Towards of multi-source data fusion framework of geo-referenced and non-georeferenced data: prospects for use in surface water bodies

Abstract

This paper proposes an approach of fusion geo-referenced and non-georeferenced data. These data can be acquired by a) unmanned surface vehicles (USVs), b) unmanned aerial vehicles (UAVs), c) airborne and submersible optical sensors, d) acoustic sensors and e) aquatic sound profiling instruments. The test scenario consisted of monitoring and morphological characterisation of a confined water body. As part of the data integration and processing process, object segmentation modules were implemented for tracking and identification of underwater fauna, accompanied by Structure-from-Motion (SfM) and multi-view stereophotogrammetry (MVS) techniques for 3D reconstruction. The results obtained make it possible to recreate morphological models of the study/test areas by coupling all the existing information above and below the water mirror. The spatial resolution obtained is of centimetric order and the resulting data have the ideal quality to be used in underwater mapping and monitoring, as well as in 3D geomorphological measurements of surface water bodies HIGHLIGHTS ∙ Multi-source geo-referenced and non-georeferenced data fusion strategy ∙ Mapping of surface water bodies using unmanned aerial/aquatic vehicles. ∙ Application of segmentation techniques for monitoring and tracking of aquatic fauna. ∙ Three-dimensional reconstruction of underwater environments at high spatial resolution ∙ Robust framework for the integration of data from passive and active sensors.