#3 gather Radon profile data
The detection of text will be easier if the typical
Radon profile of text lines is known.
A special tool might be used to gather data on the Radon
profiles of lines in typical scanned text images.
The profile of a typical line of European text looks
like an "_M_"-shaped peak and we would like to have a
more precise idea of the shape of the "_M_".
The peak statistics tool will assume that the input
bitmap contains about 10-20 lines of text (with little
or no equations or graphics; all lines have the same
font size). The tool will look for local maxima, compute
the average profile of those maxima, and print the data
on STDOUT.
This tool will be run on some scanned text images
appropriate for gathering statistics (simple and
good-quality scans of pure text in various languages),
and the results will be later averaged. The averaging
must take into account the varying resolutions and font
sizes in the text. I.e. only the relative sizes of the
"_M_"-shaped profile are to be obtained.
Possibly, another relevant datum is the relative
brightness of the "M" profile with respect to the
average brightness of the image.
