1.1.1.0 image.deviation

NRAO Home > CASA > CASA Toolkit Reference Manual

image.deviation - Function

1.1.1 Make an image based on a statistic from the input image’s pixel values, which in some cases can represent a deviation image of the original.

Description

This application creates an image that reflects the statistics of the input image. The output image has the same dimensions and coordinate system as the (selected region in) input image. The grid parameter describes how many pixels apart ”grid” pixels are. Statistics are computed around each grid pixel. Grid pixels are limited to the direction plane only; independent statistics are computed for each direction plane (ie at each frequency/stokes pixel should the input image happen to have such additional axes). Using the xlength and ylength parameters, one may specify either a rectangular or circular region around each grid point that defines which surrounding pixels are used in the statistic computation for individual grid points. If the ylength parameter is the empty string, then a circle of diameter provided by xlength centered on the grid point is used. If ylength is not empty, then a rectangular box of dimensions xlength x ylength centered on the grid pixel is used. These two parameters may be specified in pixels, using either numerical values or valid quantities with ”pix” as the unit (eg ”4pix”). Otherwise, they must be specified as valid angular quantities, with recognized units (eg ”4arcsec”). As with other region selections in CASA, full pixels are included in the computation even if the specified region includes only a fraction of that pixel. BEWARE OF MACHINE PRECISION ISSUES, because you may get a smaller number of pixels included in a region than you expect if you specify, eg, an integer number of pixels. In such cases, you probably want to specify that number plus a small epsilon value (eg ”2.0001pix” rather than ”2pix”) to mitigate machine precision issues when computing region extents.

The output image is formed by putting the statistics calculated at each grid point at the corresponding grid point in the output image. Interpolation of these output values is then used to compute values at non-grid-point pixels. The user may specify which interpolation algorithm to use for this computation using the interp parameter.

ANCHORING THE GRID

The user may choose at which pixel to ”anchor” the grid. For example, if one specifies grid=[4,4] and anchor=[0,0], grid points will be located at pixels [0,0], [0,4], [0,8] ... [4,0], [4,4], etc. This is exactly the same grid that would be produced if the user specified anchor=[4,4] or anchor=[20,44]. If the user specifies anchor=[1, 2] and grid=[4,4], then the grid points will be at pixels [1,2], [5,2], [9,2]... [5,2], [5,6], etc. and the resulting grid is the same as it would be if the user specified eg anchor=[9,10] or anchor=[21, 18]. The value ”ref”, which is the default, indicates that the reference pixel of the input image should be used to anchor the grid. The x and y values of this pixel will be rounded to the nearest integer if necessary.

SUPPORTED STATISTICS AND STATISTICS ALGORITHMS

One may specify which statistic should be represented using the stattype parameter. The following values are recognized (minimum match supported):

iqr inner quartile range (q3 - q1) max maximum mean mean medabsdevmed, madm median absolute deviation from the median median median min minimum npts number of points q1 first quartile q3 third quartile rms rms sigma, std standard deviation sumsq sum of squares sum sum var variance xmadm median absolute deviation from the median multipied by x, where x is the reciprocal of Phi1(3/4), where Phi1 is the reciprocal of the quantile function. Numerically, x = 1.482602218505602. See, eg, https://en.wikipedia.org/wiki/Median_absolute_deviation#Relation_to_standard_deviation

Using the statalg parameter, one may also select whether to use the Classical or Chauvenet/ZScore statistics algorithm to compute the desired statistic (see the help for ia.statistics() or imstat for a full description of these algorithms).

Arguments


Inputs
outfile	Output image file name. If left blank (the default), no image is written but a new image tool referencing the collapsed image is returned.
	allowed:	string
	Default:
region	Region selection. Default is to use the full image.
	allowed:	any
	Default:	variant
mask	Mask to use. Default setting is none.
	allowed:	string
	Default:
overwrite	Overwrite (unprompted) pre-existing output file? Ignored if ”outfile” is left blank.
	allowed:	bool
	Default:	false
stretch	Stretch the mask if necessary and possible? Default value is False.
	allowed:	bool
	Default:	false
grid	x,y grid spacing. Array of exactly two positive integers.
	allowed:	intarray
	Default:	11
anchor	x,y anchor pixel location. Either ”ref” to use the image reference pixel, or an array of exactly two integers.
	allowed:	variant
	Default:	variant ref
xlength	Either x coordinate length of box, or diameter of circle. Circle is used if ylength is empty string.
	allowed:	variant
	Default:	variant 1pix
ylength	y coordinate length of box. Use a circle if ylength is empty string.
	allowed:	variant
	Default:	variant 1pix
interp	Interpolation algorithm to use. One of ”nearest”, ”linear”, ”cubic”, or ”lanczos”. Minimum match supported.
	allowed:	string
	Default:	cubic
stattype	Statistic to compute. See full description for supported statistics.
	allowed:	string
	Default:	sigma
statalg	Statistics computation algorithm to use. Supported values are ”chauvenet” and ”classic”, Minimum match is supported.
	allowed:	string
	Default:	classic
zscore	For chauvenet, this is the target maximum number of standard deviations data may have to be included. If negative, use Chauvenet’s criterion. Ignored if algorithm is not ”chauvenet”.
	allowed:	double
	Default:	-1
maxiter	For chauvenet, this is the maximum number of iterations to attempt. Iterating will stop when either this limit is reached, or the zscore criterion is met. If negative, iterate until the zscore criterion is met. Ignored if algortihm is not ”chauvenet”.
	allowed:	int
	Default:	-1

Returns
image

Example

    # compute standard deviations in circles of diameter 10arcsec around
    # grid pixels spaced every 4 x 5 pixels and anchored at pixel [30, 40],
    # and use linear interpolation to compute values at non-grid-pixels
    ia.open("my.im")
    zz = ia.deviation("sigma.im", grid=[4, 5], anchor=[30, 40], xlength="10arcsec", stattype="sigma", interp="lin", statalg="cl")

    # compute median of the absolute deviations from the median values using
    # the z-score/Chauvenet algorithm, by fixing the maximum z-score to determine outliers to 5.
    # Use cubic interpolation to compute values for non-grid-point pixels. Use a rectangular region
    # of dimensions 5arcsec x 20arcsec centered on each grid point as the region in which to include
    # pixels for the computation of stats for that grid point.
    zz = ia.deviation("madm.im", grid=[4, 5], anchor=[30, 40], xlength="5arcsec", ylength="20arcsec, stattype="madm", interp="cub", statalg="ch", zscore=5)

__________________________________________________________________

[next] [prev] [prev-tail] [front] [up]

More information about CASA may be found at the CASA web page

This code is available under the terms of the GNU General Public Lincense