A Screening Application for Image Data Collected by an Acoustic Lens Sonar
Abstract
High-resolution, forward-looking sonar systems are critical tools on autonomous underwater vehicle (AUV) platforms where they are used to detect, classify, and localize objects. Data collected by these sensors can be processed automatically and used for navigation (object avoidance) or detailed object examination and discrimination tasks. Once an object is detected, size and shape parameters can be estimated based on the acoustic shadow cast behind the object when ensonified by such sensors. This report discusses the development of adaptive algorithms for highlight localization and shadow segmentation from 1.8-MHz data collected by an acoustic lens imaging sonar-DIDSON (Dual-frequency IDentification SONar). The sonar was mounted in a forward-looking configuration on a REMUS AUV. Algorithms screen large data sets for interesting frames; they demonstrate considerable effectiveness for target detection when the target images are central to the data frame and where there is little platform motion.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 2005
- Accession Number
- ADA441257
Entities
People
- J. B. Hsieh
- J. I. Olsonbaker
- W. L. Fox
Organizations
- University of Washington