NAME

ringpars -- determine the parameters of the calibration rings

USAGE

ringpars input output

PARAMETERS

input

the list of image names for the ring calibrations

output

the name of the text file to hold the ring parameters

lambda = 0

the wavelength of the standard illuminating the instrument for this ring. This parameter can only be used when a single input image is input, and must be used in conjuntion with the parameter "z". When several input images are given, the "lzfile" parameter should be used.

z = 0

the etalon gap parameter, for the instrument when obtaining the single input image. See also "lambda". The value for "z" may be in either imstrument units (near 0.5) or user units (near 500), and will be converted to instrument units on output.

lzfile = ""

A text file containing pairs of (lambda z) combinations, 1 pair per line, separated by a space or tab, and in linear association with the input image file list. This parameter is required if multiple input images are given.

average = 7

the number of image rows and columns to average while performing the analysis. For noisy data, this may need to be increased.

xstart = 0

the initial guess for the central column of the ring. If zero, the central column of the image is used.

ystart = 0

the initial guess for the central row of the ring. If zero, the central row of the image is used.

cwidth = 31

the width of the centroided region in pixels. After the peaks in the ring are located, a centroided center is found extending over this width in total.

contrast = 0.1

the minimum ratio between the strongest and weakest peaks allowable, when searching for the 2 highest peaks in a row or column cut for the ring centers.

separation = 40

the minimum number of pixels between the two highest peaks. If the rings have very small radii this may be too large a number.

threshold = 0.0

the minimum pixel value to be considered a valid peak.

verbose = no

if set to yes, the ring parameters will be printed on the terminal.

interactive = yes

if set to yes, the user may graphically view the results of the ring parameter search, and override if desired.

cursor = ""

the graphics cursor.

DESCRIPTION

The Fabry-Perot passes a single wavelength of light for a given etalon setting when used in parallel light, but due to the angles of rays coming from off-axis locations in the telescope field, the wavelength passed is a function of distance from the optical axis, as well as the etalon gap. This relationship is calibrated by illuminating the instrument with light from a standard lamp having a known wavelength. The light will fall on the detector only in a ring about the optical axis, having a radius determined by the gap setting and the wavelength of the light.

By varying the etalon setting, and illuminating with light from a lamp producing several wavelengths of light, the dispersion relations can be well determined, and are of the form:

lambda = (A + B*z) * (Cos (Atan (R/C)))

where the constants A, B, and C are to be determined, and R is the radius of the ring, and z is the etalon setting.

RINGPARS determines the ring radii by identifying the peaks in row and column cuts. The peak finder from the IDENTIFY package is used, followed by a centroiding centering algorithm. The center of the ring is deduced to be the midway point between the centers of the two peaks. A second iteration is performed at these revised row and column cuts. Two iterations appears to be adequate, even when the initial starting guesses are far from correct. The final radius is coputed by averaging the row and column diameters. It is assumed that the rings are circularly symmetric.

Upon exit, the parameters wavelength, etalon gap, radius, x-center, and y-center are written to a text file for later use by FITRINGS, and ZEROPT. The wavelength and gap are taken from the file indicated by the parameter "lzfile" and may look like:

6598.953 574
6598.953 582
6606.123 557
6606.123 592

where the first column indicates the wavelength of the lamp illumination, and the second is the etalon gap setting, in this case, in user units (the 574 value, for example, could also be given as 0.574). One line of "lz" values is needed for each input calibration image.

When the parameter "interactive" is selected, a plot of the rows and columns are displayed and the user may override the marked centers. This is frequently necessary for noisy data, or rings having small diameters. The following keystrokes are in effect when the cursor is presented (in addition to the usual system wide capital keystrokes):

c

print cursor position

h

set the high (rightmost) peak to the current cursor position

l

set the low (leftmost) peak to the current cursor position

i

reset interactive mode to no. To be used when you have many images, and you've gotten tired of looking at good automatic results.

v

print current values for the low and high peaks

?

prints minimal help

q

quit and accept the current low and high peak values for the current image.

EXAMPLES

To find the ring parameters for the neon images, prepare a file containing the lambda,z pairs called neonlz. Then run:

cl> ringpars neon*.imh neonout lzfile=neonlz

TIME REQUIREMENTS

For a 512x512 image, the time is governed by the column extraction time, which is about 3 seconds.

BUGS

The peaks are occasionally found incorrectly, especially when the peak is asymmetric. Varying the "cwidth" parameter can improve the results.

SEE ALSO

FITRING, IDENTIFY