Three-dimensional reconstruction of underwater objects from a sequence of sonar images

Benoît Zerr, Bjarne Stage

2002 · 44 citations · 5 references

Concepts

TL;DR

Sonar images contain echo signals from the object, shadows where no back‑scattering occurs, and background noise. The study proposes a method to reconstruct the three‑dimensional shape of underwater objects from sequential sonar images. Using a sector‑scanning sonar on a vehicle that circles the target, the images are segmented into echo, shadow, and background, and separate elevation and reflection maps are built from shadow and echo data before merging. The resulting 3‑D reconstruction is obtained by merging the elevation and reflection maps.

Abstract

This paper describes a method to estimate the three-dimensional aspects of underwater objects using a sequence of sonar images. The data are recorded by a sector scanning sonar, carried by an underwater vehicle which follows a circular track around the object. The sonar images are segmented into three kinds of regions : echo, shadow and background. The echo information corresponds to the signal back-scattered by the object while the shadow is due to the absence of back-scattering from behind the object. The remaining information composes the background. The reconstruction scheme processes the echo and shadow information separately. A two-dimensional elevation map is estimated from the shadow information and a two-dimensional reflection map is computed from the echoes. The final reconstruction merges these two maps.

References

5