
Saunder August 4, 2015 16:40 K23081˙C003
30
Video Cataloguing: Structure Parsing and Content Extraction
Figure 3.2 Illustration of an FR. The frame is partitioned into 8 ×10 subregions,
and the outermost round subregions are nonfocus regions, the second outermost
round subregions are the second focus regions, and the remaining subregions are
the most focused regions.
In order to efficiently represent the frame with the chosen subset of pixels that
exclude the unimportant pixels, we remove pixels in the nonfocus region, fully use
pixels in the most focused region, and select pixels with odd x-coordinates in the
second focus region. At last, we can get ...