Optical Flow Using Farneback's Algorithm

Computes a dense optical flow using the Gunnar Farneback’s algorithm.

Usage

farneback(
  image1,
  image2,
  pyr_scale = 0.5,
  levels = 3,
  winsize = 43,
  iterations = 3,
  poly_n = 7,
  poly_sigma = 1.5,
  use_init = FALSE,
  Gaussian = FALSE,
  target = "new"
)

Arguments

image1

A single-channel, 8U Image object.

image2

A single-channel, 8U Image object.

pyr_scale

Parameter, specifying the image scale (<1) to build pyramids for each image; pyr_scale = 0.5 means a classical pyramid, where each next layer is twice smaller than the previous one.

levels

Number of pyramid layers including the initial image; levels = 1 means that no extra layers are created and only the original images are used.

winsize

Averaging window size; larger values increase the algorithm robustness to image noise and give more chances for fast motion detection, but yield more blurred motion field.

iterations

Number of iterations the algorithm does at each pyramid level.

poly_n

Size of the pixel neighborhood used to find polynomial expansion in each pixel; larger values mean that the image will be approximated with smoother surfaces, yielding more robust algorithm and more blurred motion field, typically poly_n = 5 or 7.

poly_sigma

Standard deviation of the Gaussian that is used to smooth derivatives used as a basis for the polynomial expansion; for poly_n = 5, you can set poly_sigma = 1.1, for poly_n = 7, a good value would be poly_sigma = 1.5.

use_init

A logical indicating whether the content of target should be used as an initial flow approximation.

Gaussian

A logical indicating whether to use a Gaussian filter instead of a box filter for optical flow estimation; usually, this option gives a more accurate flow than with a box filter, at the cost of lower speed; normally, winsize for a Gaussian window should be set to a larger value to achieve the same level of robustness.

target

The location where the results should be stored. It can take 3 values:

"new":

a new Image object is created and the results are stored inside (the default).

An Image object:

the results are stored in another existing Image object. This is fast but will replace the content of target. target must be a two-channel, 32F Image object with the same dimensions as image1.

Value

If target="new", the function returns an Image object. If target is an Image object, the function returns nothing and modifies that Image object in place. The Image object will have two channels corresponding to the x and y coordinates (respectively) of the optical vector at each location in the image.

References

Farnebäck G. Two-Frame Motion Estimation Based on Polynomial Expansion. In: Bigun J, Gustavsson T, editors. Image Analysis. Springer Berlin Heidelberg; 2003. pp. 363–370. doi:10.1007/3-540-45103-X_50

See also

Image

Author

Simon Garnier, garnier@njit.edu

Examples

balloon <- video(system.file("sample_vid/Balloon.mp4", package = "Rvision"))
#> Error: Could not open the video.
balloon1 <- readFrame(balloon, 1)
#> Error in eval(expr, envir, enclos): object 'balloon' not found
changeColorSpace(balloon1, "GRAY", "self")
#> Error in eval(expr, envir, enclos): object 'balloon1' not found
balloon2 <- readFrame(balloon, 2)
#> Error in eval(expr, envir, enclos): object 'balloon' not found
changeColorSpace(balloon2, "GRAY", "self")
#> Error in eval(expr, envir, enclos): object 'balloon2' not found
of <- farneback(balloon1, balloon2)
#> Error in eval(expr, envir, enclos): object 'balloon1' not found