Watercore is an internal disorder that leads to breakdown of tissue and pos
sibly loss or downgrade of the product. It is very difficult to determine w
hether an apple contains watercore or not, especially in the early stages,
based solely on external information, since watercore does not alter extern
al texture until after severe internal breakdown. In this study, we explore
d the possibility of using two-dimensional (2-D) X-ray imaging to detect in
ternal watercore damage in apples. The algorithm to detect Red 'Delicious'
watercore apples consists of two stages, the first stage extracts features
from the apple x-ray. image and the second stage categorizes apples into di
fferent watercore levels using the features identified. A total of eight fe
atures were extracted from an x-ray scanned apple image and these features
were fed into neural network classifier to categorize them into three diffe
rent classes, clean, mild, and severe. The results showed that the system w
as able to correctly recognize apples into clean and severe categories with
in 5-8% false positive and negative ratios. The result also showed that the
algorithm was able to recognize apples independent of apple orientation, b
ut only if the stem-calyx axis made a fixed angle with the x-ray beam. Sort
ing at random apple orientation was not tested. The estimated speed of the
system, if implemented on a DSP board, will be fast enough to keep up with
the current apple processing line.