Suppose further that we have used a camera to digitize the visual input,
and we have isolated a single character as an array of brightness values.
How can a computer classify this data? An obvious approach is to compare
the input with a standard pattern for each class, and to choose the class
that matches best. The obvious problem with this approach is that it doesn't
say what to compare or how to measure the degree of match.
What makes pattern recognition problems hard is that there can be a large
degree of variability of inputs that belong in the same class, relative
to the differences between patterns in different classes. One way to cope
with this problem is to look for characterizing features.
On
to Features 
Up
to Models