Abstract
Bats routinely discriminate multiple glint targets whose interglint time spacings are much less than the integration time of the bat auditory system. This suggests that the bats are exploiting spectral cues in the received echos to resolve the glints. Closely spaced glints cause notches and peaks in the received signal spectrum with frequency spacings characteristic of the glint separations. A biologically inspired signal processing algorithm for resolving these echos into target images is described. The algorithm combines a coarse energy detector to locate targets with a windowed short-time cepstral analysis, or cepstrogram, to resolve the fine structure of the detected targets. The algorithm’s performance is demonstrated using data recorded in a laboratory experiment. The experiments transmit an ultrasonic chirp and record the echos from a target consisting of a pair of cylinders spaced 5 cm apart, presented at aspect angles ranging from 0 to 90 deg. These recordings are processed with the cepstrogram algorithm using parameter choices commensurate with the performance of the bat auditory system. The cepstrogram algorithm accurately estimates the spacing of the cylinders. Combining the results from the left and right receiver channels produces a binaural image of the target. [Work supported by ONR.]