perm filename NOTE.MJH[206,JMC] blob
sn#544036 filedate 1980-10-24 generic text, type C, neo UTF8
COMMENT ā VALID 00002 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002 To Marsha Jo Hannah about a draft of her thesis
C00004 ENDMK
Cā;
To Marsha Jo Hannah about a draft of her thesis
You should introduce a notion of correlation scale which determines how
much reduction is allowed without losing the match. The correlation scale
differs for different parts of the picture, but matches found at a reduced scale
can guide the search for matches at a larger scale. An experiment to give the
scales of the different areas of one of your pairs would be desirable.
The object of adjoining measures of energy in a certain frequency band
or a measure of texture to the intensity in order to get a vector signal would
be to get a larger correlation scale. Incidentally, the correlation of each
component of the vector should be normalized separately since a good match in
even one component should suggest a closer look.