Features
Let
be a set of
independent random variables (RVs) distributed
under hypothesis according to the common PDF .
The joint probability density function (PDF)
of is
Now let be ordered in decreasing order into the set
where
.
We now choose a set of indexes
, with
to form a selected collection of
order statistics
. To this set,
we add the residual “energy",
|
(8.1) |
where the set are the integers from to
not including the values
,
and is a function which is needed to insure
that has units of energy and controls
the energy statistic.
We then form the complete feature vector of length
():
By appending the residual energy to the feature vector,
we insure that contains the energy statistic.
We consider two important cases:
- if is positive intensity
or spectral data and has approximate chi-square statistics
(resulting from sums of squares of Gaussian RV),
then is sufficient.
The resulting energy statistic and reference hypotheses
are the “Exponential" in Table 3.1.
For this case, we consider to be a set of magnitude-squared DFT bin outputs,
which are exponentially distributed.
- if are raw measurements and have
approximate Gaussian statistics,
use or
.
The resulting energy statistic and reference hypotheses
are the “Gaussian" in Table 3.1.
For this case, we let be a set of absolute values of
zero-mean Gaussian RVs. Then, .