NAME:
   getdigits
 PURPOSE:   (one line only)
   Decode a single image with an IOTA-VTI time code
 DESCRIPTION:
   Scan image for IOTA-VTI time code information and decode into a
     a machine-readable form.  This can handle IOTA-VTI output only and
     only in either 720x480 or 640x480 pixel formats.
 CATEGORY:
  Video data processing
 CALLING SEQUENCE:
  getdigits,image,odd,even,otime,onum,etime,enum
 INPUTS:
  image - 3 x N x M byte image to decode
 OPTIONAL INPUT PARAMETERS:
 KEYWORD INPUT PARAMETERS:
  DEBUG - Flag, if set will generate copious output and graphics.
 OUTPUTS:
  odd   - String array with the one character decoded from each cell
             from the odd field.
  even  - String array with the one character decoded from each cell
             from the even field.
  otime - String with the odd field time, HH:MM:SS.ssss, empty if decode
             was not successful.
  onum  - Long integer with the odd field number
  etime - String with the even field time, HH:MM:SS.ssss, empty if decode
             was not successful.
  enum  - Long integer with the even field number
 KEYWORD OUTPUT PARAMETERS:
 COMMON BLOCKS:
 SIDE EFFECTS:
 RESTRICTIONS:
  Image must be 3x720x480 or 3x640x480.  Only plane 0 is read.

  Requires access to two files:
    digits_640_480.fits and digits_720_480.fits
  These files must be somewhere in your IDL_PATH but usual reside in the
    same directory as my IDL library.
 PROCEDURE:
  A hueristic procedure is used to determine the row location of the
   time code data.  The x location of the cells where characters are
   found is hard coded.  A row-summed version of the image shows a
   characteristic three-peak shape.  The middle peak is used as the center
   of the digits written to the video.  The position is required to be
   in a small range near the bottom of the image.  The results are not
   validated in any way.  The return is an array of characters and if nothing
   was found, the character is a empty string.

  The time codes are read from the even and odd fields of the images as well
   as the field counter.

 MODIFICATION HISTORY:
  2016/03/08, Written by Marc W. Buie, Southwest Research Institute
  2016/03/25, MWB, added DEBUG, XOFF, and YOFF keywords
  2016/09/01, MWB, added support for 2-d images and improved reliability