Database Reference
In-Depth Information
Fig. 7.12
Segmentation of Akiyo sequence, obtained with and without the application of the HOG
model. (
a
) Segmentation without HOG. (
b
) HOG model. (
c
) Segmentation with HOG
Fig. 7.13
Segmentation of Bus sequence, (
a
) Frame no. 1 with the user mark, (
b
) the segmentation
result at Frame no. 10
from the similarity prior probability calculation, as they match the marked bus more
than the marked background, which is mostly green. Using a weighted value of
ʻ
in Eq. (
7.21
) can resolve the problem.
The foreman sequence (Fig.
7.14
) was segmented effectively for the face.
However, the hat and the background are very similar with the contour of the hat,
and the hat was not detected by using the user marking. As shown in Fig.
7.14
b, the
segmentation of the hat can be fixed and enhanced with a shape prior information,
in addition to the pairwise luminance difference
B
(
p
,
q
)
.
Search WWH ::
Custom Search