Mapping techniques of underwater environments by optical-acoustic data integration

Moroni D, Pascali Ma, Reggiannini M, Salvetti O

Underwater Acoustic  Underwater Archaeology  Data Fusion 

A new method is proposed to integrate 3D optical and acoustic images relative to the same underwater environment. The combination of optical and acoustic sensors in terms of uniform reference system, georeferencing and time allows: (i) integration cascade (operational level), (ii) safety data acquisition in various domains (distance from ground, turbid water, vegetation, etc.), (iii) replanning of missions in progress. Furthermore, data fusion can be faced according to different approaches: (a) stratification of referenced data layers, (b) correlation of quantities of different nature, (c) comparison of extracted features: 2D geometries (segments, elementary curves) and 3D (planes, simple surfaces), repetitive patterns, (d) integration of semantic information, (e) template matching for recognizing known structures, (f) creation and refinement of probability maps as a measure of optical (geometry, texture) and acoustic (elevation or reflectivity maps) properties. Unlike most of the existing approaches that perform recognition and matching of different interesting relevant points appearing in two images, we look at the correspondence of salient features present both in optical and acoustic images representing the same scene. A set of geometrical and textural feature extraction algorithms is applied to the multi-sensor images and compared with the output results. We aim thus at emphasizing the geometric structure alignment of features (e.g., lines or different kind of curves), instead of descriptor-based individual feature matching. This is due to the fact that optical and acoustic image properties and patch statistics of corresponding features are generally quite different. Besides, the spatial layout of features is preserved in both maps.

Publisher: International Institute of Acoustics