IRAF help page for package obsolete, program oimcombine

• input
• output
• sigma =
• logfile = STDOUT
• option = average
• outtype =
• expname =
• exposure = no
• scale = no
• offset = no
• weight = no
• modesec =
• lowreject = 3., highreject = 3.
• blank = 0.

• V2.10

oimcombine input output

input

output

sigma =

logfile = STDOUT

option = average

      sum - Sum the input images
  average - Average the input images
   median - Median the input images
minreject - Reject the minimum value at each pixel
maxreject - Reject the maximum value at each pixel
minmaxrej - Reject the min and max values at each pixel
threshold - Reject pixels above and below thresholds
  sigclip - Apply a sigma clipping algorithm to each pixel
avsigclip - Apply a sigma clipping algorithm to each pixel

expname =

exposure = no

scale = no

offset = no

weight = no

modesec =

lowreject = 3., highreject = 3.

blank = 0.

The input images are combined by averaging or summing pixel by pixel to produce the output image. The input images may first be scaled to a common exposure time or mode or offset to a common mode before combining. The average may be weighted by statistical weights based on the exposure time or mode and the number of images previously combined. Deviant pixels may be detected and excluded from the average using one of a number of algorithms.

The input images are specified by a list. The images must all be the same size but may be of any dimension and pixel datatypes. If scaling by an exposure time it is given in the image headers with the name specified by the parameter expname.

The specified output image is created. Its header information is copied from the first image in the input image list. The pixel datatype may be specified by the parameter outtype. If no pixel datatype is specified then the maximum precedence datatype of the input images is used. The datatype precedences are short, integer, long, real, double, and complex in increasing order. Thus, mixed integer and real images default to a real output image. The internal arithmetic is real for all datatypes except double and complex which are done in the appropriate mode (the sigma clipping algorithms are not available for complex images). The output image header will also contain the parameter NCOMBINE giving the number of images combined. This may be used if the output image is later combined with other images. The exposure time, if specified, will be updated as the weighted average (or sum for the summing option) of the input images.

A sigma image is produced if a sigma image name is given. It is the standard deviation of the input images about the output image. Two or more images are required. The pixel datatype will be at least type real. A log is produced if a log file is specified. To print information on the standard output stream (normally the terminal) the name STDOUT is used. The log contains a time, the combining options and relevant parameters, a list of the input images, and weight parameters if used, and the name of the output image.

Except for summing the images may be scaled. There are currently three ways to scale the images. The images may be scaled to the same exposure time. If the background level or sensitivity is variable then the images may be either scaled or offset (by a constant) to the same mode. The mode is determined by sampling up to 100 points per dimension within the specified mode image section. A scaling correction is used when the overall brightness or sensitivity is varying. The offset correction is used when the background brightness is varying independently of the object brightness. When scaling by the mode, the exposure time (and setting of the exposure and offset parameters) is ignored. Except for medianing and summing, the images are combined by averaging. Combining with weighting and/or scaling is slower than if not weighting and scaling.

The relationships between the exposure time and modes determining the scaling, offsets, and weights used and printed in the log output are given below.

if scale = yes: scale = / M offset = 0. weight = (N*M)**1/2 / sum {(N*M)**1/2}

if exposure = yes and offset = no: scale = / T offset = 0. weight = (N*T)**1/2 / sum {(N*T)**1/2}

if exposure = no and offset = yes: scale = 1 offset = - M weight = (N/M)**1/2 / sum {(N/M)**1/2}

if exposure = yes and offset = yes: scale = / T offset = * ( - M/T) weight = (N*T/(M/T))**1/2 / sum {(N*T/(M/T))**1/2}

where M is the mode of an image, is the average mode over all images, N is the number of images previously combined (using the image header parameter NCOMBINE), T is the exposure time (with a minimum of 0.001 imposed), and is the average exposure time over all images. The operation **1/2 is the square root. Note that the scales average to 1, the offsets sum to zero, and the weights sum to 1.

The option parameter is the heart of the task. It selects from one of a number of algorithms for combining the images. The list below summarizes the algorithms.

	    sum - Sum the input images
	average - Average the input images
	 median - Median the input images
      minreject - Reject the minimum value at each pixel
      maxreject - Reject the maximum value at each pixel
      minmaxrej - Reject both the min. and max. values at each pixel
      threshold - Reject pixels above and below specified thresholds
	sigclip - Apply a sigma clipping algorithm to each pixel
      avsigclip - Apply a sigma clipping algorithm to each pixel

These algorithms are described in detail in the following sections. The choice of algorithm depends on the data, the number of images, and the importance of rejecting deviant pixels. The more complex the algorithm the more time consuming the operation.

The rejection parameters lowreject, highreject, and blank are used by the "threshold", "sigclip", and "avsigclip" algorithms for rejecting deviant pixels.

The input images are combined by summing. Care must be taken not to exceed the range of the short datatype when summing if the output datatype is "short". Summing is the only algorithm in which scaling and weighting are not used. Also no sigma image is produced.

The input images are combined by averaging. The images may be scaled and weighted. There is no pixel rejection. A sigma image is produced if there is more than one input image.

The input images are combined by medianing each pixel. Unless the images are at the same exposure level they should be scaled. The sigma image is based on all the input images and is only an approximation to the uncertainty in the median estimates. The median of N images is defined to be the (N+1)/2 th value of the sorted pixel values at each point in the images. Note that this is lower of the two middle values for an even number of images.

At each pixel the minimum, maximum, or both are excluded from the average. The images should be scaled and the average may be weighted. The sigma image requires at least two pixels after rejection of the extreme values. These are relatively fast algorithms and are a good choice if there are many images (>15).

The input images are combined with pixels above and below specified threshold values (before scaling) excluded. The images may scaled and the average weighted. If all pixels are rejected the specified blank value is output. The sigma image also has the rejected pixels excluded.

The input images are combined by applying a sigma clipping algorithm at each pixel. The images should be scaled. This only rejects highly deviant points and so includes more of the data than the median or minimum and maximum algorithms. It requires many images (>10-15) to work effectively. Otherwise the bad pixels bias the sigma significantly. The mean used to determine the sigmas is based on the "minmaxrej" algorithm to eliminate the effects of bad pixels on the mean. Only one iteration is performed and at most one pixel is rejected at each point in the image. After the deviant pixels are rejected the final mean is computed from all the data. The sigma image excludes the rejected pixels.

The input images are combined with a variant of the sigma clipping algorithm which works well with only a few images. The images should be scaled. For each line the mean is first estimated using the "minmaxrej" algorithm. The sigmas at each point in the line are scaled by the square root of the mean, that is a Poisson scaling of the noise is assumed. These sigmas are averaged to get a line estimate of the sigma. Then the sigma at each point in the line is estimated by muliplying the line sigma by the square root of the mean at that point. As with the sigma clipping algorithm only one iteration is performed and at most one pixel is rejected at each point. After the deviant pixels are rejected the file mean is computed from all the data. The sigma image excludes the rejected pixels.

The "avsigclip" algorithm is the best algorithm for rejecting cosmic rays, especially with a small number of images, but it is also the most time consuming. With many images (>10-15) it might be advisable to use one of the other algorithms ("maxreject", "median", "minmaxrej") because of their greater speed.

1. Simple combining with default values:

    cl> oimcombine im* imsum option=sum		# Sum images
    cl> oimcombine im* imave			# Average images
    cl> oimcombine im* immed option=median	# Median images
    cl> oimcombine im* immmr option=minmax	# Reject extreme values
    cl> oimcombine im* imsig option=sigclip	# Sigma clipping

2. To scale and weight by the exposure time given in the header by EXPTIME:

cl> oimcombine im* imave expname=exptime weight=yes

V2.10

o Vax 11/750 with FPA
o Size of images = 500 x 500 for 3-7 images
o Size of images = 500 x 50 for 50 and 100 images
  Times scaled to 500 x 500
o Input and output datatypes the same
o No sigma image output
o CPU time in seconds
			Short			 Real
Combine option	   3     5     7       3     5     7    50   100
--------------  ----- ----- -----   ----- ----- ----- ----- -----
			No weighting or scaling
sum		 15.6  22.0  28.3     8.9  13.1  18.6   270   520
average		 19.2  26.6  31.7    13.2  17.3  23.2   313   585
median		 16.9  56.2 106.0    13.0  49.3 100.6   779  1573
minreject	 30.0  39.3  51.0    22.3  33.2  43.6   449   886
maxreject	 29.0  39.6  51.3    22.6  33.6  44.7   462   885
minmaxreject	 30.6  41.1  53.4    24.7  35.8  46.3   456   891
threshold	 35.1  50.6  65.9    30.5  44.3  58.7   560  1093
sigclip		108.1 142.0 178.5    97.0 127.3 156.5  1012  1892
avsigclip	146.7 179.5 214.8   131.2 162.3 191.4  1053  1935
			Weighting and scaling
average		 33.2  49.2  66.7    28.3  43.9  61.6 
median		 42.9  95.1 161.9    35.2  85.3 151.9 
minreject	 52.6  78.2 103.4    50.0  80.4 100.8 
maxreject	 51.3  77.5 106.1    45.1  69.4  96.1 
minmaxreject	 54.2  81.1 109.2    48.6  78.5 106.4 
threshold	 56.0  86.1 116.5    53.0  80.2 111.0 
sigclip		161.7 218.3 284.8   141.0 198.3 259.6 
avsigclip	191.4 255.6 324.8   177.9 241.7 309.4 

LIMITATIONS Though this task is essentially not limited by the physical limits of the host (number of open files, amount of memory) there is a software limit in the IRAF virtual operating system of about 120 images which may be combined.

images.imcombine

imsum, ccdred.combine,