US 8,712,158 B2 · Keypoint descriptor generation by complex wavelet analysis

Abstract

The present invention relates to a method of generating a keypoint descriptor for identifying an object in an image or in a sequence of images, the keypoint descriptor being substantially invariant to a transformation of the object in the image. The method includes receiving object data representing an object for identification in an image; processing said object data to generate at least one basis function representing a feature having undergone at least one transformation or a transformation sequence across several consecutive frames, optionally using transformations that are out of the image plane to recognize objects from multiple views; modifying a prototype wavelet function based on said at least one basis function to generate a plurality of modified wavelet functions; comparing the plurality of modified wavelet functions with the at least one basis function; selecting a modified wavelet function of said plurality of modified wavelet functions based on the comparison of the plurality of modified wavelet functions with the at least one basis function; and processing an input image or input orientation field according to said selected modified wavelet function to generate the keypoint descriptor. The present invention further relates to a method of identifying an object in an image using a keypoint descriptor; and apparatus and computer software for implementing a method in accordance with the present invention.

Claims44

Dependent claims

A method of generating a keypoint descriptor for identifying an object in an image or in a sequence of images, the keypoint descriptor being substantially invariant to a transformation of the object in the image, wherein the method is performed using processing apparatus and includes:

a) receiving object data representing an object for identification in an image;
b) processing said object data to generate at least one basis function representing a feature having undergone at least one transformation or a transformation sequence across several consecutive frames, optionally using transformations that are out of the image plane to recognize objects from multiple views;
c) modifying a prototype wavelet function based on said at least one basis function to generate a plurality of modified wavelet functions;
d) comparing the plurality of modified wavelet functions with the at least one basis function;
e) selecting a modified wavelet function of said plurality of modified wavelet functions based on the comparison of step d);
and
f) processing an input image or input orientation field according to said selected modified wavelet function to generate the keypoint descriptor,
wherein the prototype wavelet function and the plurality of modified wavelet functions comprise a radial distance component and an azimuthal angle component,
wherein the prototype wavelet function and the plurality of modified wavelet functions have the formula:
Ψk(d,θ)=dnk−1e−αk d(nk−αkd) ·(cos(mkθ)+qxsin(mkθ))
where the term dnk−1e−αkd(nk−αkd) is the radial distance component and the term (cos(mkθ)+qx, sin(mkθ)) is the azimuthal angle component, and where the term qx, denotes an imaginary unit.

8 more elements

a) receiving object data representing an object for identification in an image;
b) processing said object data to generate at least one basis function representing a feature having undergone at least one transformation or a transformation sequence across several consecutive frames, optionally using transformations that are out of the image plane to recognise objects from multiple views;
c) modifying a prototype wavelet function based on said at least one basis function to generate a plurality of modified wavelet functions;
d) comparing the plurality of modified wavelet functions with the at least one basis function;
e) selecting a modified wavelet function of said plurality of modified wavelet functions based on the comparison of step d);
and
f) processing an input image or input orientation field according to said selected modified wavelet function to generate the keypoint descriptor,
wherein the object is a three dimensional object in a three dimensional image or the keypoint descriptor is arranged to identify an object in a series of images spaced over time,
wherein the prototype wavelet function and the plurality of modified wavelet functions comprise a radial distance component, an azimuthal angle component and a polar angle component,
wherein the prototype wavelet function and the plurality of modified wavelet functions have the formula:
Ψk(d,θ,φ,)=dnk−1e−αkd(nk−αkd) ·(cos(mkθ)+qxsin(mkθ)) ·(cos(lkφ)+qysin(lkφ))
where the term dnk-1e−αkd(nk−αkd) is the radial distance weighting component with radial distance, d, the term (cos(mkθ)+qx, sin(mkθ)) is the azimuthal angle component with azimuthal angle θ, and the term (cos(lkφ)+qysin(lkφ)) is polar angle component with polar angle φ, and where the terms qx, and qy, denote unit quaternions and obey the rules qx2=qy2=qz2=−1, and qx, qy=qz.

9 more elements

a) receiving object data representing an object for identification in an image;
b) processing said object data to generate at least one basis function representing a feature having undergone at least one transformation or a transformation sequence across several consecutive frames, optionally using transformations that are out of the image plane to recognise objects from multiple views;
c) modifying a prototype wavelet function based on said at least one basis function to generate a plurality of modified wavelet functions;
d) comparing the plurality of modified wavelet functions with the at least one basis function;
e) selecting a modified wavelet function of said plurality of modified wavelet functions based on the comparison of step d);
and
f) processing an input image or input orientation field according to said selected modified wavelet function to generate the keypoint descriptor,
wherein the prototype wavelet function and the modified wavelet functions are quaternion wavelet functions, the keypoint is arranged to identify an object in an image patch, the transformation comprises a rotation, the transformation comprises a translation, said at least one basis function is an eigenfunction, the object is a two-dimensional object in a temporal sequence of images or a three-dimensional object in three-dimensional volumetric data, or the modifying of step c) is performed in accordance with the formula:
α
k
,
m
k
,
n
k
,
l
k
=
arg
⁢
⁢
min
α
k
⁢
m
k
,
n
k
⁢
l
k
v
=
0
,
1
,
2
,
3
⁢
⌊

Re
⁢
{
S
⁡
(
v
)
⁢
Ψ
^
k
⁡
(
d
,
θ
,
ϕ
;
n
k
,
m
k
,
l
k
,
α
k
)
}
-
Ψ
j

2
⌋
where S(v)=δ(v)−δ(v−1)qx−δ(v−2)qy, −δ(v −3)qz, and,
where ∥Re{S(v){circumflex over (ψ)}k(d,θ,φ;nk,mk,lk,αk)}−ψj∥2, denotes find the square root of the sum of the squares of the function ƒ(nk,mk,lk,αk)=Re{S(v){circumflex over (ψ)}k(d,θ,φ;nk,mk,lk,αk)}−ψj, evaluated at predetermined discrete locations of d, θ and φ for candidate values of (nk,mk,lk,αk), and
where the terms qx, and qy, denote unit quaternions and obey the rules qx2=qy2=qz2=−1, and qx, qy=qz.

86 more elements

Dependencies are inferred from the claim text.

Keypoint descriptor generation by complex wavelet analysis

Life of this patent

Keypoint descriptor generation by complex wavelet analysis

Life of this patent

Abstract

Claims44