System and method for recognition in 2D images using 3D class models
First Claim
1. A computer-implemented method for recognizing class instances in a 2D image, comprising the steps of:
- (a) constructing a plurality of 3D class models, each model having one or more class parts, and storing the models in a database in a memory;
(b) acquiring a 2D image; and
(c) using a processor to match class instances appearing in the 2D image to the 3D class models, by identifying one or more image features in the 2D image and comparing the identified image features with one or more class parts belonging to the 3D class models.
1 Assignment
0 Petitions
Accused Products
Abstract
A system and method for recognizing instances of classes in a 2D image using 3D class models and for recognizing instances of objects in a 2D image using 3D class models. The invention provides a system and method for constructing a database of 3D class models comprising a collection of class parts, where each class part includes part appearance and part geometry. The invention also provides a system and method for matching portions of a 2D image to a 3D class model. The method comprises identifying image features in the 2D image; computing an aligning transformation between the class model and the image; and comparing, under the aligning transformation, class parts of the class model with the image features. The comparison uses both the part appearance and the part geometry.
-
Citations
35 Claims
-
1. A computer-implemented method for recognizing class instances in a 2D image, comprising the steps of:
-
(a) constructing a plurality of 3D class models, each model having one or more class parts, and storing the models in a database in a memory; (b) acquiring a 2D image; and (c) using a processor to match class instances appearing in the 2D image to the 3D class models, by identifying one or more image features in the 2D image and comparing the identified image features with one or more class parts belonging to the 3D class models. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23)
-
-
24. A non-transitory computer-readable memory containing a database of 3D class models, where:
-
the computer-readable memory contains a database with a plurality of 3D class models, wherein; (a) each 3D class model comprises a plurality of class parts; (b) each class part comprises a part geometry and a part appearance; and (c) each part appearance comprises a first part appearance descriptor that is computed from 2D information. - View Dependent Claims (25, 26, 27, 28)
-
-
29. A computer-implemented method for recognizing class instances in a 2D image, using a database of 3D class models, comprising the steps of
(a) acquiring a 2D image; -
(b) identifying with a processor one or more image features in the 2D image, each feature having a feature location and a feature appearance; (c) constructing one or more correspondence hypotheses, each correspondence hypothesis comprising a class model and a set of feature-part matches; (d) using each correspondence hypothesis to compute with the processor an aligning transformation of the class model to the 2D image; (e) evaluating each correspondence hypothesis under the aligning transformation by computing a class score; and (f) choosing correspondence hypotheses whose class scores meet selected acceptance criteria. - View Dependent Claims (30, 31)
-
-
32. A system for recognizing class instances in a 2D image, comprising
(a) a memory containing a database of 3D class models; -
(b) an image acquisition device for acquiring a 2D image; (c) a processor configured to match class instances appearing in the 2D image to the 3D class models, by identifying image features in the 2D image and comparing the image features with class parts belonging to the 3D class models. - View Dependent Claims (33, 34)
-
-
35. A non-transitory computer-readable medium having embodied thereon a program, the program being executable by a processor to perform a method comprising the steps of:
-
(a) constructing a database of 3D class models, each model having one or more class parts; (b) acquiring a 2D image; and (c) matching class instances appearing in the 2D image to the 3D class models, by identifying one or more image features in the 2D image and comparing the identified image features with one or more class parts of the 3D class models.
-
Specification