Converts various kinds of data to a pixel image

`as.im(X, …)` # S3 method for im
as.im(X, W=NULL, …,
eps=NULL, dimyx=NULL, xy=NULL,
na.replace=NULL)

# S3 method for owin
as.im(X, W=NULL, …,
eps=NULL, dimyx=NULL, xy=NULL,
na.replace=NULL, value=1)

# S3 method for matrix
as.im(X, W=NULL, …)

# S3 method for tess
as.im(X, W=NULL, …,
eps=NULL, dimyx=NULL, xy=NULL,
na.replace=NULL)

# S3 method for function
as.im(X, W=NULL, …,
eps=NULL, dimyx=NULL, xy=NULL,
na.replace=NULL,
stringsAsFactors=NULL,
strict=FALSE, drop=TRUE)

# S3 method for funxy
as.im(X, W=Window(X), …)

# S3 method for expression
as.im(X, W=NULL, …)

# S3 method for distfun
as.im(X, W=NULL, …,
eps=NULL, dimyx=NULL, xy=NULL,
na.replace=NULL, approx=TRUE)

# S3 method for nnfun
as.im(X, W=NULL, …,
eps=NULL, dimyx=NULL, xy=NULL,
na.replace=NULL, approx=TRUE)

# S3 method for data.frame
as.im(X, …, step, fatal=TRUE, drop=TRUE)

# S3 method for default
as.im(X, W=NULL, …,
eps=NULL, dimyx=NULL, xy=NULL,
na.replace=NULL)

X

Data to be converted to a pixel image.

W

Window object which determines the spatial domain and pixel array geometry.

…

Additional arguments passed to `X`

when `X`

is a function.

eps,dimyx,xy

na.replace

Optional value to replace `NA`

entries in the
output image.

value

Optional.
The value to be assigned to pixels inside the window,
if `X`

is a window.

strict

Logical value indicating whether to match formal arguments
of `X`

when `X`

is a function. If `strict=FALSE`

(the default), all the `…`

arguments are passed to
`X`

. If `strict=TRUE`

, only named arguments
are passed, and only if they match the names of formal arguments of
`X`

.

step

Optional. A single number, or numeric vector of length 2, giving the grid step lengths in the \(x\) and \(y\) directions.

fatal

Logical value indicating what to do if the resulting image
would be too large for available memory. If `fatal=TRUE`

(the
default), an error occurs. If `fatal=FALSE`

, a warning is
issued and `NULL`

is returned.

drop

Logical value indicating what to do if the result would
normally be a list of pixel images but the list contains
only one image.
If `drop=TRUE`

(the default), the pixel image is extracted
and the result is a pixel image.
If `drop=FALSE`

, this list is returned as the result.

stringsAsFactors

Logical value (passed to `data.frame`

)
specifying how to handle pixel values which
are character strings. If `TRUE`

, character values are
interpreted as factor levels. If `FALSE`

, they remain
as character strings. The default depends on the version of R.
See section *Handling Character Strings*.

approx

Logical value indicating whether to compute an approximate result at faster speed.

A pixel image (object of class `"im"`

),
or a list of pixel images,
or `NULL`

if the conversion failed.

By default, if the pixel value data are character strings, they will be
treated as levels of a factor, and the resulting image will be
factor-valued. To prevent the conversion of character strings to
factors, use the argument `stringsAsFactors=FALSE`

,
which is recognised by most of the
methods for `as.im`

, or alternatively set
`options(stringsAsFactors=FALSE)`

.

The argument `stringsAsFactors`

is a logical value (passed to `data.frame`

)
specifying how to handle pixel values which
are character strings. If `TRUE`

, character values are
interpreted as factor levels. If `FALSE`

, they remain
as character strings. The default values of `stringsAsFactors`

depends on the version of R.

In R versions

`< 4.1.0`

the factory-fresh default is`stringsAsFactors=FALSE`

and the default can be changed by setting`options(stringsAsFactors=FALSE)`

.in R versions >= 4.1.0 the default is

`stringsAsFactors=FALSE`

and there is no option to change the default.

This function converts the data `X`

into a pixel image
object of class `"im"`

(see `im.object`

).
The function `as.im`

is generic, with methods for the classes
listed above.

Currently `X`

may be any of the following:

a pixel image object, of class

`"im"`

.a window object, of class

`"owin"`

(see`owin.object`

). The result is an image with all pixel entries equal to`value`

inside the window`X`

, and`NA`

outside.a matrix.

a tessellation (object of class

`"tess"`

). The result is a factor-valued image, with one factor level corresponding to each tile of the tessellation. Pixels are classified according to the tile of the tessellation into which they fall.a single number (or a single logical, complex, factor or character value). The result is an image with all pixel entries equal to this constant value inside the window

`W`

(and`NA`

outside, unless the argument`na.replace`

is given). Argument`W`

is required.a function of the form

`function(x, y, ...)`

which is to be evaluated to yield the image pixel values. In this case, the additional argument`W`

must be present. This window will be converted to a binary image mask. Then the function`X`

will be evaluated in the form`X(x, y, ...)`

where`x`

and`y`

are**vectors**containing the \(x\) and \(y\) coordinates of all the pixels in the image mask, and`...`

are any extra arguments given. This function must return a vector or factor of the same length as the input vectors, giving the pixel values.an object of class

`"funxy"`

representing a`function(x,y,...)`

defined in a spatial region. The function will be evaluated as described above. The window`W`

defaults to the domain of definition of the function.an object of class

`"funxy"`

which also belongs to one of the following special classes. If`approx=TRUE`

(the default), the function will be evaluated approximately using a very fast algorithm. If`approx=FALSE`

, the function will be evaluated exactly at each grid location as described above.an object of class

`"distfun"`

representing a distance function (created by the command`distfun`

). The fast approximation is the distance transform`distmap`

.an object of class

`"nnfun"`

representing a nearest neighbour function (created by the command`nnfun`

). The fast approximation is`nnmap`

.an object of class

`"densityfun"`

representing a kernel estimate of intensity (created by the command`densityfun`

). The fast approximation is the Fast Fourier Transform algorithm in`density.ppp`

.an object of class

`"Smoothfun"`

representing kernel-smoothed values (created by the command`Smoothfun`

). The fast approximation is the Fast Fourier Transform algorithm in`Smooth.ppp`

.

An

`expression`

involving the variables`x`

and`y`

representing the spatial coordinates, and possibly also involving other variables. The additional argument`W`

must be present; it will be converted to a binary image mask. The expression`X`

will be evaluated in an environment where`x`

and`y`

are**vectors**containing the spatial coordinates of all the pixels in the image mask. Evaluation of the expression`X`

must yield a vector or factor, of the same length as`x`

and`y`

, giving the pixel values.a list with entries

`x, y, z`

in the format expected by the standard`R`

functions`image.default`

and`contour.default`

. That is,`z`

is a matrix of pixel values,`x`

and`y`

are vectors of \(x\) and \(y\) coordinates respectively, and`z[i,j]`

is the pixel value for the location`(x[i],y[j])`

.a point pattern (object of class

`"ppp"`

). See the separate documentation for`as.im.ppp`

.A data frame with at least three columns. Columns named

`x`

,`y`

and`z`

, if present, will be assumed to contain the spatial coordinates and the pixel values, respectively. Otherwise the`x`

and`y`

coordinates will be taken from the first two columns of the data frame, and any remaining columns will be interpreted as pixel values.

The spatial domain (enclosing rectangle) of the pixel image
is determined by the argument `W`

. If `W`

is absent,
the spatial domain is determined by `X`

.
When `X`

is a function, a matrix, or a single numerical value,
`W`

is required.

The pixel array dimensions of the final resulting image are determined by (in priority order)

the argument

`eps`

,`dimyx`

or`xy`

if present;the pixel dimensions of the window

`W`

, if it is present and if it is a binary mask;the pixel dimensions of

`X`

if it is an image, a binary mask, or a`list(x,y,z)`

;the default pixel dimensions, controlled by

`spatstat.options`

.

Note that if `eps`

, `dimyx`

or `xy`

is given, this will override
the pixel dimensions of `X`

if it has them.
Thus, `as.im`

can be used to change an image's pixel dimensions.

If the argument `na.replace`

is given, then all `NA`

entries
in the image will be replaced by this value. The resulting image is
then defined everwhere on the full rectangular domain, instead of a
smaller window. Here `na.replace`

should be a single value,
of the same type as the other entries in the image.

If `X`

is a pixel image that was created by an older version
of spatstat, the command `X <- as.im(X)`

will
repair the internal format of `X`

so that it conforms to the
current version of spatstat.

If `X`

is a data frame with `m`

columns,
then `m-2`

columns of data are interpreted as pixel values,
yielding `m-2`

pixel images. The result of
`as.im.data.frame`

is a list of pixel
images, belonging to the class `"imlist"`

.
If `m = 3`

and `drop=TRUE`

(the default), then the
result is a pixel image rather than a list containing this image.

If `X`

is a `function(x,y)`

which returns a matrix of
values, then `as.im(X, W)`

will be a list of pixel images.

Separate documentation for `as.im.ppp`

# NOT RUN { data(demopat) # window object W <- Window(demopat) plot(W) Z <- as.im(W) image(Z) # function Z <- as.im(function(x,y) {x^2 + y^2}, unit.square()) image(Z) # or as an expression Z <- as.im(expression(x^2+y^2), square(1)) # function with extra arguments f <- function(x, y, x0, y0) { sqrt((x - x0)^2 + (y-y0)^2) } Z <- as.im(f, unit.square(), x0=0.5, y0=0.5) image(Z) # Revisit the Sixties Z <- as.im(f, letterR, x0=2.5, y0=2) image(Z) # usual convention in R stuff <- list(x=1:10, y=1:10, z=matrix(1:100, nrow=10)) Z <- as.im(stuff) # convert to finer grid Z <- as.im(Z, dimyx=256) #' distance functions d <- distfun(redwood) Zapprox <- as.im(d) Zexact <- as.im(d, approx=FALSE) plot(solist(approx=Zapprox, exact=Zexact), main="") # pixellate the Dirichlet tessellation Di <- dirichlet(redwood) plot(as.im(Di)) plot(Di, add=TRUE, border="white") # as.im.data.frame is the reverse of as.data.frame.im grad <- bei.extra$grad slopedata <- as.data.frame(grad) slope <- as.im(slopedata) unitname(grad) <- unitname(slope) <- unitname(grad) # for compatibility all.equal(slope, grad) # TRUE ## handling of character values as.im("a", W=letterR, na.replace="b") as.im("a", W=letterR, na.replace="b", stringsAsFactors=FALSE) # }