Signal Processing and Acoustic Imaging Models for Animal Sonar
Abstract
A biologically inspired acoustic imaging process is obtained by considering the constraints (e.g., limited array/aperture size), and advantages (e.g., unrestricted motion, wide bandwidth) of biological sonar systems. The resulting system is easy to implement compared with conventional synthetic aperture sonar. Images are formed sequentially; an image is updated on the basis of each new echo. The processor is Doppler tolerant and is tolerant of target-induced phase shifts that occur as the result of aspect changes. The simplicity of the process leads to generalized images that can distinguish between smooth and rough surfaces and that can detect and classify low-reflectivity targets in strong volume clutter. Accuracy is improved via an associative gradient descent algorithm that can converge rapidly to a global minimum by utilizing prior information, extra acoustic information that is not directly used for imaging (e.g., resonance phenomena), and non-acoustic cues. An image-based tracker exploits sequential image formation and parallel processing to compensate for motion perturbations and to model simultaneous imaging and pursuit of prey in animal sonar systems. The imaging system is remarkably robust with respect to sparse angular sampling; image quality degrades gracefully as angular sampling intervals are increased well beyond conventional limits. Insights into neurophysiology and cognition include the utilization of top-down, bottom-up processing to obtain high resolution images from low resolution representations, as in the superior colliculus.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 20, 1999
- Accession Number
- ADA366372
Entities
People
- Richard A. Altes