High-Precision Cross-Scale Data Registration Method for Combination of AFM and White Light Interferometery

Multi-sensor fusion leverages the strengths of different measurement techniques to broaden their applicability and enhance measurement accuracy. While significant progress has been made in developing multi-sensor data fusion methods, existing approaches in surface metrology face challenges in effectively extracting matching datums from cross-scale objects, where the horizontal resolution difference between datasets can exceed a factor of ten. The main limitation is that alignment accuracy is constrained by these resolution disparities. This paper presents a novel hybrid AFM-WLI instrumentation to combinate Atomic Force Microscopy (AFM) and Vertical Scanning White Light Interferometry (WLI). Additionally, the study proposes an alignment method for the fusion of AFM and WLI datasets, despite lateral resolution differences ranging from 1:10 to 1:600. Using a new Iterative Similarity Region (ISR) algorithm, this approach replaces feature points with feature regions as matching datums. By extracting feature region descriptors at a common spatial scale from both measurement types, the method employs an attention-based alignment mechanism to precisely locate the AFM region within the WLI coordinate system. This enables cross-scale fusion of micron- and nanometer-resolution data. The experimental results demonstrate that both WLI and AFM modes in the hybrid measurement system achieve excellent height measurement accuracy. Moreover, the system enables precise localization of ultra-high-resolution AFM data within large-scale WLI datasets.