Forming regions characterized by labeled measurements

• US 9,058,666 B2
• Filed: 03/15/2013
• Issued: 06/16/2015
• Est. Priority Date: 03/15/2013
• Status: Expired due to Fees

• Current Assignee: R2 Solutions LLC (Acacia Research Corporation)
• Original Assignee: Yahoo! Inc. (Apollo Global Management, Inc.)
• Inventors: Thomee, Bart, Rae, Adam
• Primary Examiner(s): Liu, Li
• Application Number:
  US13/844,637
• Publication Number:
  US 20140270525A1
• Time in Patent Office: 823 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06F 16/29 Geographical information databases
• CPC Class Codes: G06T 2207/20016 Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
• CPC Class Codes: G06T 3/40 Scaling of whole images or parts thereof, e.g. expanding or contracting
• CPC Class Codes: G06T 5/70 Denoising; Smoothing
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/155 involving morphological operators

Abstract

Briefly, embodiments of methods and/or apparatuses for processing, at a variety of scale levels, labeled measurements in sub-regions to form a region characterized by a set of labeled measurements is described.

Claims

1. A method comprising:

computing, in parallel, a plurality of sub-regions formed from one or more smoothed overlapping histograms generated from labeled measurements encoded as physical states in a memory, said sub-regions in combination representing a closed region;

wherein said computing is performed at varying scale levels employing a similar-sized smoothing kernel at said varying scale levels by initially scaling said labeled measurements to a size of interest at a particular scale level before employing said similar-sized smoothing kernel.

2. The method of claim 1, wherein said size of interest corresponds to a size of a bin in a two-dimensional histogram.

3. The method of claim 2, said computing further comprising combining immediately adjacent histogram bins of said size of interest to obtain a different size of interest.

4. The method of claim 1, further comprising:

omitting labeled measurement density outside of selected borders of said smoothed overlapping sub-regions.

5. The method of claim 4, further comprising:

joining said plurality of overlapping sub-regions to form said closed region.

6. The method of claim 5, wherein said joining further comprises:

infilling a portion of a first sub-region of said plurality of smoothed overlapping sub-regions by smoothing one or more histograms representing a second sub-region of said plurality of smoothed overlapping sub-regions.

7. The method of claim 1, wherein said computing includes:

constructing at least two histograms representing label density of said labeled measurements for two successive of said varying scale levels.

8. The method of claim 7, wherein a first of said at least two histograms comprises approximately an inverse square of the grid points that comprise a second of said at least two histograms prior to said scaling.

9. The method of claim 8, wherein said inverse square comprise at least one of the following:

one-quarter;

one-ninth;

or one-twenty-fifth.

10. The method of claim 8, wherein said computing comprises obtaining a difference of Gaussians using differently sized Gaussian kernels to said first histogram.

11. An apparatus, comprising:

one or more processing units to;

scale, at varying scale levels, labeled measurements, encoded as physical states in a memory, to a size of interest; and

smooth, in parallel, one or more histograms representing a plurality of sub-regions characterized by said labeled measurements to generate a plurality of smoothed overlapping sub-regions, wherein said plurality of smoothed overlapping sub-regions to be aggregated to form a closed region characterized by at least one label of said labeled measurements.

12. The apparatus of claim 11, wherein said one or more processing units is additionally to:

form histograms having bins corresponding to a size of interest within said plurality of sub-regions.

13. The apparatus of claim 12, wherein said one or more processing units is additionally to:

form a plurality of combined histogram bins representing labeled measurements from labeled measurements of adjacent formed histogram bins.

14. The apparatus of claim 12, wherein said one or more processing units is additionally to:

determine conformance of at least one label of said labeled measurements of said closed region characterized by at least one label of said labeled measurements.

15. The apparatus of claim 11, wherein said one or more processing units is additionally to:

determine a difference of Gaussians among aggregated sub-regions to determine spatial frequencies among at least two of said sub-regions to be aggregated.

16. An article comprising:

a non-transitory storage medium comprising machine-readable instructions stored thereon which are executable by a special purpose computing apparatus to;

scale, at varying scale levels, labeled measurements, encoded as physical states in a memory, to a size of interest;

employ similar-sized smoothing kernels to sub-regions in a labeled measurement space, said sub-regions representing portions of a closed region characterized by a particular labeled measurement.

17. The article of claim 16, wherein said non-transitory storage medium additionally comprises machine-readable instructions stored thereon which are executable by said special purpose computing apparatus to:

form a histogram representing a distribution of labeled measurements having a particular label.

18. The article of claim 17, wherein said non-transitory storage medium additionally comprises machine-readable instructions stored thereon which are executable by said special purpose computing apparatus to:

join adjacent histogram bins.

19. The article of claim 18, wherein said non-transitory storage medium additionally comprises machine-readable instructions stored thereon which are executable by said special purpose computing apparatus to:

obtain said size of interest.

20. The article of claim 16, wherein said non-transitory storage medium additionally comprises machine-readable instructions stored thereon which are executable by said special purpose computing apparatus to:

infill a portion of a first sub-region a substantially in accordance with one or more histograms representing said first sub-region and a second sub-region to be smoothed.