Unlocking Fractures: Automation and Wavelet Multiscale Analysis in Grid Data

Eniuce Menezes (UEM)

Abstract: The identification of features in images has been investigated from different methodologies. In some areas, the interpretation of features is still a challenge and depends on trained human ability. For instance, the automatic detection and analysis of fractures associated with surface and subsurface structures are critical because regional lineaments commonly represent surface expressions of geological weak zones at tectonic boundaries of basins and plates, as well as faults and rock fractures. This study presents an alternative approach for fracture detection in grid data (satellite and unmanned aerial vehicle (UAV) imagery, alongside digital outcrop models - DOMs) that exploits inherent multiscale characteristics, enabling a comprehensive assessment of fracture aperture. The proposed method integrates wavelet multiscale decomposition with curvature analysis derived from differential geometry, operating in automated or semi-automated modes. We are going to show both components enhance fracture signal extraction and estimate attributes across multiple scales.