Similar image selection
AssigneeGoogle Inc.
InventorsShumeet Baluja, Michele Covell
Life of this patent
Estimating the term…
Abstract
An image similar to a target image is selected from among a set of candidate images. A set of image classifiers is first generated and used to create a fingerprint for each candidate image. A hash table is generated for each fingerprint segment, and an identifier for each candidate image is stored in each hash table based on the candidate image fingerprint value for the fingerprint segment associated with the hash table. A fingerprint is created for the target image using the set of classifiers. Segments of the target image fingerprints are compared to segments of the candidate image fingerprints using the hash table, and a candidate image similar to the target image is selected based on this comparison.
Claims27
A method, comprising:
- generating, by a device comprising a processor, a set of image classifiers that classify images according to characteristics of the images, wherein the generating comprises:
- iteratively generating the set of image classifiers, wherein the set of image classifiers classify a training image of a set of training images with an assigned image class index of one or more image class indexes, wherein an image class of one or more image classes corresponds to the training image;
- and
- classifying a set of candidate images using the set of image classifiers to generate fingerprints for the set of candidate images, wherein a fingerprint comprises a plurality of segments.
A method, comprising:
- generating, by a device comprising a processor, a set of image classifiers that classify images according to characteristics of the images;
- and
- classifying a set of candidate images using the set of image classifiers to generate fingerprints for the set of candidate images, wherein a fingerprint comprises a plurality of segments, wherein the image classifiers are binary classifiers and wherein the classifying the set of candidate images using the set of image classifiers to generate fingerprints for the set of candidate images comprises:
- classifying a candidate image using the binary classifiers in the set, wherein one or more of the binary classifiers produces a binary value based on classification.
A device, comprising:
- a processor, coupled to a memory, and that executes or facilitates execution of instructions to at least:
- determine a set of training images comprising images from multiple image classes;
- iteratively generate a set of image classifiers for classifying the images according to characteristics of the images, wherein the set of image classifiers, in response to classifying a training image of the set of training images, match classification of the training image to an assigned image class index corresponding to an image class of the training image;
- and
- classify a set of candidate images using the set of image classifiers to generate fingerprints for the set of candidate images, a fingerprint comprising a plurality of segments.
A non-transitory computer-readable storage medium storing instructions that, in response to execution, cause a device comprising a processor to perform operations, comprising:
- iteratively generating a set of image classifiers for classifying images according to characteristics of the images, wherein the set of image classifiers, in response to classifying a training image, match classification of the training image to an assigned image class index corresponding to an image class of the training image;
- and
- classifying a set of candidate images using the set of image classifiers to generate fingerprints for the set of candidate images, wherein a fingerprint comprises a plurality of segments.
Dependencies are inferred from the claim text.