Abstract
Fundamental techniques and commonly employed algorithms for image manipulation of
endoscopic images were reviewed, but only in a few instances was any novel information
obtained. New software was recently developed for medium-frequency band enhancement:
From the correlation distribution of image data in the RGB color space (distribution
1), only those data components depicting the fine structure patterns were extracted.
The correlation distribution of the latter components was subsequently studied in
the same space (distribution 2); the original image data were so processed that distribution
1 would be enhanced along the axis of greatest variation seen in distribution 2. This
algorithm of image processing was successful in detecting minute surface structures
of the mucosa hidden by the overlying mucus in the original images. It seems to be
the first successful algorithm of image processing that obtained specific effects
of enhancement without simultaneous noise enhancement.
