A data fusion approach for multiscale and hybrid surface topography metrology of porous materials

Novel high-performance porous materials enable lightweight, resource-efficient components with locally tailored functionalities, yet their complex multiscale microstructures challenge reliable metrology. No single optical technique provides a large field of view (FOV), steep-flank fidelity and practical acquisition durations. Focus variation (FV) enables rapid coverage but fails on near-vertical walls, while confocal laser scanning microscopy (CLSM) enhances axial resolution at the expense of speed and field size. This study introduces a geometry-related, adaptive multiscale framework by combining the individual strengths of each method. The three-stage pipeline first stitches 10× FV tiles into an overview topography, then quantifies local reliability via axial resolution mapping and slope-based Gaussian mixture modeling to identify systematic pore-flank failures and derive data-driven critical slope threshold. Finally, regions exceeding this threshold trigger targeted 20× CLSM re-probing with optimized FOVs and vertical ranges. FV and CLSM point clouds are then merged such that each modality contributes selectively according to local reliability. The resulting surfaces achieve superior axial resolution and geometric fidelity in steep pore walls, while reducing acquisition duration around 5 times versus full-field 20× CLSM, with only a minimal increase in surface topography data size. By integrating local geometry into measurement planning and merging, this framework enables precise, efficient topographic characterization of complex porous materials, providing a robust foundation for serial sectioning workflows and high-fidelity 3D volumetric reconstruction.