of magnetic maps that include artifacts.

13. The method according to claim 10 , wherein a loss function of the autoencoder comprises a similarity-based loss function.

14. The method according to claim 13 , wherein the similarity-based loss function comprises a structure-based similarity index or a feature-based similarity index.

15. A system, comprising:

one or more processors; and

a non-transitory computer readable medium coupled to the one or more processors and storing instructions that, when executed by the one or more processors, cause the system to perform operations comprising:

obtaining a first geophysical magnetic map of a navigable region of the earth;

determining, based on a first machine learning model, one or more areas in the first geophysical magnetic map, wherein each of the one or more areas includes one or more artifacts in the first geophysical magnetic map, and determining the one or more areas comprises:

generating, based on the first machine learning model and the magnetic map, a second geophysical magnetic map by the first machine learning model; and

comparing the second geophysical magnetic map with the first geophysical magnetic map to determine the one or more areas; and

removing, from the first geophysical magnetic map, the one or more artifacts in each of the one or more areas, wherein removing the one or more artifacts comprises;

determining, based on a second machine learning model, a respective set of revised map values for at least one of the one or more artifacts; and

replacing the map values associated with at least one of the one or more artifacts with the respective set of revised map values,

wherein the second machine learning model comprises a variational autoencoder that takes as input first geophysical magnetic map data and processes the one or more artifacts in the first geophysical magnetic map to generate the respective set of revised geophysical magnetic map values for the at least one of the one or more artifacts.

16. The system according to claim 15 , wherein removing the one or more artifacts comprises replacing map values associated with each of the one or more artifacts with a respective set of revised map values.

17. An artifact-free geophysical magnetic map without artifact magnitudes above a specified threshold made by a process, the process comprising:

obtaining a first geophysical magnetic map of a navigable region of the earth;

determining, based on a first machine learning model, one or more areas in the first geophysical magnetic map, wherein each of the one or more areas includes one or more artifacts in the first geophysical magnetic map, and determining the one or more areas comprises:

generating, based on the first machine learning model and the magnetic map, a second geophysical magnetic map by the first machine learning model; and

comparing the second geophysical magnetic map with the first geophysical magnetic map to determine the one or more areas; and

generating the artifact-free geophysical magnetic map by removing, from the first geophysical magnetic map, the one or more artifacts in each of the one or more areas, wherein removing the one or more artifacts comprises;

determining, based on a second machine learning model, a respective set of revised map values for at least one of the one or more artifacts; and

replacing the map values associated with at least one of the one or more artifacts with the respective set of revised map values,

wherein the second machine learning model comprises a variational autoencoder that takes as input first geophysical magnetic map data and processes the one or more artifacts in the first geophysical magnetic map to generate the respective set of revised geophysical magnetic map values for the at least one of the one or more artifacts.

Assignments (1)

References Cited (17)