Abstract
New generation of underwater 2-D forward-look sonar video systems operating at over 1 MHz frequency offer images with enhanced target detail. Within limited range of only tens of meters, they are the most suitable imaging systems for conducting visually guided missions in turbid waters. The reconstruction of 3-D target shapes or simply establishing the 3-D spatial location of interest points is a highly desired capability for automating many tasks. As achieved with 2-D optical images, multiple images from nearby positions may be utilized. This paper investigates some key issues in the accurate estimation of 3-D point locations from overlapping images of two forward-scan (FS) sonar systems in stereo configuration. We first present a stereo calibration algorithm and assess its accuracy. We then analyze the sonar stereo epipolar geometry. Beyond reducing the correspondence problem to a 1-D search along epipolar curves, this reveals unique properties associated with sonar measurements of range and azimuth angle. In particular, the search window to locate feature matches can vary significantly with stereo geometry, posing key tradeoffs. These issues are demonstrate using representative examples, and experiments with real data.