Granted Patent B2

US 9,946,955 · App. 14/883,158 · Granted Apr 17, 2018

Abstract

Claims (64)

1. An image registration method comprising:

obtaining a first reference image by photographing a reference target from a first viewpoint;

obtaining a second reference image by photographing the reference target from a second viewpoint;

obtaining reference coordinate-difference values that indicate difference in coordinates of corresponding pixels between the first reference image and the second reference image;

obtaining parallax registration-error values based on a distance of a photographing target from the first viewpoint and the second viewpoint;

obtaining a registration result of a first image of the photographing target taken from the first viewpoint and a second image of the photographing target taken from the second viewpoint based on the reference coordinate-difference values; and

correcting the registration result based on the parallax registration-error values,

wherein the first viewpoint is different from the second viewpoint, and

wherein the registration result is obtained after the parallax registration-error values are obtained.

2. The image registration method of claim 1 , wherein

a viewing angle, the distance of the photographing target, and a distance of the reference target are applied equally to the first viewpoint and the second viewpoint,

the reference coordinate-difference values and the parallax registration-error values are calculated by number of pixels,

the parallax registration-error values comprise a registration-error value of an x-axis of a screen and a registration-error value of a y-axis of the screen, and

the reference coordinate-difference values comprise an x-axis difference value and a y-axis difference value.

3. The image registration method of claim 2 , wherein the obtaining the reference coordinate-difference values comprises:

obtaining a first projected reference image by converting three-dimensional (3D) components of the first reference image into two-dimensional (2D) components;

obtaining a second projected reference image by converting 3D components of the second reference image into 2D components; and

calculating difference in coordinates of corresponding pixels between the first projected reference image and the second projected reference image.

4. The image registration method of claim 3 , wherein the obtaining the parallax registration-error values comprises:

calculating a parallax registration-error value Ppara of the x-axis by an equation

Ppara

=

(

Dobj

⁢

⁢

2

Dobj

⁢

⁢

1

-

1

)

×

Dcam

×

Pfhd

2

×

Dobj

⁢

⁢

2

×

tan

⁡

(

Tcam

2

)

,

wherein the distance of the reference target is Dobj 1 , the distance of the photographing target is Dobj 2 , a distance between the first viewpoint and the second viewpoint is Dcam, a number of pixels in a horizontal line of the first image or the second image is Pfhd, and the viewing angle is Tcam.

5. The image registration method of claim 3 , wherein the obtaining the registration result comprises:

obtaining a 2D homography matrix for position conversion of the respective pixels of the second reference image using the first projected reference image and the second projected reference image; and

registering the second image with the first image based on the obtained 2D homography matrix.

6. The image registration method of claim 5 , wherein the registering the second image with the first image comprises converting an x-coordinate value and a y-coordinate value of a pixel of the second image,

wherein the converting comprises:

obtaining a first x-coordinate value and a first y-coordinate value as a registration result of the pixel by substituting an x-coordinate difference value and a y-coordinate difference value of the pixel for elements of the obtained 2D homography matrix;

obtaining an x-coordinate correction value and a y-coordinate correction value of the pixel by substituting the registration-error value of the x-axis and the registration-error value of the y-axis for elements in the obtained the 2D homography matrix;

obtaining a second x-coordinate value by correcting the first x-coordinate value based on the x-coordinate correction value;

obtaining a second y-coordinate value by correcting the first y-coordinate value based on the y-coordinate correction value; and

setting the second x-coordinate value and the second y-coordinate value as final coordinate values of the pixel.

7. The image registration method of claim 1 , wherein the first image is taken by a first camera and the second image is taken by a second camera.

Assignments (8)