Upon computing the FRFT, the result contains an impulseshaped
peak that corresponds to the desired chirp. This chirp is
then filtered out from other sounds, by only retaining the bin
with the highest peak along with its few nearby bins and setting
all other bins to zero (illustrated in Figure 8.c). The ratio of the
peak value to the mean value of all zeroed bins prior to zeroing
provides a good measure for the quality of the perceived chirp
and the reliability of the obtained bearing measurement. Finally,
the filtered chirp in the FRFT domain is transformed back to the
time domain by computing the inverse FRFT. Steps and result
of the chirp extraction procedure for a detected chirp in a real
world experiment with two flying drones, one chirping drone
and one observing drone, is illustrated in Figure 8.