SCALE-INVARIANT SUPERPIXEL REGION EDGES

• US 20140119656A1
• Filed: 10/31/2013
• Published: 05/01/2014
• Est. Priority Date: 10/31/2012
• Status: Active Grant

• Current Assignee: Environmental Systems Research Institute Incorporated
• Original Assignee: Environmental Systems Research Institute Incorporated
• Inventors: Lilje, Anneliese, Kyle, Jennifer Reiber, McGlinchy, Joseph
• Granted Patent: US 9,299,157 B2
• Time in Patent Office: Days
• US Class Current: 382/173
• CPC Class Codes: G06T 2207/10032 Satellite or aerial image; Remote sensing
• CPC Class Codes: G06T 2207/20016 Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
• CPC Class Codes: G06T 2207/30184 Infrastructure
• CPC Class Codes: G06T 7/12 Edge-based segmentation

Abstract

Certain embodiments of the invention relate to a computer-implemented method that comprises analyzing an image over a plurality of different scales, where the analyzing includes determining spectral characteristics of the image at each of the plurality of different scales, determining spatial characteristics of the image at each of the plurality of different scales, and determining a segmentation pattern (i.e., superpixel) for the image at each of the plurality of different scales based on the spectral and spatial characteristics of that particular scale. The method further includes identifying objects in the image based on portions of the segmentation patterns that are scale-invariant over the plurality of different scales. In some cases, the method can include determining statistical, textural, and/or intensity characteristics of the image at each scale, where determining the segmentation patterns for the image at each scale is further based on one or more of these characteristics.

Claims

1. A computer-implemented method comprising:

receiving an image of a region of interest;

receiving an input corresponding to a plurality of scales;

for each of the plurality of scales;

determining a segmentation pattern to divide the image into a plurality of segments, each segment including a boundary containing a plurality of adjacent pixels, wherein the step of determining a segmentation pattern is performed based on at least one characteristic of pixels of the image and the particular scale; and

generating a segmentation pattern for each of the plurality of scales.

2. The method of claim 1 further comprising:

detecting scale-invariant boundaries or portions thereof over two or more of the segmentation patterns; and

identifying objects based on the scale-invariant boundaries.

3. The method of claim 1, wherein the at least one characteristic of pixels includes at least one of spectral characteristics, spatial characteristics, texture characteristics, statistical characteristics, and intensity characteristics.

4. A computer implemented method comprising:

analyzing an image over a plurality of different scales, wherein analyzing includes;

determining spectral characteristics of the image at each of the plurality of different scales;

determining spatial characteristics of the image at each of the plurality of different scales;

determining a segmentation pattern for the image at each of the plurality of different scales based on the spectral and spatial characteristics of that particular scale; and

identifying objects in the image based on portions of the segmentation patterns that are scale-invariant over the plurality of different scales.

5. The system of claim 4, further comprising:

determining statistical characteristics of the image at each scale,

wherein determining the segmentation pattern for the image at each scale is further based on the statistical characteristics.

6. The system of claim 4, further comprising:

determining texture characteristics of the image at each scale,

wherein determining the segmentation pattern for the image at each scale is further based on the texture characteristics.

7. The system of claim 4, further comprising:

determining texture characteristics, statistical characteristics, and intensity characteristics of the image at each scale,

wherein determining the segmentation pattern for the image at each scale is further based on the texture characteristics, statistical characteristics, and intensity characteristics.

8. The method of claim 4 wherein each of the segmentation patterns includes a plurality of segments, each segment having a boundary, and wherein identifying objects in the image further includes:

super-positioning the segmentation patterns determined at each of the plurality of different scales;

determining which of the boundaries or portions thereof are overlapping over one or more of the super-positioned segmentation patterns, wherein the portions of the segmentation patterns that are scale-invariant correspond to the overlapping boundaries or portions thereof.

9. A computer-implemented system, comprising:

one or more processors; and

one or more non-transitory computer-readable storage mediums containing instructions configured to cause the one or more processors to perform operations including;

analyzing an image over a plurality of different scales, wherein analyzing includes;

determining spectral characteristics of the image at each of the plurality of different scales;

determining spatial characteristics of the image at each of the plurality of different scales;

determining a segmentation pattern for the image at each of the plurality of different scales based on the spectral and spatial characteristics of that particular scale; and

identifying objects in the image based on portions of the segmentation patterns that are scale-invariant over the plurality of different scales.

10. The system of claim 9, further comprising:

determining statistical characteristics of the image at each scale,

wherein determining the segmentation pattern for the image at each scale is further based on the statistical characteristics.

11. The system of claim 9, further comprising:

determining texture characteristics of the image at each scale,

wherein determining the segmentation pattern for the image at each scale is further based on the texture characteristics.

12. The system of claim 9, further comprising:

determining texture characteristics, statistical characteristics, and intensity characteristics of the image at each scale,

wherein determining the segmentation pattern for the image at each scale is further based on the texture characteristics, statistical characteristics, and intensity characteristics.

13. The system of claim 9 wherein each of the segmentation patterns includes a plurality of segments, each segment having a boundary, and wherein identifying objects in the image further includes:

super-positioning the segmentation patterns determined at each of the plurality of different scales;

determining which of the boundaries or portions thereof are overlapping over one or more of the super-positioned segmentation patterns, wherein the portions of the segmentation patterns that are scale-invariant correspond to the overlapping boundaries or portions thereof.

14. The system of claim 9 wherein a range of the plurality of different scales that are used is based on a size of one or more target objects in the image.

15. A non-transitory computer-program product, tangibly embodied in a machine-readable non-transitory storage medium, including instructions configured to cause a data processing apparatus to:

analyze an image over a plurality of different scales, wherein analyzing includes;

determine spectral characteristics of the image at each of the plurality of different scales;

determine spatial characteristics of the image at each of the plurality of different scales;

determine a segmentation pattern for the image at each of the plurality of different scales based on the spectral and spatial characteristics of that particular scale; and

identify objects in the image based on the portions of the segmentation patterns that are scale-invariant over the plurality of different scales.

16. The computer-program product of claim 15, further comprising instructions configured to cause the data processing apparatus to:

determine statistical characteristics of the image at each scale,

wherein determining the segmentation pattern for the image at each scale is further based on the statistical characteristics.

17. The computer-program product of claim 15, further comprising instructions configured to cause the data processing apparatus to:

determine texture characteristics of the image at each scale,

wherein determining the segmentation pattern for the image at each scale is further based on the texture characteristics.

18. The computer-program product of claim 15, further comprising instructions configured to cause the data processing apparatus to:

determine texture characteristics, statistical characteristics, and intensity characteristics of the image at each scale,

wherein determining the segmentation pattern for the image at each scale is further based on the texture characteristics, statistical characteristics, and intensity characteristics.

19. The computer-program product of claim 15 wherein each of the segmentation patterns includes a plurality of segments, each segment having a boundary, and wherein identifying objects in the image further includes instructions configured to cause the data processing apparatus to:

super-position the segmentation patterns determined at each of the plurality of different scales;

determine which of the boundaries or portions thereof are overlapping over the one or more super-positioned segmentation patterns, wherein the portions of the segmentation patterns that are scale-invariant correspond to the overlapping boundaries or portions thereof.

20. The computer-program product of claim 15 wherein a range of the plurality of different scales that are used is based on a size of one or more target objects in the image.