results obtained from the machine vision system and the manual
method were not significant. This means that an insufficient
detection rate is not related to the weakness of the
proposed system. In order to enhance the system, other technologies
which use the sense of touch can be used, because
the prime difference between Tamar and Rotab on the verge
of Tamar is the softness of texture that cannot be sensed by
machine vision. The accuracy of sorting in the Khalal stage
for machine vision and human vision was 99.66% and
100%, respectively and differences between human and machine
vision were not significant. Therefore the performance
of the proposed system was satisfactory, bearing in mind that,
detection of all Khalals accurately was one of our objectives.
The accuracies of machine vision and human vision for sorting
the Rotab stage were 97% and 100%, respectively and differences
between human and machine vision were significant. The
reason for misclassification of the Rotab stage was mentioned
before (see Fig. 11). Lee et al. (2008) succeeded in obtaining a
detection rate of 74.53% for yellow-head (Rotab) Date fruits.
Therefore some improvement in algorithm is needed; otherwise
some modification is necessary in the conveyor section of the
system in such a way that all sides of a fruit could be exposed
to a camera.