lattice::levelplot() functions for hyperSpec objects. An image or map of a summary value of each spectrum is plotted. Spectra may be identified by mouse click.

plotmap(object, model = spc ~ x * y, func = mean, func.args = list(), ...)

# S4 method for formula,hyperSpec
levelplot(
x,
data,
transform.factor = TRUE,
...,
contour = FALSE,
useRaster = !contour
)

# S4 method for hyperSpec,missing
levelplot(x, data, ...)

map.identify(
object,
model = spc ~ x * y,
voronoi = FALSE,
...,
tol = 0.02,
warn = TRUE
)

plotvoronoi(object, model = spc ~ x * y, use.tripack = FALSE, mix = FALSE, ...)

## Arguments

object, data

the hyperSpec object

model, x

formula specifying the columns of object that are to be displayed by lattice::levelplot()

func, func.args

Before plotting, plotmap applies function func with the arguments given in the list func.args to each of the spectra. Thus a single summary value is displayed for each of the spectra.

This can be suppressed manually by setting func to NULL. It is automatically suppressed if .wavelength appears in the formula.

...

further arguments are passed down the call chain, and finally to lattice::levelplot()

transform.factor

If the color-coded variable is a factor, should trellis.factor.key() be used to compute the color coding and legend?

contour, useRaster

see lattice::levelplot()

voronoi

Should the plot for identifying spectra by mouse click be produced by plotmap (default) or plotvoronoi?

tol

tolerance for map.identify as fraction of the viewport (i.e. in "npc" units)

warn

should a warning be issued if no point is within the specified tolerance? See also details.

use.tripack

Whether package tripack should be used for calculating the Voronoi polygons. If FALSE, package deldir is used instead. See details.

mix

For Voronoi plots using package tripack, I experienced errors if the data was spatially ordered. Randomly rearranging the rows of the hyperSpec object circumvents this problem.

## Value

map.identify returns a vector of row indices into object of the clicked points. The other functions return a lattice object.

## Details

The model can contain the special column name .wavelength to specify the wavelength axis.

plotmap, map.identify, and the levelplot methods internally use the same gateway function to lattice::levelplot(). Thus transform.factor can be used with all of them and the panel function defaults to lattice::panel.levelplot.raster() for all three. Two special column names, .rownames and .wavelength may be used.

levelplot plots the spectra matrix.

plotvoronoi calls plotmap with different default settings, namely the panel function defaults to latticeExtra::panel.voronoi(). latticeExtra::panel.voronoi() depends on either of the packages 'tripack' or 'deldir' being installed. For further information, please consult the help page of latticeExtra::panel.voronoi(). On the faux_cell() data set, plotmap is roughly 5 times faster than plotvoronoi using tripack, and ca. 15 times faster than plotvoronoi using deldir. Package tripack, however, is free only for non-commercial use. Also, it seems that tripack version hang (R running at full CPU power, but not responding nor finishing the calculation) for certain data sets. In this case, mix = TRUE may help.

map.identify calls plotmap and plotvoronoi, respectively and waits for (left) mouse clicks on points. Other mouse clicks end the input.

Unlike lattice::panel.identify(), the indices returned by map.identify are in the same order as the points were clicked. Also, multiple clicks on the same point are returned as multiple entries with the same index.

map.identify uses option debuglevel similar to identify_spc(): debuglevel == 1 will plot the tolerance window if no data point was inside (and additionally labels the point) while debuglevel == 2 will always plot the tolerance window.

The map.sel.* functions offer further interactive selection, see map.sel.poly().

vignette(plotting), vignette(hyperSpec)

plot()

lattice::levelplot()

trellis.factor.key() for improved color coding of factors

hyperSpec options() identify_spc() map.sel.poly()

latticeExtra::panel.voronoi()

C. Beleites

## Examples

if (FALSE) {
vignette(plotting)
vignette(hyperSpec)
}

levelplot(spc ~ y * x, faux_cell[, , 1003]) # properly rotated
#> Warning: device support for raster images unknown, ignoring 'raster=TRUE'

plotmap(faux_cell[, , 1003])
#> Warning: device support for raster images unknown, ignoring 'raster=TRUE'

# plot spectra matrix
levelplot(spc ~ .wavelength * t, laser, contour = TRUE, col = "#00000080")

plotmap(faux_cell, region ~ x * y)
#> Warning: device support for raster images unknown, ignoring 'raster=TRUE'

# Voronoi plots
smpl <- sample(faux_cell, 300)
plotmap(smpl, region ~ x * y)
#> Warning: device support for raster images unknown, ignoring 'raster=TRUE'

if (require(tripack)) {
plotvoronoi(smpl, region ~ x * y)
}
#> Warning: device support for raster images unknown, ignoring 'raster=TRUE'

if (require(deldir)) {
plotvoronoi(smpl, region ~ x * y,
use.tripack = FALSE
)
}
#> deldir 1.0-6      Nickname: "Mendacious Cosmonaut"
#>
#>      The syntax of deldir() has had an important change.
#>      The arguments have been re-ordered (the first three
#>      are now "x, y, z") and some arguments have been
#>      eliminated.  The handling of the z ("tags")
#>      argument has been improved.
#>
#>      The "dummy points" facility has been removed.
#>      This facility was a historical artefact, was really
#>      of no use to anyone, and had hung around much too
#>      long.  Since there are no longer any "dummy points",
#>      the structure of the value returned by deldir() has
#>      changed slightly.  The arguments of plot.deldir()
#>      have been adjusted accordingly; e.g. the character
#>      string "wpoints" ("which points") has been
#>      replaced by the logical scalar "showpoints".
#>      The user should consult the help files.
#> Warning: device support for raster images unknown, ignoring 'raster=TRUE'