Pixel images are handled by many different software packages.
In virtually all of these, the pixel values are stored in a matrix,
and are accessed using the row and column indices of the matrix.
However, different pieces of software use different conventions for
mapping the matrix indices $[i,j]$ to the spatial coordinates
$(x,y)$.
-
In the Cartesian convention, the first matrix index $i$
is associated with the first Cartesian coordinate $x$,
and $j$ is associated with $y$. This convention is used in
image.default
.
-
In the European reading order convention, a matrix is displayed
in the spatial coordinate system as it would be printed in a page of text:
$i$ is effectively associated with the negative $y$ coordinate,
and $j$ is associated with $x$. This convention is used in some
image file formats.
-
In the
spatstat
convention, $i$
is associated with the increasing $y$ coordinate,
and $j$ is associated with $x$. This is also used in some
image file formats.
To convert between these conventions, use the function
transmat
. If a matrix m
contains pixel image data
that is correctly displayed by software that uses the Cartesian convention,
and we wish to convert it to the European reading convention, we can type
mm <- transmat(m, from="Cartesian", to="European")
.
The transformed matrix mm
will then be correctly displayed by
software that uses the European convention.
Each of the arguments from
and to
can be one of the names
"Cartesian"
, "European"
or "spatstat"
(partially matched)
or it can be a list specifying another convention. For example
to=list(x="-i", y="-j")!
specifies that rows of the output matrix
are expected to be displayed as vertical columns in the plot,
starting at the right side of the plot, as in the traditional
Chinese, Japanese and Korean writing order.