Mobile image-based information retrieval system
First Claim
Patent Images
1. An image-based information retrieval system, comprising:
- a database configured to store images in association with object representations and text identifiers, the object representations of the images comprising feature vectors, wherein the feature vectors for at least one image are parameterized according to a corresponding illumination condition of the at least one image; and
a recognition server system configured to receive an image, and further configured to match the received image using the feature vectors of an object representation in the database, and further configured to forward the associated text identifier for a matching object representation to a media server.
3 Assignments
0 Petitions
Accused Products
Abstract
An image-based information retrieval system, including a mobile telephone, a remote recognition server, and a remote media server, the mobile telephone having a built-in camera and a communication link for transmitting an image from the built-in camera to the remote recognition server and for receiving mobile media content from the remote media server, the remote recognition server for matching an image from the mobile telephone with an object representation in a database and forwarding an associated text identifier to the remote server, and the remote media server for forwarding mobile media content to the mobile telephone based on the associated text identifier.
287 Citations
38 Claims
-
1. An image-based information retrieval system, comprising:
-
a database configured to store images in association with object representations and text identifiers, the object representations of the images comprising feature vectors, wherein the feature vectors for at least one image are parameterized according to a corresponding illumination condition of the at least one image; and a recognition server system configured to receive an image, and further configured to match the received image using the feature vectors of an object representation in the database, and further configured to forward the associated text identifier for a matching object representation to a media server. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14)
-
-
15. A method of recognizing an object in an image using an image recognition system, the method comprising:
-
storing, by the image recognition system, for each of a plurality of objects, a set of linked feature vectors representing images of the object from a plurality of viewing angles; storing, by the image recognition system, in association with the set of linked feature vectors of at least a first object of the plurality of objects, times of day at which the images represented by the linked feature vectors were created; receiving, by the image recognition system, an image having at least one object without a known viewing angle; extracting, by the image recognition system, feature vectors from the received image; and recognizing, by the image recognition system, at least the first object in the received image from at least one viewing angle by a finding a closest matching set of linked feature vectors to the extracted feature vectors and by comparing the times of day associated with the closest matching set of linked feature vectors to a time of day of the received image. - View Dependent Claims (16, 17, 18, 19, 20, 21, 22, 23)
-
-
24. A method of recognizing an image, comprising:
-
storing images in a database in association with object representations and text identifiers, the object representations of the images comprising feature vectors, wherein the feature vectors for at least one image are parameterized according to a corresponding illumination condition of the at least one image; receiving an image; and matching the received image with an object representation in the database using the feature vectors of the object representation in the database. - View Dependent Claims (25, 26, 27, 28)
-
-
29. A computer-readable storage medium storing a computer program executable by a processor for recognizing an image, the actions of the computer program comprising:
-
storing images in a database in association with object representations and text identifiers, the object representations of the images comprising feature vectors, wherein the feature vectors for at least one image are parameterized according to a corresponding illumination condition of the at least one image; receiving an image; and matching the received image with an object representation in the database using the feature vectors of the object representation in the database. - View Dependent Claims (30, 31, 32)
-
-
33. An image recognition system, comprising:
-
a database configured to store; for each of a plurality of objects, a set of linked feature vectors representing images of the object from a plurality of viewing angles, and in association with the set of linked feature vectors of at least a first object of the plurality of objects, times of day at which the images represented by the linked feature vectors were created; and a recognition server system configured to; receive an image having at least one object without a known viewing angle, extract feature vectors from the received image, and recognize at least the first object in the received image from at least one viewing angle by a finding, in the database, a closest matching set of linked feature vectors to the extracted feature vectors, and by comparing the times of day associated with the closest matching set of linked feature vectors to a time of day of the received image. - View Dependent Claims (34, 35)
-
-
36. A computer-readable storage medium storing a computer program executable by a processor for recognizing a first object in an image, the actions of the computer program comprising:
-
storing, for each of a plurality of objects, a set of linked feature vectors representing images of the object from a plurality of viewing angles; storing, in association with the set of linked feature vectors of at least a first object of the plurality of objects, times of day at which the images represented by the linked feature vectors were created; receiving an image having at least one object without a known viewing angle; extracting feature vectors from the received image; and recognizing at least the first object in the received image from at least one viewing angle by a finding a closest matching set of linked feature vectors to the extracted feature vectors and by comparing the times of day associated with the closest matching set of linked feature vectors to a time of day of the received image. - View Dependent Claims (37, 38)
-
Specification