Systems and methods for used learned representations to determine terrain type

Patent metadata

Field	Value
Publication number	US-11275940-B1
Application number	US-202016924409-A
Country	US
Kind code	B1
Filing date	Jul 9, 2020
Priority date	Jun 19, 2018
Publication date	Mar 15, 2022
Grant date	Mar 15, 2022

Abstract

Disclosed herein are embodiments of systems, methods, and products comprise an analytic server, which provides a terrain segmentation and classification tool for synthetic aperture radar (SAR) imagery. The server accurately segments and classifies terrain types in SAR imagery and automatically adapts to new radar sensors data. The server receives a first SAR imagery and trains an autoencoder based on the first SAR imagery to generate learned representations of the first SAR imagery. The server trains a classifier based on labeled data of the first SAR imagery data to recognize terrain types from the learned representations of the first SAR imagery. The server receives a terrain query for a second SAR imagery. The server translates the second imagery data into the first imagery data and classifies the second SAR imagery terrain types using the classifier trained for the first SAR imagery. By reusing the original classifier, the server improves system efficiency.

First claim

What is claimed is:

1. A computer-implemented method comprising:
receiving, by a computer, from a client computer a request including a query imagery generated by a sensor device of a first device type;
translating, by the computer, the query imagery from the first device type to a learned representation for a second device type by applying an autoencoder on the query imagery, the autoencoder trained to generate learned representations of the second device type;
identifying, by the computer, one or more terrain types in the learned representation of the query imagery by applying a classifier on the learned representation, the classifier trained to determine the one or more terrain types based upon one or more learned representations of the second device type; and
generating, by the computer, a graphical user interface to display the one or more terrain types in the learned representation at the client computer.

2. The method according to claim 1 , further comprising training, by the computer, the autoencoder by performing unsupervised learning on a unlabeled dataset of a training imagery of the second device type to generate learned representations of the a training imagery.

3. The method according to claim 2 , wherein the computer trains a new autoencoder in response to receiving a new training imagery of a new device type.

4. The method according to claim 2 , wherein the training imagery comprises the unlabeled dataset containing original imagery data obtained by a second sensor device of the second device type.

5. The method according to claim 1 , further comprising training, by the computer, the classifier by performing supervised learning on a labeled dataset of a training imagery of the second device type, wherein the labeled dataset comprises the one or more terrain types of the training imagery.

6. The method according to claim 5 , further comprising receiving, by the computer, the labeled dataset of the training imagery of the second device type.

7. The method according to claim 5 , wherein the labeled dataset includes one or more reference objects corresponding to the one or more terrain types.

8. The method according to claim 7 , wherein the reference objects provide a correspondence relationship between the training imagery and the query imagery.

9. The method according to claim 5 , further comprising receiving, by a computer, the training imagery from one or more sensor devices of the second device type.

10. The method according to claim 1 , further comprising retrieving, by the computer, from a database the classifier trained for the second device type.

11. A system comprising:
a first sensor device of a first device type;
a second sensor device of a second device type; and
a server comprising a processor configured to:
receive from a client computer a request including a query imagery generated by a sensor device of a first device type;
translate the query imagery from the first device type to a learned representation for a second device type by applying an autoencoder on the query imagery, the autoencoder trained to generate learned representations of the second device type;
identify one or more terrain types in the learned representation of the query imagery by applying a classifier on the learned representation, the classifier trained to determine the one or more terrain types based upon one or more learned representations of the second device type; and
generate a graphical user interface to display the one or more terrain types in the learned representation at the client computer.

12. The system according to claim 11 , wherein the server is further configured train the autoencoder by performing unsupervised learning on a unlabeled dataset of a training imagery of the second device type to generate learned representations of the a training imagery.

13. The system according to claim 12 , further comprising a new sensor device of a new device type, wherein the server trains a new autoencoder in response to receiving the new training imagery of the new device type.

14. The system according to claim 12 , wherein the training imagery comprises the unlabeled dataset containing original imagery data obtained by the second sensor device of the second device type.

15. The system according to claim 11 , wherein the server is further configured to train the classifier by performing supervised learning on a labeled dataset of a training imagery of the second device type, wherein the labeled dataset comprises the one or more terrain types of the training imagery.

16. The system according to claim 15 , wherein the server is further configure to receive the labeled dataset of the training imagery of the second device type.

17. The system according to claim 15 , wherein the labeled dataset includes one or more reference objects corresponding to the one or more terrain types.

18. The system according to claim 17 , wherein the reference objects provide a correspondence relationship between the training imagery and the query imagery.

19. The system according to claim 15 , wherein the server is further configure to receive the training imagery from one or more sensor devices of the second device type.

20. The system according to claim 11 , further comprising a processor configured to store one or more classifiers, wherein the server is further configure to retrieve from the database the classifier trained for the second device type.

Assignees

• Architecture Tech Corp

Inventors

• Nicotera Paul
• Joyce Robert
• Powers Judson
• Mcardle Daniel

Classifications

• G06V10/764 Primary

  using classification, e.g. of video objects · CPC title

• G06V10/82

  using neural networks · CPC title

• G06V20/13

  Satellite images · CPC title

• G06T2207/10032

  Satellite or aerial image; Remote sensing · CPC title

• G06T2207/30188

  Vegetation; Agriculture · CPC title