4. Image analysis
To compare the contrast between the frozen/defrosted and raw
fruit in the transmission and dark-field signal, respectively, an
image contrast measure is introduced in the form of a contrast-tonoise
ratio (CNR) which is defined as (Song et al., 2004):
CNR ¼ jmF mRj=
wFs2F
þ wRs2
R
1=2
where m denotes the mean value and s2 the variance of the
frozen/defrosted (F) and raw fruit (R) in the transmission and
dark-field images, respectively. The variances are weighted with
the factor w which is the ratio of the number of the frozen/
defrosted fruit (F) or raw fruit (R), relative to the total number of
the two, i.e. wF ¼ NF/(NF þ NR) where N is the number of pixels in
the area, and wR ¼ 1 wF. If the two areas are equal, w will take
the value one-half. A high CNR value means that the contrast
between frozen/defrosted and the raw fruit is higher than the
noise in the image whereas a low value means that there is no
significant contrast.
4. Image analysis
To compare the contrast between the frozen/defrosted and raw
fruit in the transmission and dark-field signal, respectively, an
image contrast measure is introduced in the form of a contrast-tonoise
ratio (CNR) which is defined as (Song et al., 2004):
CNR ¼ jmF mRj=
wFs2F
þ wRs2
R
1=2
where m denotes the mean value and s2 the variance of the
frozen/defrosted (F) and raw fruit (R) in the transmission and
dark-field images, respectively. The variances are weighted with
the factor w which is the ratio of the number of the frozen/
defrosted fruit (F) or raw fruit (R), relative to the total number of
the two, i.e. wF ¼ NF/(NF þ NR) where N is the number of pixels in
the area, and wR ¼ 1 wF. If the two areas are equal, w will take
the value one-half. A high CNR value means that the contrast
between frozen/defrosted and the raw fruit is higher than the
noise in the image whereas a low value means that there is no
significant contrast.
การแปล กรุณารอสักครู่..
