System and method for improved computational processing efficiency in the HSEG algorithm

• US 8,526,733 B2
• Filed: 06/01/2011
• Issued: 09/03/2013
• Est. Priority Date: 06/03/2010
• Status: Active Grant

• Current Assignee: United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration
• Original Assignee: The United States of America As Represented By The Secretary of Agriculture
• Inventors: Tilton, James C.
• Primary Examiner(s): COUSO, JOSE L
• Application Number:
  US13/150,316
• Publication Number:
  US 20110299777A1
• Time in Patent Office: 825 Days
• Field of Search: 382/173, 382175-177, 382180-181, 382190-191, 382/199, 382203-204
• US Class Current: 382/173
• CPC Class Codes: G06T 2207/10032 Satellite or aerial image; Remote sensing
• CPC Class Codes: G06T 2207/10036 Multispectral image; Hyperspectral image
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/187 involving region growing; involving region merging; involving connected component labelling
• CPC Class Codes: G06V 20/13 Satellite images

Abstract

Disclosed herein are systems, methods, and non-transitory computer-readable storage media for processing image information using a more computationally efficient version of a previously developed HSEG (Hierarchical Segmentation) algorithm. The HSEG algorithm includes iterations intertwining two forms of region growing to produce a set of hierarchically related image segmentations. The first form is a commonly used approach in which regions are grown by merging the most similar spatially adjacent regions. The second form is a unique approach in which regions are grown by also merging spatially non-adjacent regions that are at least as similar as the spatially adjacent regions merged by the first form in the same iteration. The improved HSEG algorithm limits the regions considered for non-adjacent region merging in the second form of operation to regions having at least a minimum number of pixels. This minimum is dynamically controlled to optimize processing efficiency and image segmentation quality.

Claims

1. A system comprising:

a processor;

a memory storing instructions for controlling the processor to perform steps comprising;

receiving image data;

in a first operating mode yielding first merged regions;

identifying spatially adjacent regions in the image data;

merging spatially adjacent regions having a spectral similarity above a minimum spectral similarity threshold;

in a second operating mode yielding second merged regions;

computing a spectral similarity value for all regions, including spatially non-adjacent regions and spatially adjacent regions, in the image data having at least a minimum number of pixels;

weighting the minimum spectral similarity threshold to yield a weighted minimum spectral similarity threshold;

merging spatially non-adjacent regions by comparing respective spectral similarity values to the weighted minimum spectral similarity threshold; and

outputting a combination of the first merged regions and the second merged regions.

2. The system of claim 1, wherein the minimum spectral similarity threshold is weighted by a factor greater than 0 and less than or equal to 1.

3. The system of claim 1, wherein merging spatially non-adjacent regions further comprises merging a pair of spatially non-adjacent regions.

4. The system of claim 1, wherein the minimum number of pixels is dynamically determined based on a targeted range of a number of regions.

5. The system of claim 1, wherein the minimum number of pixels is dynamically determined based on a desired level of improved image segmentation accuracy.

6. The system of claim 1, wherein the minimum number of pixels is dynamically determined based on a desired level of computational speed.

7. The system of claim 1, further comprising alternating between the first operating mode and the second operating mode.

8. A non-transitory computer-readable storage medium storing instructions which, when executed by a computing device, cause the computing device to perform steps comprising:

receiving image data;

in a first operating mode yielding first merged regions;

identifying spatially adjacent regions in the image data;

merging spatially adjacent regions having a spectral similarity above a minimum spectral similarity threshold;

in a second operating mode yielding second merged regions;

computing a spectral similarity value for all regions, including spatially non-adjacent regions and spatially adjacent regions, in the image data having at least a minimum number of pixels;

weighting the minimum spectral similarity threshold to yield a weighted minimum spectral similarity threshold;

merging spatially non-adjacent regions by comparing respective spectral similarity values to the weighted minimum spectral similarity threshold; and

outputting a combination of the first merged regions and the second merged regions.

9. The non-transitory computer-readable storage medium of claim 8, wherein the minimum spectral similarity threshold is weighted by a factor greater than 0 and less than or equal to 1.

10. The non-transitory computer-readable storage medium of claim 8, wherein merging spatially non-adjacent regions further comprises merging a pair of spatially non-adjacent regions.

11. The non-transitory computer-readable storage medium of claim 8, wherein the minimum number of pixels is dynamically determined based on a targeted range of a number of regions.

12. The non-transitory computer-readable storage medium of claim 8, wherein the minimum number of pixels is dynamically determined based on a desired level of improved image segmentation accuracy.

13. The non-transitory computer-readable storage medium of claim 8, wherein the minimum number of pixels is dynamically determined based on a desired level of computational speed.