Skip to contents

Perspective Transformation Between Two Planes

Source: R/transform.R
findHomography.Rd

findHomography computes a perspective transformation between two planes.

Usage

findHomography(
  from,
  to,
  from_dim,
  to_dim = from_dim,
  method = "RANSAC",
  ransac_reproj_th = 3,
  max_it = 2000,
  conf = 0.95
)

Arguments

from

A Nx2 matrix indicating the location (x, y) of N points in the source image.

to

A Nx2 matrix indicating the location (x, y) of N points in the destination image. The order of the points must correspond to the order in from.

from_dim

A vector which first two elements indicate the number of rows and columns of the source image.

to_dim

A vector which first two elements indicate the number of rows and columns of the destination image. If not specified, from_dim will be used as a default.

method

A character string indicating the method used to compute a homography matrix. It can be one of the followings: "LS" (least-square), "RANSAC" (RANSAC-based robust method; the default), "LMEDS" (Least-Median robust method), or "RHO" (PROSAC-based robust method).

ransac_reproj_th

Maximum allowed reprojection error to treat a point pair as an inlier (used in the RANSAC and RHO methods only). If `from` and `to` are measured in pixels, it usually makes sense to set this parameter somewhere in the range of 1 to 10.

max_it

The maximum number of RANSAC iterations.

conf

Confidence level, between 0 and 1.

Value

A 3x3 homography matrix.

See also

getPerspectiveTransform, getAffineTransform, warpPerspective

Author

Simon Garnier, garnier@njit.edu

Examples

src <- cbind(x = c(0, 1, 1, 0), y = c(0, 0, 1, 1))
dst <- src + 1
findHomography(src, dst, c(10, 10))
#>      [,1] [,2] [,3]
#> [1,]    1    0    1
#> [2,]    0    1   -1
#> [3,]    0    0    1