Back to results
US 7,460,735 B1Granted
Systems and methods for using image duplicates to assign labels to images
AssigneeGoogle Inc.
InventorsHenry A. Rowley, Shumeet Baluja, Arungundram C. Narendran
Life of this patent
Estimating the term…
Abstract
A system analyzes multiple images to identify similar images using histograms, image intensities, edge detectors, or wavelets. The system retrieves labels assigned to the identified similar images and selectively concatenates the extracted labels. The system assigns the concatenated labels to each of the identified similar images and uses the concatenated labels when performing a keyword search of the plurality of images.
Claims33
An automated method, comprising:
- retrieving a plurality of images;
- analyzing the retrieved images to identify a set of duplicate images by using at least one of histograms, image intensities, edge detectors, wavelets;
- extracting one or more labels, where at least some of the labels are assigned to an image of the identified set of duplicate images;
- concatenating selected ones of the one or more extracted labels;
- assigning the concatenated labels to at least some of the images of the identified set of duplicate images;
- and
- storing information regarding the concatenated labels.
4 more elements
An automated method, comprising:
- performing image analysis on a plurality of images, where at least some of the images are associated with data comprising a set of labels;
- identifying similar images from the plurality of images based on the image analysis;
- selectively combining at least some of the labels associated with at least some of the similar images into a superset of labels by forming a partial union of the at least some of the labels associated with at least some of the similar images;
- assigning at least a portion of the superset of labels to at least some of the similar images;
- storing information regarding the at least a portion of the superset of labels;
- and
- using the assigned at least a portion of the superset of labels when performing a keyword search of the plurality of images.
4 more elements
A computer-readable memory device containing instructions for controlling at least one processor to perform a method, the method comprising:
- analyzing images of a plurality of digital images, where each digital image is associated with data comprising a set of labels;
- identifying similar images from the plurality of digital images based on the image analysis;
- selectively combining at least some of the labels associated with at least some of the similar images into a superset of labels by forming at least a partial union of the at least some of the labels associated with at least some of the similar images;
- assigning at least a portion of the superset of labels to each of the similar images;
- storing information regarding the at least a portion of the superset of labels, and
- using the assigned at least a portion of the superset of labels when performing a keyword search of the plurality of images.
3 more elements
A system implemented in a computer device, comprising:
- means for analyzing a plurality of images where at least some of the images are associated with data comprising a set of labels;
- means for identifying duplicate images from the plurality of images based on the image analysis;
- means for combining at least some of the labels associated with at least some of the duplicate images into a superset of labels by forming a partial union of at least some of the labels associated with at least some of the duplicate images;
- means for assigning at least a portion of the superset of labels at least some of the duplicate images;
- means for storing the at least a portion of the superset of labels. And means for using the assigned at least a portion of the superset of labels when performing a keyword search of the plurality of images.
2 more elements
An automated method, comprising:
- analyzing at least some of a plurality of digital images to identify similar digital images using at least one of histograms, image intensities, edge detectors, or wavelets;
- retrieving labels assigned to the identified similar digital images;
- selectively concatenating the extracted labels;
- assigning the concatenated labels to each of the identified similar digital images;
- and
- using the concatenated labels when performing a keyword search of the plurality of digital images.
3 more elements
Dependencies are inferred from the claim text.