background[/qualifiers]
The method used consists in dividing the image in n equal small square boxes where the local background intensity is estimated. The event distribution in each box is checked for spatial uniformity and consistency with the expected statistical distribution (Poisson or Gauss, depending on image type). If inconsistency between the observed and expected distribution is found the box is rejected. The distribution of the average background values in all boxes is then compared to a Gaussian distribution with mean equal to the mean of all the background values found. All boxes where the background level is more than three standard deviations away from the mean are rejected. If more than 70 % of the boxes are rejected the background routine stops and an error message is generated. The value of the background intensity returned is the average of the values measured in all accepted boxes.
display all background boxes which have NOT been used for the estimation of the background
defines the size (in image pixels) of the boxes where the image background is estimated. Allowed background box sizes are powers of two, and values of 16, 32, 64, and 128 are recommended.
A P(D) distribution is obtained using the /npd=n option. n is the number of count rate intervals into which the distribution is divided. Use /plot_pd to get a plot. The results are written to a file pd.qdp, or this a file specified by /file_pd=.
Go through a range of background boxes and determine the optimum box size to give a minimum backgorund counting rate.
Plot the P(D) distribution written using background/npd=n.
Write the P(D) distribution to a file, otherwise the default pd.qdp is used.
Examples:
background/opt ! calculate the optimum background
background/box=32 ! calculate the average image background
! using square boxes of size=32 image pixels.
background/draw ! use the default box size and draw the
! boxes that have been rejected
background/npd=20/plot ! do a P(D) distribution into 20 count rate
! ct bins and plot the result