Recognition by parts using adaptive and robust correlation filters
First Claim
1. A correlation filter for determining if a physical test target in at least one test image obtained using an imaging device matches a physical training target in at least one training image, comprising:
- a) a conjugate transpose module configured to generate a conjugate transpose SH of an image matrix S that includes at least one training image vector, each of the at least one training image vector including a vectorized representation of the at least one training image;
b) a power spectrum module configured to generate a power spectrum matrix Dx using at least one test image vector x, each of the at least one test image vector x including a vectorized representation of the at least one test image; and
c) a correlation processor configured to generate correlation-peak-strength and distance-from-origin data using the conjugate transpose SH and power spectrum matrix Dx according to h=(Dx+ε
I)−
1S[SH(Dx+ε
I)−
1S]−
1d; and
wherein;
i) ε
is a regularization constant;
ii) I is an identity matrix; and
iii) Dx is a vector desired response with elements corresponding to at least one of the at least one training image vector in matrix S.
2 Assignments
0 Petitions
Accused Products
Abstract
A recognition-by-parts authentication system for determining if a physical test target represented in test image(s) obtained using an imaging device matches a physical training target represented in training image(s). The system includes a multitude of adaptive and robust correlation filters. Each of the adaptive and robust correlation filters is configured to generate correlation-peak-strength and distance-from-origin data using a multitude of related images. Each of the multitude of related images representing a similar part of a larger image. The related images originate from the test image(s) and training image(s).
110 Citations
18 Claims
-
1. A correlation filter for determining if a physical test target in at least one test image obtained using an imaging device matches a physical training target in at least one training image, comprising:
-
a) a conjugate transpose module configured to generate a conjugate transpose SH of an image matrix S that includes at least one training image vector, each of the at least one training image vector including a vectorized representation of the at least one training image; b) a power spectrum module configured to generate a power spectrum matrix Dx using at least one test image vector x, each of the at least one test image vector x including a vectorized representation of the at least one test image; and c) a correlation processor configured to generate correlation-peak-strength and distance-from-origin data using the conjugate transpose SH and power spectrum matrix Dx according to h=(Dx+ε
I)−
1S[SH(Dx+ε
I)−
1S]−
1d; andwherein; i) ε
is a regularization constant;ii) I is an identity matrix; and iii) Dx is a vector desired response with elements corresponding to at least one of the at least one training image vector in matrix S. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10)
-
-
11. A recognition-by-parts authentication system for determining if a physical test target represented in at least one test image obtained using an imaging device matches a physical training target represented in at least one training image, comprising:
- a multitude of adaptive and robust correlation filters, each of the multitude of adaptive and robust correlation filters configured to generate correlation-peak-strength and distance-from-origin data using a multitude of related images, each of the multitude of related images representing a similar part of at least one larger image, at least one of the related images originating from at least one of the at least one test image, at least one of the related images originating from at least one of the at least one training image, wherein at least one of the multitude of adaptive and robust correlation filters includes;
a) a conjugate transpose module configured to generate a conjugate transpose SH of an image matrix S that includes at least one training image vector, each of the at least one training image vector including a vectorized representation of the at least one training image; b) a power spectrum module configured to generate a power spectrum matrix Dx using at least one test image vector x, each of the at least one test image vector x including a vectorized representation of the at least one test image; and c) a correlation processor configured to generate correlation-peak-strength and distance-from-origin data using the conjugate transpose SH and power spectrum matrix Dx according to h=(Dx+ε
I)−
1S[SH(Dx+ε
I)−
1S]−
1d; andwherein; i) ε
is a regularization constant;ii) I is an identity matrix; and iii) Dx is a vector desired response with elements corresponding to at least one of the at least one training image vector in matrix S. - View Dependent Claims (12, 13, 14, 15, 16, 17, 18)
- a multitude of adaptive and robust correlation filters, each of the multitude of adaptive and robust correlation filters configured to generate correlation-peak-strength and distance-from-origin data using a multitude of related images, each of the multitude of related images representing a similar part of at least one larger image, at least one of the related images originating from at least one of the at least one test image, at least one of the related images originating from at least one of the at least one training image, wherein at least one of the multitude of adaptive and robust correlation filters includes;
Specification