US 9,286,656 B2Granted

Homography estimation apparatus and method

AssigneeChung-Ang University Industry-Academy Cooperation Foundation

InventorsSangkeun Lee, Sangwon Seo, Soowoong Jeong

Life of this patent

Estimating the term…

Loading patent…

Abstract

Disclosed are a homography estimation apparatus and method. A down-sampling unit reduces the reference image and the target image to the same size to generate down-sampled images. A feature point extraction unit extracts feature points from the reference image and the target image down-sampled to the same size, respectively. An outlier removal unit matches the feature points extracted from the reference image with the feature points extracted from the target image, and detects and removes outliers, which are feature points that do not match, from the matched feature points. A homography estimation unit estimates a homography using feature points from which outliers are removed. A reference image correction unit corrects the reference image using the estimated homography.

Claims9

Dependent claims

A homography estimation apparatus comprising:

a processor configured to execute operations as follows:
(a) reduce a reference image and a target image;
(b) generate a down-sampled image of the reference image and a down-sampled image of the target image;
(c) extract feature points from the down-sampled image of the reference image and the down-sampled image of the target image, respectively;
(d) match the feature points extracted from the down-sampled image of the reference image with the feature points extracted from the down-sampled image of the target image to detect and remove outliers, which are feature points that do not match, from the matched feature points;
(e) estimate a homography using feature points from which the detected outliers have been removed;
and
(f) repeat execution of operations (a) through (e) n times, where n is a positive integer greater than or equal to two, wherein
in each of the n iterations of operations (a) through (e), the processor reduces the reference image and the target image to a different size, and estimates a homography using the different reduced sizes of the reference image and the target image, resulting in n different homographies, and
the processor calculates a homography error for each of the estimated n homographies, the homography error being a difference between coordinates of the reference image and coordinates of the estimated homography, selects a homography corresponding to the smallest error among the homography errors, and corrects the reference image using the selected homography based on an equation:
H
=
[
1
0
0
0
1
0
0
0
1
/
m
]
⁢
H
′
⁡
[
1
0
0
0
1
0
0
0
m
]
=
[
h
11
h
12
h
13
h
21
h
22
h
23
h
31
h
32
h
33
]
,
wherein H is a corrected homography, H′ is a homography calculated from a down-sampled image, m is a scale factor that indicates the extent to which an image is reduced when H′ is calculated, h13 and h23 are values obtained by multiplying by m, and h31 and h32 are values obtained by multiplying by 1/m.

60 more elements

A homography estimation method performed by a homography estimation apparatus, comprising:

(a) reducing a reference image and a target image;
(b) generating a down-sampled image of the reference image and a down-sampled image of the target image;
(c) extracting feature points from the down-sampled image of the reference image and the down-sampled image of the target image, respectively;
(d) matching the feature points extracted from the down-sampled image of the reference image with the feature points extracted from the down-sampled image of the target image to detect and remove outliers, which are feature points that do not match, from the matched feature points;
(e) estimating a homography using feature points from which the detected outliers have been removed;
and
(f) repeat execution of operations (a) through (e) n times, where n is a positive integer greater than or equal to two, wherein
in each of the n iterations of operations (a) through (e), the reference image and the target image are reduced to a different size, and a homography is estimated using the different reduced sizes of the reference image and the target image, resulting in n different homographies, and
the homography estimation method further comprises:
calculating a homography error for each of the estimated n homographies, the homography error being a difference between coordinates of the reference image and coordinates of the estimated homography, selecting a homography corresponding to the smallest error among the homography errors, and correcting the reference image using the selected homography based on an equation:
H
=
[
1
0
0
0
1
0
0
0
1
/
m
]
⁢
H
′
⁡
[
1
0
0
0
1
0
0
0
m
]
=
[
h
11
h
12
h
13
h
21
h
22
h
23
h
31
h
32
h
33
]
,
wherein H is a corrected homography, H′ is a homography calculated from a down-sampled image, m is a scale factor that indicates the extent to which an image is reduced when H′ is calculated, h13 and h23 are values obtained by multiplying by m, and h31 and h32 are values obtained by multiplying by 1/m.

60 more elements

Dependencies are inferred from the claim text.

Facts

Application: 14/134,501Filed 19 Dec 2013
Priority: KR 10-2012-014947820 Dec 2012
Published: US 2014/0177968 A126 Jun 2014
Granted: 15 Mar 2016
Term ends: Not estimatedThe source records don’t support an estimate.

Classification7 codes

G06T 3/14
Geometric image transformations in the plane of the image: transformations for image registration, e.g. adjusting or mapping for alignment of images
G06T 3/4038
Scaling of whole images or parts thereof, e.g. expanding or contracting: image mosaicing, e.g. composing plane images from plane sub-images
G06T 3/4084
Scaling of whole images or parts thereof, e.g. expanding or contracting: in the transform domain, e.g. fast Fourier transform [FFT] domain scaling
G06T 7/337
Using feature-based methods: involving reference images or patches