Method and apparatus to recover scene data using re-sampling compressive sensing

• US 10,024,969 B2
• Filed: 08/22/2016
• Issued: 07/17/2018
• Est. Priority Date: 02/24/2014
• Status: Active Grant

• Current Assignee: Perspecta Labs Inc. (Perspecta, Inc.)
• Original Assignee: Vencore Labs, Inc.
• Inventors: Lau, Chi Leung, Woodward, Ted K.
• Primary Examiner(s): Ye, Lin
• Assistant Examiner(s): Morehead, III, John H
• Application Number:
  US15/243,014
• Publication Number:
  US 20170045619A1
• Time in Patent Office: 694 Days
• CPC Class Codes: G01S 17/89 for mapping or imaging
• CPC Class Codes: G01S 7/4808 Evaluating distance, position or velocity data
• CPC Class Codes: G06T 3/40 Scaling of whole images or parts thereof, e.g. expanding or contracting

Abstract

Methods, computer program products, and computer systems for recovering data of scene are provided. The techniques include: obtaining, by at least one processor, data of a scene; re-sampling, by the at least one processor, the data of the scene to obtain re-sampled sensing data of the scene and a corresponding sensing matrix; and constructing, by the at least one processor, recovered data of the scene using the re-sampled sensing data of the scene and the corresponding sensing matrix. In one embodiment, the method includes enhancing occluded data of the scene to construct the recovered data of the scene, where the recovered data of the scene facilitates identification of the scene. In another embodiment, the method includes compressing original data of a scene, the compressing including sampling the original data of the scene to select the data of the scene.

Claims

1. A method comprising:

obtaining data of a scene;

re-sampling, the data of the scene to obtain re-sampled sensing data of the scene and a corresponding sensing matrix, the re-sampling using a plurality of entries, corresponding to a sampling rate for the data of the scene, and a probability parameter to control creation of the sensing matrix, the creation of the sensing matrix comprising probabilistically assigning values of the sensing matrix by evaluating the entries in the plurality of entries to determine whether a next entry of the plurality of entries indicates that a next value for the sensing matrix is to be a first value with a first probability or a second value with a second probability; and

constructing recovered data of the scene using the re-sampled sensing data of the scene and the corresponding sensing matrix.

2. The method of claim 1, wherein the data of the scene comprises occluded data of the scene, and the constructing comprises:

enhancing the occluded data of the scene to construct the recovered data of the scene, wherein the recovered data of the scene facilitates identification of the scene.

3. The method of claim 1, further comprising:

identifying the scene using the recovered data of the scene.

4. The method of claim 1, wherein the obtaining comprises:

compressing original data of a scene, the compressing comprising sampling the original data of the scene to select the data of the scene, wherein the original data of the scene has a first size and the data of the scene has a second size, the second size being smaller than the first size.

5. The method of claim 4, further comprising:

transmitting the data of the scene.

6. The method of claim 1, wherein the constructing comprises:

minimizing an L1 norm of the recovered data of the scene, wherein the re-sampled sensing data of the scene is equal to an inner product of the sensing matrix and the recovered data of the scene.

7. The method of claim 1, wherein the constructing comprises:

minimizing an L1 norm of the recovered data of the scene.

8. The method of claim 1, wherein the constructing comprises:

selecting the re-sampled sensing data of the scene to be equal to an inner product of the sensing matrix and the recovered data of the scene.

9. The method of claim 1, wherein the obtaining comprises:

randomly sampling the scene to determine the data of the scene.

10. The method of claim 9, wherein the random sampling comprises non-Gaussian sampling.

11. A computer system comprising:

a memory; and

at least one processor in communication with the memory, wherein the computer system is configured to perform a method, said method comprising;

obtaining data of a scene;

re-sampling the data of the scene to obtain re-sampled sensing data of the scene and a corresponding sensing matrix, the re-sampling using a plurality of entries, corresponding to a sampling rate for the data of the scene, and a probability parameter to control creation of the sensing matrix, the creation of the sensing matrix comprising probabilistically assigning values of the sensing matrix by evaluating the entries in the plurality of entries to determine whether a next entry of the plurality of entries indicates that a next value for the sensing matrix is to be a first value with a first probability or a second value with a second probability; and

constructing recovered data of the scene using the re-sampled sensing data of the scene and the corresponding sensing matrix.

12. The computer system of claim 11, wherein the data of the scene comprises occluded data of the scene, and the constructing comprises:

enhancing the occluded data of the scene to construct the recovered data of the scene, wherein the recovered data of the scene facilitates identification of the scene.

13. The computer system of claim 11, wherein said method further comprises:

identifying the scene using the recovered data of the scene.

14. The computer system of claim 11, wherein the obtaining comprises:

compressing original data of a scene, the compressing comprising sampling the original data of the scene to select the data of the scene, wherein the original data of the scene has a first size and the data of the scene has a second size, the second size being smaller than the first size.

15. The computer system of claim 11, wherein the constructing comprises:

minimizing an L1 norm of the recovered data of the scene, wherein the re-sampled sensing data of the scene is equal to an inner product of the sensing matrix and the recovered data of the scene.

16. A computer program product comprising:

a non-transitory computer readable storage medium readable by a processing circuit and storing instructions for execution by the processing circuit for performing a method comprising;

obtaining data of a scene;

re-sampling the data of the scene to obtain re-sampled sensing data of the scene and a corresponding sensing matrix, the re-sampling using a plurality of entries, corresponding to a sampling rate for the data of the scene, and a probability parameter to control creation of the sensing matrix, the creation of the sensing matrix comprising probabilistically assigning values of the sensing matrix by evaluating the entries in the plurality of entries to determine whether a next entry of the plurality of entries indicates that a next value for the sensing matrix is to be a first value with a first probability or a second value with a second probability; and

constructing recovered data of the scene using the re-sampled sensing data of the scene and the corresponding sensing matrix.

17. The computer program product of claim 16, wherein the data of the scene comprises occluded data of the scene, and the constructing comprises:

enhancing the occluded data of the scene to construct the recovered data of the scene, wherein the recovered data of the scene facilitates identification of the scene.

18. The computer program product of claim 16, wherein said method further comprises:

identifying the scene using the recovered data of the scene.

19. The computer program product of claim 16, wherein the obtaining comprises:

compressing original data of a scene, the compressing comprising sampling the original data of the scene to select the data of the scene, wherein the original data of the scene has a first size and the data of the scene has a second size, the second size being smaller than the first size.

20. The computer program product of claim 16, wherein the constructing comprises:

minimizing an L1 norm of the recovered data of the scene, wherein the re-sampled sensing data of the scene is equal to an inner product of the sensing matrix and the recovered data of the scene.