A comprehensive computational model of animal biosonar signal processing

Computational models of animal biosonar seek to identify critical aspects echo processing responsible for the superior, real-time performance echolocating bats and dolphins in target tracking clutter rejection. The Spectrogram Correlation Transformation (SCAT) model replicates imaging both species by wideband sounds echoes with auditory mechanisms identified from experiments bats. acquires broadband broadcasts echoes, represents them as time-frequency spectrograms using parallel bandpass filters, translates filtered signals into ten amplitude threshold levels, then operates on resulting time-of-occurrence values at each frequency estimate overall range delay. It uses structure spectrum depicting it a series local nulls arranged regularly along axis after dechirping relative broadcast. Computations take place entirely timing threshold-crossing events threshold-events Threshold-crossing times account amplitude-latency trading, physiological feature absent conventional digital signal processing. Amplitude-latency trading transposes profile amplitudes across frequencies time-registrations frequencies. Target shape is extracted spacing object’s individual acoustic reflecting points, or glints, mutual interference pattern peaks spectrum. These are merged range-delay produce delay-based reconstruction distance well its glints. Clutter indiscriminately activate multiple parts null-detecting system, which produces equivalent glint-delay spacings images, thus blurring echo-delay estimates adding spurious glint delays image. Blurring acts an anticorrelation process that rejects intrusion perceptions.