Finds the topologically-connected components of a spatial object, such as the connected clumps of pixels in a binary image.
connected(X, ...)# S3 method for owin
connected(X, ..., method="C", connect=8)
# S3 method for im
connected(X, ..., background = NA, method="C", connect=8)
A pixel image (object of class "im") with factor values.
The levels of the factor correspond to the connected components.
A spatial object such as a pixel image (object of class "im")
or a window (object of class "owin").
Optional. Treat pixels with this value as being part of the background.
String indicating the algorithm to be used. Either "C"
or "interpreted". See Details.
Arguments passed to as.mask to determine the
pixel resolution.
The connectivity of the pixel grid: either 8 or 4.
It may be hard to distinguish different components in the default plot because the colours of nearby components may be very similar. See the Examples for a randomised colour map.
The algorithm for method="interpreted"
can be very slow for large images (or images where
the connected components include a large number of pixels).
Original R code by Julian Burgos, University of Washington. Adapted for spatstat by Adrian Baddeley Adrian.Baddeley@curtin.edu.au and Rolf Turner rolfturner@posteo.net.
The function connected is generic, with methods
for pixel images (class "im") and windows (class "owin")
described here. There are also methods for
tessellations (connected.tess),
point patterns (connected.ppp and
connected.lpp),
and linear networks (connected.linnet).
The functions described here compute the connected component transform
(Rosenfeld and Pfalz, 1966)
of a binary image or binary mask. The argument X is first
converted into a pixel image with logical values. Then the algorithm
identifies the connected components (topologically-connected clumps
of pixels) in the foreground.
Two pixels belong to the same connected component if they have the value
TRUE and if they are neighbours.
This rule is applied repeatedly until it terminates.
Then each connected component
contains all the pixels that can be reached by stepping from neighbour
to neighbour.
Pixels are defined to be neighbours if they are physically adjacent to
each other. If connect=4, each pixel has 4 neighbours,
lying one step above or below, or one step to the left or right.
If connect=8 (the default), each pixel has 8 neighbours,
lying one step above or below, or one step to the left or right,
or one diagonal step away. (Pixels at the edge of the image have fewer
neighbours.) The 8-connected algorithm is the default
because it gives better results when the pixel grid is coarse. The
4-connected algorithm is faster and is recommended when the pixel grid
is fine.
If method="C", the computation is performed by a compiled C language
implementation of the classical algorithm of Rosenfeld and Pfalz
(1966). If method="interpreted", the computation is performed
by an R implementation of the algorithm of Park et al (2000).
The result is a factor-valued image, with levels that correspond to the connected components. The Examples show how to extract each connected component as a separate window object.
Park, J.-M., Looney, C.G. and Chen, H.-C. (2000) Fast connected component labeling algorithm using a divide and conquer technique. Pages 373-376 in S.Y. Shin (ed) Computers and Their Applications: Proceedings of the ISCA 15th International Conference on Computers and Their Applications, March 29-31, 2000, New Orleans, Louisiana USA. ISCA 2000, ISBN 1-880843-32-3.
Rosenfeld, A. and Pfalz, J.L. (1966) Sequential operations in digital processing. Journal of the Association for Computing Machinery 13 471-494.
connected.ppp,
connected.tess,
im.object,
tess
d <- distmap(cells, dimyx=256)
X <- levelset(d, 0.07)
plot(X)
Z <- connected(X)
plot(Z)
# or equivalently
Z <- connected(d <= 0.07)
# number of components
nc <- length(levels(Z))
# plot with randomised colour map
plot(Z, col=hsv(h=sample(seq(0,1,length=nc), nc)))
# how to extract the components as a list of windows
W <- tiles(tess(image=Z))
Run the code above in your browser using DataLab