280 CHAPTER 8. APPLICATIONS
Referring to Figure 8.21, the collection of insects is a random sample that is
representative of the various insect types that may be present in the agricultural
setting such as a large cotton farm. As such, statistical methods for analyzing
the data could be applied. Samples of insects can be collected from various parts
of a farm, and population estimates of each class of arthropod can be obtained.
Figure 8.21. Integrated pest management scheme
Such estimates may be easy to obtain for small farm areas, but certainly
present a formidable task for large farms that could be thousands of acres in
size. For this reason, we need some automated means of collecting insects and
analyzing the sampled collection for various classes of insects. For this, a com-
puter vision-based approach can be used for classifying insects into two basic
classes, namely, desirable and undesirable insects. By desirable, we mean insects
that belong to the predator species, and undesirable implying pests.
In addition, we are also suggesting that each of the basic classes be further
divided into specific insect types that form the biological control species and pest
species, respectively. Intuitively then, a ratio of desirable to undesirable insects
should provide a basis for developing decision rules for pesticide application.
It is clear that these decision rules would be fuzzy ìIf...then...î rules. For
example, a typical set of fuzzy rules might be of the form