findpars
threshold = 4.0 (sigma)
The object detection threshold above local background in units of
datapars.sigma.
nsigma = 1.5
The semi-major axis of the Gaussian convolution kernel used
to computed the density enhancement and mean density images, in Gaussian sigma,
where gsigma = min (2.0, 0.42466 * nsigma * datapars.fwhmpsf /
datapars.scale) pixels.
ratio = 1.0
The ratio of the sigma of the Gaussian convolution kernel along the minor axis
direction to the sigma along the major axis direction.
Ratio defaults to 1.0 in which case the image is convolved with a circular
Gaussian.
theta = 0.0
The position of the major axis of the elliptical Gaussian. Theta is
measured counter-clockwise from the x axis.
sharplo = .2, sharphi = 1.0
Sharplo and sharphi are numerical cutoffs on the image sharpness
statistic, designed to eliminate
brightness maxima which are due to bad pixels, rather than to
astronomical objects.
roundlo = -1.0 roundhi = 1.0
Roundlo and roundhi are numerical cutoffs on the image roundness
statistic designed to eliminate
brightess maxima which are due to bad rows or columns, rather
than to astronomical objects.
mkdetections = no
Mark the positions of the detected objects on the image display?
This option does
not currently work on images displayed on the image display, but does work
if a contour plot is substituted for the displayed image.
The object detection algorithm parameters control the action of the detection algorithm. The default parameter values have been chosen to give reasonable results in the majority of cases. Note that some of the detection parameters scale with two quantities which are data dependent, the full-width half-maxima of the image point spread function and the standard deviation of the background pixels. Both of these quantities are parameters in the DATAPARS task.
In the first step of the detection process, the detection algorithm computes a linear function of the brightness values in an elliptical array of pixels within a semi-major axis of max (2.0, .42466 * nsigma * datapars.fwhmpsf / datapars.scale) pixels of the pixel of interest. The equivalent mathematical operation is a convolution of the original data with a truncated, lowered elliptical Gaussian function, specified by datapars.fwhmpsf, ratio and theta and computed in such a way as to be the mathematical equivalent of fitting a Gaussian stellar profile to the image data by least-squares. The coefficients of the linear function sum to zero so that the overall bias level (local sky) cancels out exactly. Since the function is symmetric about a single pixel, a smooth gradient in the sky brightness also cancels exactly. Therefore the user does not have to specify an absolute brightness threshold.
After the convolution step the detection algorithm goes through the convolved data looking for local maxima in the brightness enhancements, which are greater than threshold * datapars.sigma above background and brighter than any neighboring pixels which are within the bounds of the convolution kernel. As the algorithm finds potential candidates it computes a couple of shape characteristics. The limits on these parameters roundlo, roundhi, sharplo, sharphi are set to weed out bad pixels and brightess enhancements that are elongated in x and y.
Finally the approximate x and y centroids of the the detected objects are computed. A rough magnitude relative to threshold * datapars.sigma is computed and the object is added to the output file.
1. List the object detection parameters.
da> lpar findpars
2. Edit the object detection parameters.
da> findpars
3. Edit the FINDPARS parameters from within the DAOFIND task.
da> epar daofind ... edit a few daofind parameters ... move to the findpars parameter and type :e ... edit the findpars parameter and type:q ... finish editing the daofind parameters and type ^Z
4. Save the current FINDPARS parameter set in a text file fndnite1.par. This can also be done from inside a higher level task as in the previous example.
da> findpars ... edit the parameters ... type ":w fndnite1.par" from within epar
daofind