Method and apparatus for calibrating projector-camera system
First Claim
1. A computer implemented method for automatically calibrating a system having a camera with a camera image frame, a projector with a source image frame and a projection surface with a projection image frame, comprising the steps of:
- arbitrarily placing the camera and the projector;
intersecting a camera field of view of the camera with a projector field of projection of the projector on the projection surface;
projecting an image of a calibration pattern with a plurality of calibration pattern feature points onto the projection surface;
capturing the image of the calibration pattern by the camera;
identifying locations of calibration pattern feature points in the camera image frame of the captured image using a feature extraction algorithm; and
obtaining parameters for mapping utilizing techniques of linear algebra, given a location of a calibration pattern feature point in the source image frame and a corresponding location of the calibration pattern feature point in the camera image frame such that the system is calibrated, wherein the location of the calibration pattern feature point, i, in the camera image frame is (Xi, Yi), the corresponding location of the calibration pattern feature point in the source image frame is (xi,yi), the parameters of the mapping are represented by {overscore (p)}=(p1. . . p9), {overscore (p)} is a unit vector resulting in eight degrees of freedom, the {overscore (p)} that best maps the calibration pattern feature point from the camera image frame to the source image frame is given by the eigenvector corresponding to the smallest eigenvalue of the matrix B, and given the mapping {overscore (p)}, any given point (X, Y) in the camera image frame is transformable to its corresponding point (x,y) in the source image frame by the following equation,
1 Assignment
0 Petitions
Accused Products
Abstract
The present invention enables a user to automatically calibrate a projector-camera system to recover the mapping from a given point in the source (pre-projection) image and its corresponding point in the camera image, and vice versa. One or more calibration patterns are projected onto a flat surface with possibly unknown location and orientation by a projector with possibly unknown location, orientation and focal length. Images of these patterns are captured by a camera mounted at a possibly unknown location, orientation and with possibly unknown focal length. Parameters for mapping between the source image and the camera image are computed. The present invention can become an essential component of a projector-camera system, such as automatic keystone correction and vision-based control of computer systems.
256 Citations
46 Claims
-
1. A computer implemented method for automatically calibrating a system having a camera with a camera image frame, a projector with a source image frame and a projection surface with a projection image frame, comprising the steps of:
-
arbitrarily placing the camera and the projector;
intersecting a camera field of view of the camera with a projector field of projection of the projector on the projection surface;
projecting an image of a calibration pattern with a plurality of calibration pattern feature points onto the projection surface;
capturing the image of the calibration pattern by the camera;
identifying locations of calibration pattern feature points in the camera image frame of the captured image using a feature extraction algorithm; and
obtaining parameters for mapping utilizing techniques of linear algebra, given a location of a calibration pattern feature point in the source image frame and a corresponding location of the calibration pattern feature point in the camera image frame such that the system is calibrated, wherein the location of the calibration pattern feature point, i, in the camera image frame is (Xi, Yi), the corresponding location of the calibration pattern feature point in the source image frame is (xi,yi), the parameters of the mapping are represented by {overscore (p)}=(p1. . . p9), {overscore (p)} is a unit vector resulting in eight degrees of freedom, the {overscore (p)} that best maps the calibration pattern feature point from the camera image frame to the source image frame is given by the eigenvector corresponding to the smallest eigenvalue of the matrix B, and given the mapping {overscore (p)}, any given point (X, Y) in the camera image frame is transformable to its corresponding point (x,y) in the source image frame by the following equation, - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13)
thresholding by intensity the pixel to create a binary image such that a centroid of the pixel is the location of a corresponding calibration pattern feature point.
-
-
6. The method of claim 1 wherein the calibration pattern includes four differently colored spots on a black background, with each colored spot located near a corner of the source image frame.
-
7. The method of claim 1 wherein the calibration pattern includes a set of images, with each image formed as a differently sized white rectangle on a black background.
-
8. The method of claim 7 further including the step of:
computing intersections of lines of corners of each differently sized white rectangle.
-
9. The method of claim 1 wherein the feature extraction algorithm includes the steps of:
-
thresholding by intensity the pixel to create a binary image;
utilizing a connected components algorithm to identify the largest region of bright pixels in the image corresponding to the differently sized white rectangle in the calibration pattern;
extracting the edges of the largest region; and
computing the intersections between adjacent edges, yielding the locations of the corners to sub-pixel accuracy.
-
-
10. The method of claim 1 wherein the location of the calibration pattern feature points is manually adjustable by a user.
-
11. The method of claim 1 wherein inverse mapping from the source image frame to the camera image frame exchanges (xi,yi) and (Xi, Yj) in all cases.
-
12. The method of claim 1 wherein the projection surface includes a plurality of planar patches, a calibration pattern having at least four calibration pattern feature points is projected onto each of the planar patches, and parameters for mapping each planar patch is independently derived.
-
13. The method of claim 1 wherein the parameters for mapping are restricted between calibration pattern feature points in the camera image frame and the corresponding calibration pattern feature points in the projection image frame.
-
14. An automatic calibration system comprising:
-
a camera with a camera image frame and a camera field of view;
a projector with a source image frame and a projector field of projection;
a projection surface with a projection image frame;
a means for performing a feature extraction algorithm; and
a means for mapping parameters utilizing techniques of linear algebra, wherein the camera and the projector are arbitrarily placed, a camera field of view of the camera intersects a projector field of projection of the projector on the projection surface, the projector projects an image of a calibration pattern with a plurality of calibration pattern feature points onto the projection surface, the camera captures the image of the calibration pattern, the locations of calibration pattern feature points in the camera image frame of the captured image are identified using the feature extraction algorithm, and the parameters for mapping are obtained, given a location of a calibration pattern feature point in the source image frame and a corresponding location of the calibration pattern feature point in the camera image frame;
wherein the location of the calibration pattern feature point, i, in the camera image frame is (Xi, Yi), the corresponding location of the calibration pattern feature point in the source image frame is (xi,yi), the parameters of the mapping are represented by {overscore (p)}=(p1. . . p9), {overscore (p)} is a unit vector resulting in eight degrees of freedom, the {overscore (p)} that best maps the calibration pattern feature point from the camera image frame to the source image frame is given by the eigenvector corresponding to the smallest eigenvalue of the matrix B, and given the mapping {overscore (p)}, any given point (X, Y) in the camera image frame is transformable to its corresponding point (x,y) in the source image frame by the following equation, - View Dependent Claims (15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46)
-
Specification