(Massachusetts Institute of Technology and Woods Hole Oceanographic Institution, 1990-03)
Cobra, Daniel Távora de Queiroz
This thesis introduces a new procedure for the enhancement of acoustic images of the
bottom of the sea produced by side-scan sonars. Specifically, it addresses the problem of
estimating and correcting geometric distortions frequently observed in such images as a
consequence of motion instabilities of the sonar array. This procedure estimates the geometric
distortions from the image itself, without requiring any navigational or attitude
measurements. A mathematical model for the distortions is derived from the geometry
of the problem, and is applied to estimates of the local degree of geometric distortion
obtained by cross-correlating segments of adjacent lines of the image. The model parameters
are then recursively estimated through deterministic least-squares estimation. An
alternative approach based on adaptive Kalman filtering is also proposed, providing a
natural framework in which a priori information about the array dynamics may be easily
incorporated. The estimates of the parameters of the distortion model are used to rectify
the image, and may also be used for estimating the attitude parameters of the array. A
simulation is employed to evaluate the effectiveness of this technique and examples of its
application to high-resolution side-scan sonar images are provided.