Methods and apparatuses for facilitating object recognition

Patent metadata

Field	Value
Publication number	US-9286544-B2
Application number	US-201013522486-A
Country	US
Kind code	B2
Filing date	Jan 29, 2010
Priority date	Jan 29, 2010
Publication date	Mar 15, 2016
Grant date	Mar 15, 2016

Abstract

Methods and apparatuses are provided for facilitating object recognition. A method may include accessing data for a first object and data for a second object. The method may additionally include comparing the first and second objects based at least in part upon a reference set and training results generated based at least in part upon the reference set and training data. The method may further include determining whether the first object and the second object are the same object based at least in part upon the comparison. Corresponding apparatuses are also provided.

First claim

What is claimed is:

1. A method comprising:
accessing data for a first object and data for a second object;
extracting a reference feature (P ji a ) from each of the first object and the second object, wherein the reference feature (P ji a ) is represented as:
P ji a =S a T ( Tr ji −m a )

where Tr ji is training data comprising a discriminative subspace and Tr ji ={Tr j1 , Tr j2 , . . . , Tr ji , . . . |j≦G} among G face classes,
where m a is a reference mean, and S a is a reference manifold and S a T is a transformation matrix of S a ;
comparing, by a processor, the first and second objects based at least in part upon the reference features from each of the first and second object, wherein the comparing comprises:
mapping the extracted first and second reference features on the discriminative subspace to obtain a first merged feature and a second merged feature;
determining a distance between the first and second merged features; and
determining whether the first object and the second object are the same object comprises determining whether the first object and the second object are the same object based at least in part upon the determined distance; and

determining, using a processor, whether the first object and the second object are the same object based at least in part upon the comparison.

2. The method of claim 1 , wherein determining whether the first object and the second object are the same object comprises determining that the first object and the second object are the same object when the determined distance is less than a predetermined threshold distance.

3. The method of claim 1 , further comprising generating the reference set from a set of reference objects.

4. The method of claim 1 , further comprising:
determining a discriminative subspace from the extracted reference features using a supervised learning technique, wherein the training results comprise the determined discriminative subspace.

5. The method of claim 1 , wherein the first object comprises a first face and the second object comprises a second face, and wherein determining whether the first object and the second object are the same object comprises determining whether the first face and the second face are sampled from the same person.

6. An apparatus comprising at least one processor and at least one memory storing computer program code, wherein the at least one memory and stored computer program code are configured to, with the at least one processor, cause the apparatus to at least:
access data for a first object and data for a second object;
extract a reference feature (P ji a ) from each of the first object and the second object, wherein the reference feature (P ji a ) is represented as:
P ji a =S a T ( Tr ji −m a )

where Tr ji is training data comprising a discriminative subspace and Tr ji ={Tr j1 , Tr j2 , . . . , Tr ji , . . . |j≦G} among G face classes,
where m a is a reference mean, and S a is a reference manifold and S a T is a transformation matrix of S a ;
compare the first and second objects based at least in part upon the reference features from each of the first and second object, wherein causing the apparatus to compare the first and second objects comprises causing the apparatus to:
map the extracted first and second reference features on the discriminative subspace to obtain a first merged feature and a second merged feature; and
determine a distance between the first and second merged features; and
the at least one memory and stored computer program code are configured to, with the at least one processor, cause the apparatus to determine whether the first object and the second object are the same object by determining whether the first object and the second object are the same object based at least in part upon the determined distance; and

determine whether the first object and the second object are the same object based at least in part upon the comparison.

7. The apparatus of claim 6 , wherein the at least one memory and stored computer program code are configured to, with the at least one processor, cause the apparatus to determine whether the first object and the second object are the same object by determining that the first object and the second object are the same object when the determined distance is less than a predetermined threshold distance.

8. The apparatus of claim 6 , wherein the at least one memory and stored computer program code are configured to, with the at least one processor, cause the apparatus to generate the reference set from a set of reference objects.

9. The apparatus of claim 6 , wherein the at least one memory and stored computer program code are configured to, with the at least one processor, further cause the apparatus to:
determine a discriminative subspace from the extracted reference features using a supervised learning technique, wherein the training results comprise the determined discriminative subspace.

10. The apparatus of claim 9 , wherein the supervised learning technique comprises linear discriminant analysis.

11. The apparatus of claim 6 , wherein the first object comprises a first face and the second object comprises a second face, and wherein the at least one memory and stored computer program code are configured to, with the at least one processor, cause the apparatus to determine whether the first object and the second object are the same object by determining whether the first face and the second face are sampled from the same person.

12. A computer program product comprising at least one non-transitory computer-readable storage medium having computer-readable program instructions stored therein, the computer-readable program instructions comprising program instructions configured to:
access data for a first object and data for a second object;
extract a reference feature (P ji a ) from each of the first object and the second object, wherein the reference feature (P ji a ) is represented as:
P ji a =S a T ( Tr ji −m a )

where Tr ji is training data comprising a discriminative subspace and Tr ji ={Tr j1 , Tr j2 , . . . , Tr ji , . . . |j≦G} among G face classes,
where m a is a reference mean, and S a is a reference manifold and S a T is a transformation matrix of S a ;
compare the first and second objects based at least in part upon the reference features from each of the first and second object, wherein the program code instructions configured to compare the first and second objects comprises program code instructions to:
map the extracted first and second reference features on the discriminative subspace to obtain a first merged feature and a second merged feature;
determine a distance between the first and second merged features; and
determine whether the first object and the second object are the same object comprises determining whether the first object and the second object are the same object based at least in part upon the determined distance; and

determine whether the first object and the second object are the same object based at least in part upon the comparison.

13. The method of claim 1 , wherein the object reference mean is calculated by the formula

m
i

=

∑

g
=
1

Assignees

• Li Jiangwei
• Wang Kongqiao
• Xu Lei
• Nokia Technologies Oy

Inventors

• Li Jiangwei
• Wang Kongqiao
• Xu Lei

Classifications

• G06V10/7715

  Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods · CPC title

• G06V10/771

  Feature selection, e.g. selecting representative features from a multi-dimensional feature space · CPC title

• G06V40/172 Primary

  Classification, e.g. identification · CPC title

• G06F18/2135

  based on approximation criteria, e.g. principal component analysis · CPC title

• G06F18/211

  Selection of the most significant subset of features · CPC title